Numerical Modeling of Anti-icing Systems and Comparison to Test Results on a NACA 0012 Airfoil

NASATechnical Me._m_orandq.m 1_05975 : /"_ _' _- 3 __f ....

AIAA-93-0170 ............ __ 2f ..... #,/_/

............... " iing--Sf-_ti:Icing = Sy _ _

..... Numenca! Mode .... _ stems

=_i_-:?:and Comparison toTestRe_u!ts on a .... i_ i

Kame! M, AI-Khali[ and Mark t31 Potfip6zuk

Lewis Research Center .......................

........ Cleveland, Ohi6

Prepared for the ....

3!st Aer6space S cje_ce_Meefing and Exhibit .................

=sponsored by the American Institute of Aero-nauficS and Astronautics .........

Reno, Nevada,January 11-14, 1993 ....... :- - - .... ....... ---

N93-16345 (NASA-TM-IO5975) NUMERICAL MODELING OF ANTI-ICING SYSTEMS AND COMPARISI]N TO T_ST RESULTS ON A uncl_s NASA O012 AIRFOIL (NASA) 14 p G3/03 0136230 Numerical Modeling of Anti-icing Systems and Comparison to Test Results on a NACA 0012 Airfoil Kamel M. AI-Khalil* and Mark G. Potapczuk t NASA Lewis Research Center Cleveland, Ohio 44135 A series of experimental tests were conducted in the NASA Lewis IRT on an electro-thermally heated NACA 0012 airfoil. Quantitative comparisons between the experimental results and those predicted by a computer simulation code were made to assess the validity of a recendy developed anti-icing model. An infrared camera was utilized to scan the instantaneous temperature contours of the skin surface. Despite some experimental difficulties, good agreement between the numerical predictions and the experimental results were generally obtained for the surface temperature and the possibility for the runback to freeze. Some recommendations were given for an efficient operation of a thermal anti-icing system.

Nomenclature 0 = rivulet contact angle with the solid surface p = density pressure coefficient Cp z = aerodynamic shear force at the skin surface friction coefficient

c I --

F = ratio of wetted area to the total surface at a Subscriots particular streamwise location FF = freezing fraction: ratio of liquid water that surf = outer surface of the aircraft skin freezes to the total liquid entering a control oo = frec stream volume LWC = I. Introduction liquid Water Content (g/m 3) MVD= mean volume droplet diameter (pro) Aircraft icing problems have been encountered for m ° ._ runback water mass flow rate several decades during flights in adverse weather conditions. Aerodynamic penalties as well as the P = static pressure dramatic accidents associated with ice accreted on critical aircraft surfaces have led many researchers to study this c[' = rate of heat transfer per unit area at the of the phenomenon. Extensive studies are required in order to skin surface better understand the icing process and to accurately S surface distance from stagnation (positive on develop analytical models and computational codes for the upper surface and negative on the lower) use as analysis and design tools. The cost involved to develop these codes is far less than the cost of obtaining a T temperature complete experimental database.

V = Velocity De-icing systems have been mainly developed to ratio of chordwise distance to the airfoil chord x/c = protect aircraft wing surfaces and rotor craft blades. These systems may not be suitable to protect engine inlets droplet collection efficiency since any ice shed from leading nacelle surfaces can cause ratio of rivulet width to wetness factor, or serious damage to engine components due to the high distance between two surface streamlines force of impact of ice particles. Furthermore, ice accretion on inlet nacelles distorts the intake airflow, thereby reducing the engine propulsion efficiency.

* NRC Research Associate at NASA Lewis Research Center, Therefore, anti-icing systems are generally used to Member AIAA continuously maintain inlet surfaces free from ice.

"t" Aerospace Engineer, Member AIAA This study is concerned with anti-icing systems, and in particular the thermal type (hot air and electro-thermal Copyright © 1993 American Institute of Aeronautics and Astronautics, Inc. All rights reserved.

heaters). But the modeldeveloped may alsobe skin. The surface tension phenomenon derives from intermolecular cohesive forces. The molecules in the

incorporated in theanalysis of water, fromthermally de-

iced surfaces, that runs back downstream onto unprotected

thin layer of water of the impingement region are mainly surfaces.

subjected to attractive forces between neighboring molecules. At the water-air interfacial surface, an Recently, a runback model was developed [1,2,3] to unbalanced cohesive force exists and is directed toward the better simulate the physics of water flow on aircraft liquid side. This unbalance causes the interfacial water surfaces. This was accomplished by incorporating a molecules to move inwards. Consequently, the interface rivulet model for the streams of water often observed on tends to contract which may lead to the breakup of the airfoil surfaces during "running wet" anti-icing test cases.

water film into beads (macroscopic drops) or rivulets A computer code was developed to simulate running wet (shallow and narrow streams). These phenomena are and evaporative aircraft anti-icing systems. This model is referred to as capillary phenomena. This explains many based on the analysis of the thin liquid water layer which of the observations made earlier by Olsen and Walker [4], forms in regions of direct impingement and, then, breaks and later repeated by Reehorst et. al [5,6], in the NASA up into rivulets in the vicinity of the impingement Lewis Icing Research Tunnel.

limits. The wetness factor distribution, F, resulting from the film breakup, and the rivulet configuration on the In the neighborhood of the stagnation line, relatively surface are predicted using a stability analysis theory and small water drops form due to the coalescence of many the laws of mass and energy conservation. The rivulets microscopic cloud droplets impacting the surface because formed were assumed to be equally spaced as illustrated in of surface tension effects, as explained above. This is Fig. 1. The value of F is updated from that location only true if the water is not evaporated by a thermal anti- onward up to the point where total evaporation occurs, icing system. When the drops grow large enough such where the value off becomes zero.

that aerodynamic forces overcome the adhesion forces between the drops and the solid surface, they are swept Several observations were made during experimental back downstream on the aircraft skin. However, the tests conducted in November 1991, in the NASA Lewis surface in the direct impingement region is still Icing Research Tunnel (IRT). Most of the data acquired considered entirely wetted. This is due to the incoming were qualitative, some of which were reported in droplets that continuously tend to wet the surface between Reference [3]. Two of these observations are shown in the water beads with a very thin liquid layer. This Figs. 2 and 3. Figure 2 illustrates several rivulets phenomenon was observed in the IRT during the running back to downstream regions on the surface of a previous [3] test as well as during the current tests.

thin leading edge body during one of the test cases.

However, at the limits of impingement, the liquid film is Upstream of the location where rivulet formation started, most probably unstable due to surface tension forces, and the entire surface was totally wet. Similarly, Fig. 3 it can no longer wet the entire surface. This causes the shows a more definitive breakup of the thin liquid film surface water to coalesce into individual streams, referred into rivulets on an engine inlet cowl. Rivulets in the to as rivulets, separated by dry spaces.

latter case were more consistently established due to the larger leading edge radius of the cowl. This triggers the In the analytical model, the rivulets formed are significance of scaling when testing a sub scale model in assumed to be equally spaced, with the cross-section of a refrigerated spray tunnel. The qualitative experimental each rivulet having the shape of a portion of a cylinder.

data agreed well with the code predictions. The angle between the surface and the line tangent to the curve representing the liquid-air interface at the triple In the current work, a series of experimental tests point, is called the contact angle 0 (see Fig. 1). It is a were conducted in the NASA Lewis IRT on an electro- function of the liquid and the surface properties. The thermally heated NACA 0012 airfoil. This study mainly water flow on the surface is driven by the aerodynamic includes quantitative comparisons between the friction on the aircraft skin, and the flow is assumed to be experimental and the numerical surface temperature along surface streamlines. Details on the hydrodynamics predictions to assess the validity of the current computer and stability analysis of shear driven thin liquid films anti-icing model. An infrared camera was utilized to may be found in Ref. [1], and its application to anti- measure the instantaneous temperature profile of the outer icing systems modeling in Ref. [2].

skin. A general overview of the analytical model The runback model utilized herein is based on a two- development will first be given. Then, discussions on dimensional mathematical formulation. The surface the experimental setup as well as the numerical and test results will follow.

water and the solid structure temperatures vary across their thicknesses and in the flow direction along a streamline on the surface. Spanwise temperature II. Analytical Model variation is assumed to be small. However, the latter is accounted for by performing energy balances on control ILl Runback Water volumes whose spanwise widths extend between two adjacent streamlines on the aircraft surface. The latter are The rate of water striking leading aircraft surfaces is determined from the flowfield solution.

relatively small. Consequently, its behavior is predominantly controlled by aerodynamic effects, water surface tension, and the physical properties of the aircraft transfer across the interface of different property layers, Additional Feature: and at the domain boundaries. The method also possesses a high numerical stability.

The computer model was originally developed for anti-icing applications such that the heat required to keep The temperature is assumed to vary in the the runback water from freezing is supplied to the surface, streamwise direction and across the thickness of the wall.

at the minimum. This is because a two-dimensional It is further assumed to be constant in the spanwise phase-change model was found to be inappropriate since direction. However, the variation of the width of each freezing will normally start at the liquid-air interface, strip in the spanwise direction (i.e., the distance between which creates a problem in modeling the flow two adjacent surface streamlines) is accounted for in the characteristics of the unfrozen water. However, since the energy balance equation of each control volume. For temperature drop across the film thickness is small, the example, this width is constant for a two-dimensional temperature may be assumed to be uniform across the flow over a 2D airfoil, and is variable for flow about an layer. Therefore, when a freezing temperature, or lower, engine inlet nacelle or a tapered wing.

is attained during the calculation process, an alternate method is used.

III. Computational Scheme The method used consists of performing a A simultaneous numerical solution of the governing macroscopic energy balance on the surface water to obtain energy and mass conservation equations must be obtained the freezing fraction. This is defined as the ratio of water in the following three regions: (1) runback water; (2) that freezes in a control volume to the total mass of solid structure; and, (3) anti-ice bleed air, if it is used.

liquid water entering the control volume. Nevertheless, This is carried out using an iterafive type of solution due the rivulet configuration and the method used for to the dependency of the boundary conditions of each prediction of the wetness factor remains unchanged. This region on the final solution of its adjacent region.

enables one to predict the amount and location of ice accumulation during a specified period of exposure time.

A fully implicit method was used to numerically solve the runback water energy equation because of its 11.2 Anti-icing systems: positive stability properties. Backwards differencing in the streamwise direction, and central differencing in the The current computer model was designed to handle transverse direction were employed. This produces a two different types of thermal anti-icing systems: the hot temperature distribution within the water layer thickness air type, and the electro-thermal type. In the hot air type at each streamwise location. This distribution is then systems, hot air is generally drawn from an intermediate averaged locally to produce a single value of the local or high stage compressor bleed port or provided by a runback water temperature.

heating system through a heat exchanger. The air is then The governing energy equation of the anti-ice bleed ducted through passages to the leading edges of wings, air, is a first order ordinary differential equation (ODE).

empennages, engine nacelles or other critical areas. Since the anti-ice air temperature distribution depends on the A very accurate and frequently used technique for solving ODE's, is the fourth order Runge-Kutta method.

solid wall temperature distribution which also depends on the runback water solutions, the energy equations of Knowing the temperature distribution in the wall, from the most recent iteration, the latter method is used to those three regions must be solved simultaneously.

predict or update the hot air temperature distribution in In the electro-thermal type systems, the heating the cowl. The result is subsequently used in the wall elements are modeled as individual layers in the aircraft temperature solution at the next iteration.

composite body structure. The mathematical formulation also allows anti-icing to be achieved by means of simply Finally, the control volume approach was chosen to applying a specified heat flux distribution at the wall represent the heat balance on the wall structure as inner surface. This is provided as a user input option to mentioned earlier. Difference equations are derived by the code. In addition, any combination of the three performing an energy balance on each control volume methods of heating is permitted. This type of corresponding to a particular node. Details of the formulation was chosen to give the flexibility of numerical techniques and the iterative solution procedure between the regions of interest are presented in Ref [2].

modeling different systems and to provide a tool that can be used in the design and analysis stages of the system development. IV. Experimental Setup 11.3 Wall Structure IV.1 Test Model Several different approaches are available to The current tests were conducted in the NASA Lewis numerically simulate the temperature variations within Research Center Icing Research Tunnel ORT) during the the aircraft skin structure. The method used herein is the last two weeks of August, 1992. The test section in the control volume approach. An equation is written for each IRT is 9 ft wide, and 6 ft high. The test model used is a node in the discretized region using the laws of energy NACA 0012 airfoil with 21 inch (0.533 m) chord, and conservation. This is an accurate method due to its 6 feet (1.829 m) span. It is centered vertically on the conservation properties, specially when dealing with heat turntable of the IRT Test Section to control the airfoil the water vapor in the atmosphere. However, energy absorption is almost negligible in the long wave range angle of attack with the air flow. It comprises a (8-14 pro) of the electromagnetic spectral band.

removable leading edge that measures 35 inches spanwise which fits on the center of the model. The 18.5 inch on The instrument used herein for surface temperature each side of the center piece are non-active, while the measurements is the Inframetrics Model 600 IR system center section is built to accommodate the ice protection which is designed to intercept radiant energy in the band system to be tested.

of 8 to 14 Bm. Generally, most non-metallic materials have high emittance in this region. However, the The center section has an active region where the majority of metallic surfaces have a low emittance. For surface is ice protected along a 32 inch strip along the example, Aluminum has an emittance value Of as low as span. It is a 0.25 inch thick fiberglass material which is covered with a thin layer of epoxy/glass insulation. The 0.04, depending on its purity and surface polish. In the heater mats are then laid on top of the insulation current experimental setup, the airfoil model had an material. The latter were divided into five isolated zones Aluminum abrasion shield with high reflectance. In order to increase its emissivity and to eliminate any reflectance as illustrated in Fig. 4. The heating rate in each zone is from other objects, the surface was coated with a thin individually controlled by a voltage Variac. The zone layer of flat black paint. Consequently, the emissivity of setup is as follows: Zone A is 0.5 inch (1.27 cm) wide the surface is increased to about 0.95.

centered at the leading edge, Zone B is on the upper side and zone C is on the lower side of A, 1.0 inch (2.54 cm) The infrared scanner and analyzer setup used in the wide each; and, finally, Zones D&E, adjacent to B&C, test is the Inframetrics Model 600. This model has a respectively, at 1.5 inches (3.81 cm) wide each.

very high frame rate which allows data to be viewed The heater mats were insulated on each side with a without any sacrifice of time delay or accuracy. The system provides a video output signal that is fully temperature/electrical resistance tape. An aluminum abrasion shield was laid on top of the heater mats. In compatible with the NTSC television standard.

between the two layers, a heat transfer grease was used to The IR camera was set in the IRT control room to make a good thermal contact between the heaters and the monitor the surface of the airfoil (almost aluminum shield to increase the heat transfer performance perpendicularly). The center window of the tunnel and reduce the possibility of delamination of the layers.

observation room was replaced with a Plexiglass window with a hole drilled in it to allow viewing the model with The current test served two purposes. Technology the IR scanner. A 4.75 inch diameter CLEARTRAN TM Dynamics Group, Inc., the developers of the new technology heater mats, were interested in testing the clear window (water clear zinc sulfide) access port was inserted in that hole. It possesses attractive optical operation and performance of their product as an anti/de- properties since it can transmit radiant energy in the icer system, while one of NASA Lewis's interests was to visible through the infrared range (0.4-16 microns) of the expand their existing experimental database to support in- spectrum. Its transmittance for that range is house computer code validation.

approximately 0.% IV.2 Infrared Temperature Measurements A few problems were encountered with using the CLEARTRAN TM window. The cold temperature inside There exist several commercial types of non-contact the tunnel test section caused a tremendous amount of temperature measurement instruments. Most of them water condensation on the window of the control room rely on the same basic principles. All objects emit side. To overcome this problem, a heat gun was used to radiant energy which increases with increasing surface keep the surface clean. However, the amount of heat temperature of the object.

required to achieve that objective caused an increase in the The emitted energy is distributed over a band of window temperature which affected the IR camera surface wavelengths in the electromagnetic spectrum. This temperature readings. This may be associated with the distribution may be represented by a curve known as the high reflectance of the window surface as well as its heat Planck function. The peak of this curve shifts towards absorption.

the shorter wavelength regions as the object surface Finally, the best solution to these problems was to temperature increases.

use a small sliding window that may be opened during Objects which are at near room temperature have the data acquisition and closed afterwards to reduce the noise level in the control room. Thus, no obstructions spectral energy distributions that peak near the middle of exist between the scanner and the airfoil surface.

the infrared region (i.e., close to 10 microns). The radiated energy is received by a radiant sensitive Consequently, no calibration was required for the scanned instrument which is calibrated to translate the energy temperature readings except for the surface emissivity and intercepted into temperatures using the appropriate radiant energy absorbed by the water droplet cloud.

relationship between the two variables.

Eventually, a portion of the radiated energy is absorbed by the atmospheric medium that exists between the object and the temperature reading instrument.

Typically, the majority of the absorbed energy is due to resulting curvilinear body fitted mesh was composed of: IV. Experimental Results and Comparison to (309 grid points along the surface) by (65 nodes in the Numerical Calculations nearly normal direction).

The droplet trajectory and impingement module of Although the tests consisted of de-icing as well as NASA's ice accretion code LEWICE [8], originally anti-icing runs, only the latter are of concern in the developed by FWG Associates under a contract for NASA present discussion. In these cases, the airfoil was Lewis, was used to generate the collection efficiency mounted vertically on the turntable in the IRT such that curves. The code was modified by Potapczuk [7] to read the angle of attack of the airfoil with the flow was and interpolate the flowfield velocities from the outputs 4 degrees. The lower surface (pressure side) was visually generated by ARC2D. Minor modifications were also and instrumentally monitored from the control room required for the code to handle viscous flows when the through heated windows and access ports.

flowfield code is run in a viscous mode (Navier-Stokes equations) as opposed to non-viscous compressible mode The complete numerical solution to the problem was resolved in three major steps: (1) flowfield calculations, (Euler equations).

including the viscous layer near the wall; (2) individual The particle trajectories were subsequently used to water droplet trajectory calculations using the velocities compute the local collection efficiencies. The cloud was calculated in the previous step; and, finally, (3) the heat assumed to be mono-dispersed with a droplet size of transfer calculations to determine the surface water and either MVD = 15 ttm, or MVD = 20 lam, depending skin temperature distributions.

on the particular case being considered. The The flowfield around the airfoil must be calculated corresponding results are respectively shown in Figs. 7 such that viscous effects near the wall are accounted for.

and 8 for the different air speeds used during the tests.

This can be accomplished with a non-viscous flow solver In an earlier discussion, it was mentioned that when which includes a boundary layer model. On the other runback water exists on the surface it will flow back as hand, it may be accomplished using a code that rivulets, or a combination of water beads and rivulets, numerically solves the full or a simplified version of the beyond the impingement region. Previous Navier-Stokes equations. ARC2D, a "thin layer" Navier- observations [3] showed that pure rivulet flow structure Stokes solver developed at NASA Ames, was used for is more probable to exist on blunt bodies than on thin this purpose. It has been used lately for icing-studies at leading edge bodies as illustrated in Figures 2 and 3. As NASA Lewis by Potapczuk [7], and was found to give expected, since the airfoil used in the current test may be satisfactory results for low angles of attack. The model considered as a thin leading edge body, the current is adequate for high Reynolds number, and for attached observations agreed with the above conclusion.

and mildly separated flows where the viscous terms associated with derivatives in the direction parallel to the Individual beads were seen to flow on the surface in body are negligible. The output format from the code cases of low mass flow rates. These cases correspond to was tailored to the inputs requirements of the current heat at least one of the following conditions: (1) low liquid transfer code.

water contents, (2)low freestream velocities, and, finally, (3)high rates of evaporation caused by high The resulting pressure coefficient and friction power densities applied to the surface in the impingement coefficient distributions from ARC2D are illustrated in regions. Another reason for the formation of water beads Figs. (5) and (6), respectively. These coefficients are is the highly smooth solid surface which decreases the defined as follows: adhesion forces between the drops and the surface, P-P..

consequently increasing the contact angle 0 (see Fig. I).

Ct, - (1) Once rivulets were seen to leave the direct impingement region, they froze in downstream regions for two reasons: (1) the downstream regions were thermally unprotected, (2) less heat was absorbed by the rivulet due to the ano, C/= "r decreased wetting factor.

£ _ v" 2 (2) Moreover, when large beads formed on the smooth and thin leading edge body surface, they moved slowly in The first coefficient may be used to calculate properties at the downstream direction near the stagnation line.

the edge of the boundary layer, and the second is used to However, once the beads moved slightly beyond the compute the wall shear stress that causes the water to run stagnation region, they ran back at a much faster speed back.

onto unprotected regions. This is due to the rapid increase in the local aerodynamic shearing forces as was Due to the effects of viscosity near the solid surfaces, illustrated in Fig. 6. The drawback in this situation is large gradients of the variables exist in these regions.

that the amount of heat absorbed by those beads This requires a fine grid spacing in the directions normal diminishes drastically due to the decrease in residency to the rigid body which allows the flow solver to capture time of the drops in the heated region, normally the all the necessary details in those regions. To region of direct impingement and slightly further accomplished this, a hyperbolic grid generator was used to produce a C-type grid structure around the airfoil. The downstream. Consequently, total evaporation should be carefully accomplished as early as possible without Nevertheless, the cases that were considered in this inducing any probable damage to the heater mats or the report were chosen such that most of the problems structure.

mentioned earlier either did not exist or had been eliminated. Those cases are summarized in Table 1.

Several operating conditions were considered in the current tests. In the earlier runs, the model failed after They represent five different operating conditions: several hours of operation. One of the main reasons freestream temperature, velocity, liquid water content, associated with this failure, burnout at different spots, mean volume droplet diameter, and power densities. The was that the substrate was only able to handle a airfoil was set at 4 degrees angle of attack in all cases.

maximum temperature of about 250°F. In addition, the Since only the lower surface was monitored by the heater mats were initially laid on the substrate surface infrared camera for surface temperature measurements, the without a heat insulating material. This problem was results shown in the following sections correspond to resolved by using a layer or two of glass epoxy as an that side only of the airfoil surface. In the following insulating material. figures, "s" represents the distance from stagnation on the surface and is negative on the lower surface (pressure Another problem encountered was the relatively long side).

period required for the system to reach a steady state when running in an anti-icing mode. This problem was not The summary of results are shown in the last expected to occur, and, therefore, many data points were column of Table 1. In Cases 1 and 2, the table clearly collected at an earlier time than required for the system to shows that the system, when running in an anti-icing stabilize. This problem was not detected until later mode, was able to completely evaporate the water during the test from thermocouple temperature readings. impinging on the surface before it ran back to the The cause for this long period that was required for a unprotected downstream regions. However, in cases 3 stable condition to be reached was mainly associated with through 5, the heat supplied to the surface was only able a heat transfer grease that was applied between the heater to partially evaporate the surface water, which eventually ran back and froze at some distance downstream.

mats and the abrasion shield, as mentioned earlier.

The idea behind the use of the grease was to make a Figures 10 and I 1 represent the surface temperature better physical contact between the heater mats and the distribution, and, the runback mass flow rate and freezing abrasion shield, and to eliminate the possibility of the fraction distributions for Case 1, respectively. The existence of air gaps. This, supposedly would increase corresponding variables are plotted in Figs. 12 and 13 for the system performance. However, although the grease Case 2, respectively. These results show that the had a very good thermal conductivity property, its experimental surface temperature readings are in good thermal capacitance (product of density and specific heat) agreement with the code predictions in Case 2. The was very high. Consequently, the grease took a long differences shown in Fig. 10 for Case 1 are slightly time to warm up when the heaters were fired on. Later, above the experimental margin of errors. This may be when the power density is reduced or the heaters shut off, associated with the poor code prediction of the heat transfer coefficient distribution between the surface and the grease also took a longer period to cool down due to the large amount of heat initially absorbed. the freestream air. However, the agreement was perfect Subsequently, at least five minutes were allowed for the regarding the prediction of total evaporation. Figures 11 system to stabilize before recording the actual data. and 13 show that only a very small amount of liquid water (because FF = 0) exists in direct impingement One of the difficulties with measuring the surface region in these two cases. However, total evaporation of temperature using an IR camera is calibration. The IR the water is achieved upstream of the impingement scanner has different settings, user defined, for the range limits.

of temperature to be measured. A separate control is then provided to set the center temperature within the In Cases 3 and 4, as shown in Figs. 14 through 17, predefined range. The smallest range that may be used is the runback water froze downstream of the heated regions.

5°C. However, this is not sufficient enough for such The location where freezing begins corresponds to the measurements since the surface temperature may vary location where the freezing fraction FF goes above the from about freestream temperature to the neighborhood of value of zero. This corresponds to the computer predicted 25°C. This temperature range can be greater than 50°C. surface distance of approximately 0.095 m from the The problem with such a high range settings on the IR stagnation point for Case 3, as shown in Fig. 15. The scanner is that the error in temperature readings increases close-up video from the experimental observations with increasing temperature range settings. This had to showed that freezing for that case occurred at a distance of be taken into account with some degree of difficulty with about 0.11 m. In Case 4, the freezing location was the calibration. Mainly, the temperatures were calibrated predicted at about 0.11 m, and the corresponding against regions downstream of the heated zone, where the experimental observations showed that this occurred at surface temperature would be known to be in the about 0.1 m. This distance varied spanwise during the neighborhood of the freestream temperature. Figure 9 experiment by at least 0.01 m in all the cases.

illustrates a sample video output from the IR scanner (the Therefore, the computer predictions are considered to be original scan is in color). in very good agreement with the test results.

The surface temperature predictions for the latter two

V. Concluding Remarks cases are shown in Figs. 14 and 16. The agreement is very good in Case 3 and is relatively good in Case 4.

The analysis developed earlier confirmed that the Note that in Case 4, the numerical code predicts a plateau liquid water film which forms in regions of direct droplet region where the surface temperature is zero degrees impingement will breakup into beads and/or rivulets in Celsius. This is due to the assumption that the water/ice the neighborhood the impingement limits. Despite some mixture (freezing fraction is larger than zero and smaller experimental difficulties, very good agreement was than one) remains at zero degrees Celsius until all the generally obtained between the numerical predictions of water freezes. However, this is not what was actually the computer code and the experimental results of surface predicted by the IR camera. It seems that the temperature temperature. For cases where the freestream temperature continuously decreases in some fashion.

was exlremely cold, the agreement was acceptable. The The same is true for the surface temperature location where the runback water may freeze, if predictions of Case 5, as shown in Fig. 18. A couple of conditions favor it, was well predicted by the computer reasons may be associated with the disagreement of model.

results. One reason may be due to the fact that the An efficient anti-icing system design may be accumulated ice is cooling below freezing, while the achieved with a sufficient heat supply to the regions of water that remains unfrozen runs back in the valleys direct impingement within the tolerable limits of the between frozen rivulets. The other reason, which is less aircraft structural materials for two reasons. The first is likely to tremendously affect the measurements, is that due to the fact that runback water flows fast past the the IR scanner is calibrated for surface temperatures of heated regions such that its residency time in those above -20 °C. Notice that the cases where marginal regions is not sufficient to achieve total evaporation.

agreement occurred between the numerical and The other reason is that more heat is transferred to the experimental results are those which correspond to a very water to evaporate it in the impingement region and less cold freestream condition (Too = - 17.78 °C). These heat is lost to the ambient compared to the heat loss correspond to Cases 1, 4, and 5. This also causes the through dry areas on the surface in the downstream outer layer of the frozen ice to cool faster than the liquid regions. Under extreme weather conditions, the heated water running between the valleys.

zones should also be extended further downstream to Inspection of the operating conditions given in achieve complete evaporation of the runback water and Table 1 shows that the only difference between Case 4 prevent it from freezing.

and Case 5, is that the freestream velocity in the latter Future work should include a larger set of case is twice that of Case 4. This generally means that experimental data now that many of the issues regarding the rate of water catch on the surface is almost doubled, the IR scanner and packaging of the anti-icing system which explains why more runback froze downstream.

have been addressed and mostly resolved. This is vital in The freezing also occurred earlier (shorter distance from the validation of computational simulation computer stagnation) within the heated region. The reason for this codes. Furthermore, the details of the water and ice is the increase in the heat transfer coefficient to the behavior at the freezing point should be carefully looked ambient. Thus, the heater in zone E was turned off since at to determine the temperature behavior of the ice/water it would not help with the power density settings of mixture, and the distance over which the frozen rivulets 20 W/in 2 which were used in Zone A and Zone C. In extend and to what thickness for a given exposure time.

this case, higher power densities would be required.

However, in Case 4, the runback froze beyond the heated Acknowledgments zones. In both cases, it is recommended that the heated zone be extended a little further downstream. This is This work was done while the first author held a specially true for high LWC and freestream velocity, National Research Council Research Associateship at since increasing the power density alone to that required NASA Lewis Research Center. The sponsorship of these to achieve total evaporation would be extremely high two organizations is gratefully acknowledged. Also, the such that the heater mats and/or the structure could be cooperation of George Grauer from Technology damaged from the elevated surface temperature.

Dynamics Group, Inc., and Tomas Bond from NASA Finally, it should also be noted that the first tWO Lewis with the experimental tests is well appreciated.

cases, where total evaporation was achieved, represent References cloud conditions of relatively low liquid water content (LWC = 0.5 g/m 3) compared to the other cases 1. Al-Khalil, K.M., Keith, T.G., and De Witt, K.J., (LWC _- 1.0 g/m3). The lower temperature conditions "Development of an Anti-Icing Runback Model," contributed less than the high LWC to runback beyond AIAA Paper 90-0759, Jan. 1990.

the heaters. This may be seen by comparing results of Case 1 and Case 4 where the freestream temperature was 2. AI-Khalil, K.M., Keith, T.G., and De Witt, K.J., - 17.78 °C and the major difference between the two "Further Development of an Anti-Icing Runback cases was the high LWC in Case 4 where the runback Model," AIAA Paper 91-0266, Jan. 1991.

froze.

3. AI-Khalil, K.M., Keith, T.G., and De Witt, K.J., 6. Hansman, R.J., Breuer, K., Hazan, D., Reehorst, A., "Development of an Improved Model for Runback and Vargas, M., "Analysis of Surface Roughness Water on Aircraft Surfaces," AIAA Paper 92-0042, Generation in Aircraft Ice Accretion," AIAA Paper 93- January 1992. 0029, Jan 1993.

4. Olsen, W., and Walker, E., "Experimental Evidence 7. Potapczuk, Mark G., "LEWICE/E: An Euler Based Ice for Modifying the Current Physical Model for Ice Accretion Code," AIAA Paper 92-0037, Jan 1992.

Accretion on Aircraft Surfaces," NASA TM-87184, 8. Ruff, G.A., and Berkowitz, B.M., "Users Manual for May 1986.

the NASA Lewis Ice Accretion Prediction Code 5. Hansman, R.J., Reehorst, A., and J. Sims, "Analysis (LEWICE)," NASA CR-185129, May 1990.

of Surface Roughness Generation in Aircraft Ice Accretion," AIAA Paper 92-0298, Jan 1992.

Table 1: Operating Conditions for the different Cases.

Case T** V_ LWC MVD Power density (Wfm 2) Comments # ('%3 (mph) (g/m 3) (ttm) Zone A Zone C Zone E on results 1 - 17.78 100 0.50 15 15.0 15.0 7.5 a 2 -6.11 100 0.50 15 15.0 15.0 5.0 a 3 -6.11 100 1.00 20 15.0 15.0 5.0 b 4 - 17.78 100 1.00 20 20.0 20.0 10.0 b 5 - 17.78 200 1.00 20 20.0 20.0 0.0 b a- Total evaporation of the impinging water was achieved.

b- Partial evaporation and freezing runback.

Flow is in the z-direction

. a_.lm a_._ It. i _. mII.lt_ O It*. a U. e It. _It. all_ e It, e It. * _l It. * _ Fig. 1: Resulting rivulets after breakup of the film in the impingement region.

ORIGINAL r-',,'._" ,9LACK AND WHITE P.HOTOGRA._H Fig. 2: Runback rivulets on a thin leading edge body in the IRT.

Fig. 3: Runback rivulets on an engine inlet cowl in the IRT.

upper surface Air flow Lower surface Fig. 4: Heater zones on the airfoil leading edge.

1.5 0.8 • * ' I ' * ' I ' ' I ' ' ' I ' ' ' -( ,4 0.7 1.0 -- V = 200 mph --- v= '50mph I /f_ [_ I--- V.= 150mph i ] 0.6 --V.= 200mph I MVD =2OWn ...... m_ 0.5 ...... .= 0.5 0.4 -0.5 0.3

c, o,oI

-1.0 0.2 -1.5 0.1 = 0.0 _0 L- I I ±. ' ' I = , , l , _ • t I I -2. " 0.2 0.4 0.6 0.8 -0.07 -0.05 -0.03 o0.01 0.01 0.03 1.0 s (m) x/c Fig. 8: Droplet collection efficiency Fig. 5: Surface pressure coefficient.

0.020 , ' o ' ' i ' ' , ' ' ' , ' • ' : i:i:::: _i. :::;':':" ":_:_i:_._ i: ::._::_ii:: :_::..

m

l--V= 200mph 0.015 I--- V== 150mph !: ', I ...... V = !00mph I: 0.010 _" _ n:

ll

cl

0.005 0.000 -0.005 B!!__:i:i" .?._!:-:'::!:::::!:i:!:!:[:!;!: "._._::!i?_..

, . l I , , , I 1 [ i l = • • l • _ ::::: _:.:::::_:;.;.:.:.:.:.:.:_: _:::_:_:::_ -0.010 0.0 0.2 0.4 0.6 0.8 1.0 x/c Fig. 6: Surface friction coefficient.

Fig. 9: An Example of IR Temperature Reading from the Airfoil Surface.

0.8 .... I ' • • I ' ' ' t ' ' ' I • • ' 30 "_--'_-'-k.' m ' ' ' , ' ' ' , ' ' ' 0.7 I[--V== 200 mph MVD = 15 p,m CASE # 1 [I- -- v = 15o mphI

...... v'; Omph A

_ 10 0.3 0.2 _. o i___nomor,oa, I ,..

l- -_ - Experiment " ,,.

I

0.1 -10

°°o.o_''-o.o5' -o.o3 -O.Ol o.0i ''o.o3

s (m) 0.00 0.02 0.04 0.06 0.08 0.10 - s (m) Fig. 7: Droplet collection efficiency (MVD = 15 IJ.m) Fig. 10: Surface temperature for Case # 1.

' ' ' I " ' ' 1 ' 1 I 1 1 ' I l I I 1 1.000 1 20 I ._ .I CASE # 3 CASE # 1 2O I00 0.800 I_m_°' ic_' I 0 80 _, 15 o _ 0.600 [-..

0 40 o 0.400 E 0.200 0 , t 1 I i _._L_._-J-_ _ 1 L : • [ t t I -5 0.000 -0 20 0.00 0.02 0.04 0.06 0.08 O. 10 0.00 0.02 0.04 0.06 0.08 0.10 - s (m) - s (m) Fig. 11: Predicted runback mass flowrate & Fig. 14: Surface temperature for Case # 3.

freezing fraction for Case # 1 (lower surface).

1.00 .... I ' 1Ti I ' ' ' ' t I " ]-T'_ 12 30 ., l,. ,_.. l... _...

CASE # 3 25 _l_ CASE # 2 0.80 20 08 0.60 0,1 xo G 15 .

o 0.40 ["" 5 E 0.20 0 I-_- Experiment I -5 0.00 -I0 ''''',.'''' '''' _ ' ' ' 0.00 0.10 0.15 0.20 0.25 0.05 0.0( 0.02 0.04 0.06 0.08 O. 10 - s (m) - s (m) Fig. 15: Predicted runback mass flowrate & Fig. 12: Surface temperature for Case # 2. freezing fraction for Case # 3 (lower surface).

30 ,:%, , , , .... , .... I''' * r_-"-'- 1.00 12 • ' ' I ' ' ' I ' _ ' I ' • ' I ' ' " ..._ q CASE#4 CASE # 2 1.0 0.80 20 < "'_ [ -- Numoncal I 0.8 o 0.60 o 0.6 o 0.40 0.2 -10 0.20 .....

0.0 -20 , _ , l L. '. t I L , t I • • t 0.00 -0.2 0.00 o.os O.lO oJ5 0.20 0.25 0.0( 0.02 0.04 0.06 0.08 0.10 - s (m) - s (m) Fig. 13: Predicted runback mass flowrate & Fig. 16: Surface temperature for Case # 4.

freezing fraction for Case # 2 (lower surface).

1.0 1.2 CASE # 4 1.0 0.8 0,8 %- 0.6 I 0.6

M

i + 0,4 O 0.4 I + 0,2 i t 0.2 O0 0.0 -0,2 0.00 0.04 0.08 0.12 0.16 0.20 - s (m) Predicted runback mass flowrate & Fig. 17: freezing fraction for Case # 4 (lower surface).

" ' ' I ' ' I ' " I * ' ' I ' " " \ CASE # 5 q -Numerical J r.)

o -5 - Experiment_ -I0 -15 -20 0.00 0.02 0.04 0.06 0.08 0.10 - s (m) Fig. 18: Surface temperature for Case # 5.

1.0 ' + ' I ' ' ' l ' ' ' I ' ' ' I ' ' ' 1.2 CASE # 5 1.0 0.8 O r 8 %- 0.6 e..)

0,6 0.4 0.4 0 i r 0.2 0.2 0.0 I .I . , | , . , 0.0 -0.2 0.00 0.02 0.04 0.06 0.08 0.10 - s (m) Fig. 19: Predicted runback mass flowrate & freezing fraction for Case # 5 (lower surface).

Form Approved REPORT DOCUMENTATION PAGE OMB No. 0704-0188 Public reporting burden for this collection of information is estimated to average 1 hour per response, including the time for reviewing instructions, searching existing data sources, gathering and maintaining the data needed, and completing and reviewing the collection of information. Send comments regarding this burden estimate or any other aspect of this collection of information, Including suggestions for reducing this burden, to Washington Headquarters Services. Directorate for information Operations and Reports, 1215 Jefferson Davis Highway. Suite 1204, Arlington, VA 22202_1302, and to the Office of Management and Budget, Paperwork Reduction Project (0704-0188), Washington, DC 20503.

1. AGENCY USE ONLY (Leave blank) 2. REPORT DATE 3. REPORTTYPE AND DATES COVERED January 1993 Technical Memorandum 4. TITLE AND SUBTITLE 5. FUNDING NUMBERS Numerical Modeling of Anti-Icing Systems and Comparison to Test Results on a NACA 0012 Airfoil WU-505-68-10 6. AUTHOR(SI Kamel M. AI-Khalil and Mark G. Potapczuk 8. PERFORMING ORGANIZATION 7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES) REPORT NUMBER National Aeronautics and Space Administration Lewis Research Center E-7498 Cleveland, Ohio 44135-3191 10. SPONSORING/MONITORING 9. SPONSORING/MONITORING AGENCY NAMES(S) AND ADDRESS(ES) AGENCY REPORT NUMBER National Aeronautics and Space Administration NASA TM- 105975 Washington, D.C. 20546-0001 AIAA-93-0170 11. SUPPLEMENTARY NOTES Prepared for the 31st Aerospace Sciences Meeting and Exhibit sponsored by the American Institute of Aeronautics and Astronautics, Reno, Nevada, January 11-14, 1993. Kamel M. AI-Khalil, National Research Council-NASA Research Associate at Lewis Research Center. Mark G. Potapczuk, NASA Lewis Research Center. Responsible person, Kamel M. AI-Khalil, (216) 433-4601.

12b. DISTRIBUTION CODE 12S. DISTRIBUTION/AVAILABILITY STATEMENT Unclassified - Unlimited Subject Categories 02 and 03 13. ABSTRACT (Maximum 200 words) A series of experimental tests were conducted in the NASA Lewis IRT on an electro-thermally heated NACA 00I 2 airfoil.

Quantitative comparisons between the experimental results and those predicted by a computer simulation code were made to assess the validity of a recently developed anti-icing model. An infrared camera was utilized to scan the instantaneous temperature contours of the skin surface. Despite some experimental difficulties, good agreement between the numerical predictions and the experiment results were generally obtained for the surface temperature and the possibility for the runback to freeze. Some recommendations were given for an efficient operation of a thermal anti-icing system.

15. NUMBER OF PAGES 14. SUBJECT TERMS Anti-icing; Ice protection 16. PRICE CODE 20. LIMITATION OF ABSTRACT 19. SECURITY CLASSIFICATION 17. SECURITY CLASSIFICATION 18. SECURITY CLASSIFICATION OF REPORT OF THIS PAGE OF ABSTRACT Unclassified Unclassified Unclassified Standard Form 298 (Rev. 2-89) NSN 7540-01-280-5500 Prescribed by ANSI Std. Z39-1B 298-102