An Experimental Study of Dynamic Ice Accretion Process on Aero-Engine Spinners

AIAA 2017-0551 AIAA SciTech Forum 9 - 13 January 2017, Grapevine, Texas 55th AIAA Aerospace Sciences Meeting

An Experimental Study of Dynamic Ice Accretion Process on Aero-engine Spinners

Linkai Li1, Hui Hu2() Department of Aerospace Engineering, Iowa State University, Ames, Iowa, 50011

Ice accretion on aero-engine spinners has posed significant safety and performance concerns when aircraft operate in cold weather. In the present study, a comprehensive experimental study was conducted to investigate the transient ice accretion process on a rotating aero-engine fan with different spinner profiles (i.e., conical, coniptical, and elliptical). The experiments were performed in the Iowa State University Icing Research Tunnel (ISU-IRT) with a scaled aero-engine fan model operated under a variety of icing conditions. The transient details in the unsteady icing processes were resolved using a high-speed imaging technique, while the input power required to drive the fan at constant rotation rate was measured during the ice accretion process. The experiments demonstrated how varying the spinner profiles modifies the ice accretion process. It was found that the elliptical spinner is more sensitive to ice accretion, as indicated by the more dramatic increase in the required input power when ice accretes.

Nomenclature δ = blade trailing edge thickness, mm Subscripts θ = half apex angle of spinner, ° Δ = ice thickness, mm 0 = initial value ε = threshold used for ice boundary detection f = fan η = mechanical efficiency of the brush-less motor i = image index ρ = density, kg/m3 ins = instantaneous value σ = standard deviation of boundary detection, mm max = maximum value A = cross sectional area of test section, m2 no-ice = no ice status D = diameter, mm ratio = non-dimensional ratio HE = standard exposure distance, nmi rime = rime ice Ii = ith image in the high-speed image sequence s = spinner J = advance ratio of aero-engine fan water = sprayed water L = length, mm ∞ = freestream LWC = liquid water content, g/m3 MVD = mean volume diameter, μm N = number of blades P = input power of brush-less motor, W R = rotation speed, rpm t = time, s Downloaded by IOWA STATE UNIVERSITY on February 13, 2017 | http://arc.aiaa.org DOI: 10.2514/6.2017-0551 T = temperature, ℃ U = velocity, m/s XLE = ice boundary coordinate at X direction

I. Introduction Aero-engine icing is still a common aviation danger for aircraft operation in cold weather. When aircrafts flies in a cold climate, some of super-cooled droplets would impinge and freeze on the exposed surfaces of aero-engines, like the inlet tips, spinners and fan rotor blades. Ice accretions on such components can significantly degrade the

1 Graduate Student, Department of Aerospace Engineering, AIAA Student Member 2 Martin C. Jischke Professor, Department of Aerospace Engineering, AIAA Associate Fellow, Email: [email protected]

Copyright © 2017 by the American Institute of Aeronautics and Astronautics, Inc. The U.S. Government has a royalty-free license to exercise all rights under the copyright claimed herein for Governmental purposes. All other rights are reserved by the copyright owner.

aerodynamic performances of aero-engines that may cause an unstable compressor operation. Meanwhile, the ice accumulated over the fan blades and spinner would result in imbalance of the rotor, which would cause serious mechanical vibrations. More problematically, the ice shedding from the inlet, spinner or rotor blades may damage the fan rotor and the components behind the fan, even be suck into the core of the engine that cause stall and surge.1 Although many efforts have been devoted to icing-related issues on aero-engine since 1940s,2 further studies are stilled needed for a better understanding about the underlying physics pertinent to the icing phenomenon, which would lead to more effective and robust anti-/de-icing strategies that ensure safer and more efficient operation of aero-engine in cold weather. The spinner of the aero-engine is more sensitive to ice accretion due to its relative low spinning velocity and larger imping area for the super-cooled water droplets. A number of previous studies have been conducted in recent years to investigate the icing phenomena on aero-engines, and some anti/de-icing strategies have been suggested as well. Bidwell et al. used NASA’s LEWICE3D code to examine the ice issues on Energy Efficient Engine (E3), and found that an increased water drop size would result in an increased impingement rates on the spinner and fan.3 Dong et al. conducted a numerical study to investigate the ice accretion on rotating aero-engine spinner and found that rotating speed has light effect on the ice shape, but the temperature of freestream would affect the thickness of ice over the spinner.4 Wang et al. studied the ice accretion and shedding on a rotating spinner in an icing wind tunnel, they found that a mixture of different styles of ice grow like a feather on the surface and shed eventually.5 Dong and Hu conducted both numerical and experimental studies to evaluate a newly-developed hot-air anti-icing system applied in a full-scale cone model and they found that the hot-air film anti-icing method works well under designed icing conditions.6,7 Gilchrist et al. invented a nose cone anti-icing system using a rotating heat pipe filling with water or ethanol, which could improve the nose cone anti-icing performance and provide more feasibility.8 Although these methods are effective for anti/de-icing to a certain extent, the heating air or liquids from low-pressure compressor (LPC) or high-pressure compressor (HPC) would inevitably cause performance reduction of the aero-engine. While most of the previous studies focused on ice shape prediction, heat transfer process and performances of anti/de-icing system. Very few can be found in literature to address the effect of spinner’s geometry shape on the dynamics of ice accretion and aerodynamic performance of the engine fan. The shape of the spinner would not only affect the aerodynamic efficiency of the fan, but also determine the tendency for ice accretion on the spinner, which can be leveraged to design more effective anti/de-icing systems for aero-engine. Three typical spinner shapes, which are elliptical, conical and coniptical, are widely used in aero-engines. As described in the chapter of Linke-Diesinger’s,1 although the elliptical spinner is easier to accumulate ice at a lower humidity level in comparison to other spinner designs, it has a better aerodynamic efficiency. Oppositely, the conical spinner requires higher level of humidity to accumulate ice. Its aerodynamic efficiency, however, is not as high as the elliptical one. Hu et al. investigated the effect of spinner angle on ice accretion of a scaled rotating conical spinner, and demonstrated that the tip angle of the spinner would significantly affect the ice accretion over the spinner.9 Both these previous studies, however, lack enough quantitative measurement data to support the conclusions and the fundamental physics behind experimental observations is still unclear. In this study, a comprehensive experimental investigation was conducted to examine the dynamic transient ice accretion process over different kinds of aero-engine fan spinners. The experiments were performed in an icing research tunnel available at Iowa State University (i.e., ISU-IRT) with a scaled aero-engine fan model under different controlled icing conditions. In addition to recording the transient behavior of water runback and dynamic ice accretion Downloaded by IOWA STATE UNIVERSITY on February 13, 2017 | http://arc.aiaa.org DOI: 10.2514/6.2017-0551 process on the rotating spinners by using a high-speed imaging system, instantaneous input power of the driving motor was acquired through a DAQ system. From the acquired time-resolved image sequences of icing events, the important microphysical processes, including the impingement area of the super-cooled water droplets and ice accretion shape on the rotating spinner, were revealed in detail. Moreover, based on the high-speed images quantification method from Waldman and Hu,10 one images processing technique used to identify the ice profiles of rime ice accretion and extract the development of ice thickness on the spinners is demonstrated in present study.

II. Experiments

A. Aero-engine Fan Model The model aero-engine fan used in the present study is designed based on the Boeing 18-inch fan rig11 and its design parameters are summarized in Table 1.

Table 1. Design parameters of the aero-engine fan model

Fan Diameter, Df (mm) 200 Hub/Tip Ratio 0.402

Spinner Diameter, Ds (mm) 80 Aspect Ratio 1.65 Number of Blades, N 18 Tip Max Thickness/Chord 0.025 Thickness of trailing edge, δ (mm) 0.18 Hub Max Thickness/Chord 0.100

Figure 1 shows the schematic of the three kinds of aero-engine spinners examined in the present study. The test models are made of a hard plastic material (i.e., VeroWhitePlus, RGD835, manufactured by Stratasys, Inc.) by using a rapid prototyping machine (i.e., 3D printer) that builds the model layer-by-layer with a resolution of about 25 lm. The external surface of the test models was processed with fine sandpaper (i.e., up to 2000 grit) to achieve a very smooth, glossy finish.

Figure 1. Design of three types of spinners and the aero-engine fan tested in the present study B. Experimental Setup and Instruments The experimental study was performed in ISU-IRT located in the Aerospace Engineering Department of Iowa State University. The ISU-IRT, which was originally donated by UTC Aerospace System (formerly Goodrich Corporation) to Iowa State University, is a newly refurbished research-grade multifunctional icing research tunnel.10 The test section is 2.0 m long with 400mm (height) × 400 mm (width) in cross section. The walls of the test section are optically transparent. ISU-IRT has a capacity of generating a maximum wind speed of 60 m/s and lowest airflow temperature of -25°C. The turbulence level of the oncoming airflow at the entrance of the test section was found to be about 3.0 %, as measured by a hot wire anemometer. An array of pneumatic atomizing spray nozzles was installed at the entrance of the contraction to inject micro-sized water droplets (10~100 μm in size) into the airflow. By adjusting the water flowrate through the spray nozzles, the liquid water content (LWC) level in the tunnel can be adjusted as required; thus, ISU-IRT can be run over a wide range of icing conditions to duplicate various atmospheric icing

Downloaded by IOWA STATE UNIVERSITY on February 13, 2017 | http://arc.aiaa.org DOI: 10.2514/6.2017-0551 phenomena (i.e., from relatively dry rime ice to extremely wet glaze ice conditions). Figure 2 shows the diagram of the experimental setup used in the present study. The engine fan model with rotor blades and spinner is installed in the center of the test section, and the rotor disk of the fan is set to be normal to the freestream of the incoming airflow.

Figure 2. Experimental setup to study ice accretion process on aero-engine fan spinner In present study, a direct current power supply (BK PRECISION 1692) was used to power a brush-less motor (Scorpion, SII-4020-420KV) to drive the fan model, and an ESC (Scorpion, Commander 15V 60A) was used to adjust the rotational speed of the fan by means of changing the duty cycle of PWM control signal. During the experiment, the rotation speed of fan was measured by using a tachometer (MONARCH PLT200) that generates a pulse signal to a digital delay generator (BNC 577) to trigger the high-speed imaging system to achieve “phase-locked” imaging of the aero-engine spinner. Also, the rotation speed was kept constant during the icing process through a PID feedback control program based on LabView platform. The electric current in the electric circuit was measured by using a current transducer (Crmagnetics, CR5410-50), which was then used to calculate the input power of the brush-less motor. A high-speed imaging camera (PCO.dimax S4 with maximum 2016 pixels × 2016 pixels in resolution) with a 24mm lens (Nikon, 24 mm Nikkor 2.8D) for a larger view field or a 105mm lens (Nikon, 105mm Micro Nikkor 2.8D) for a smaller view field was mounted at one side of the test section to record the ice accretion process. To provide high quality imaging data, low-flicker illumination was provided by a pair of 100 W Studio-LED lights (RPS Studio Light, Model RS-5610 and RS-5620). An aluminum tube with an airfoil shape in cross section, which was used to support the fan model, was connected to the test section wall.

III. Experimental Conditions Selection From the airworthiness certification’s point of view, each commercial aircraft or aero-engine with an icing Downloaded by IOWA STATE UNIVERSITY on February 13, 2017 | http://arc.aiaa.org DOI: 10.2514/6.2017-0551 protection system must satisfy the strict icing operational requirements included in 14 CFR Part 25 (Airworthiness Standards: Transport Category Airplanes) or Part 33 (Airworthiness Standards: Aircraft Engines).12-14 And the in- flight icing conditions, which are used to design ice protection systems for airframe or aero-engine, are defined in 14 CFR Part 25, Appendix C and Appendix O (Super-cooled Large Drop Icing Conditions). Traditionally, continuous maximum conditions have been applied to airframe ice protection and intermittent maximum conditions have been applied to engine ice protection.15 In order to simulate a natural icing environment on the atmosphere, the icing conditions used for present study are selected based on the “intermittent maximum” envelope--the fourth figures of document 4 CFR Part 25 Appendix C,12 which is shown in Fig. 3. In order to study the ice accretion process over aero- engine spinner under glaze ice, mixing ice and rime ice, three typical test conditions (i.e., red stars) are selected from Fig. 3, which are also listed in Table 2.

Figure 3. Intermittent maximum atmospheric icing conditions from 14 CFR Part 25 Appendix C (Liquid water content vs mean effective drop diameter)

Table 2. Icing test conditions

T∞ LWC t R0 U∞ MVD Mwater Case Ice style (°C) (g/m3) (s) (rpm) (m/s) (µm) (g) 1 -5°C 2.4 135 Glaze 2 -15°C 2.0 163 2500 15 10~50 800 Mixing 3 -10°C 1.1 308 Rime

In present study, the mass of water sprayed into the icing wind tunnel for each case is a constant, which is set as 0.8 Kg. Therefore, the icing time t for different LWC case is determined by Eq. (1): M water = U ∞A LWC  t= Const (1)

where U∞ is the velocity of freestream, A is the area of cross section of test section, LWC is liquid water content, and t is the icing time. Meanwhile, the aero-engine fan is simulated to running at the cruise status with a typical advanced ratio (J) equal to 1.8 based on the specification of a typical CFM International CFM56 turbofan aero-engine.16 The rotation speed of the aero-engine fan and the velocity of freestream are determined by Eq. (2) with the pre-determined advanced ratio.

U J = ∞ (2) RD0 f Downloaded by IOWA STATE UNIVERSITY on February 13, 2017 | http://arc.aiaa.org DOI: 10.2514/6.2017-0551 where R0 is the expected rotation speed of aero-engine fan and Df is the diameter of the fan rotor.

IV. Results and Discussions A. The Ice Accretion Process In the present study, a typical glaze ice accretion trail is examined firstly. Figure 4 shows the time-resolved image 3 sequences of ice accretion over the spinners under this condition, i.e., T∞= -5 ºC, LWC=2.4 g/m . The t is the icing time while t= 0s means the super-cooled water droplets first impact with the model surface, which is determined by the icing videos. From the images shown in Fig. 4 (a), for the conical spinner, the super-cooled water droplets impinge the spinner tip area firstly but are not frozen immediately. A water film is developed on the surface and the water runback rivulets can be observed from the snapshots images. When the impinged water is driven downstream along the surface, the ice start to grow like feathers standing around the spinner with the combination of aerodynamics shear stress and inertial

force, i.e., centrifugal force. During the icing process, the lengths of these feather-like icicles grow dramatically. After 135 seconds of icing process, these feather-like icicles almost occupy the three-quarters area of the surface, i.e., from X=0.75 Ds to X=0 Ds as shown by the red dash lines in the last image of Fig. 4 (a). Figure 4 (b) presents the ice accretion over the coniptical spinner. Similarly, the water droplets impinge the tip area and form a thin water film that is developed to the rivulets. The ice accretes on the tip area exhibits characteristics in agreement with the phenomenon shown on previous conical spinner, which are both rough ice buildup. Behind the tip area, the feather-like icicles start to grow around the surface as well. However, there is no obvious ice accretion near the root area of coniptical spinner, and most of the icicles only accumulate at the range from X=0.8 Ds to X=0.4 Ds, shown by the red dash lines in the last image of Fig. 4 (b). Its range is shorter than the feather-like icicles area of conical spinner, but the position is closer to the tip. Figure 4 (c) shows the ice accretion process of the elliptical spinner, the similar feather-like icicles start to build up at the very front area of the spinner, i.e. from X=0.9 Ds to X =0.6 Ds, shown by the red dash lines in the last image of Fig. 4 (c). In addition, the ice accreted at tip area of elliptical spinner also become rough like previous results. For the rest area, i.e., X=0.6 Ds to X=0 Ds, no apparent ice is observed, which is similar to the characteristics of coniptical spinner.

(a) Conical

(b) Coniptical Downloaded by IOWA STATE UNIVERSITY on February 13, 2017 | http://arc.aiaa.org DOI: 10.2514/6.2017-0551

3 Figure 4. Snapshots of ice accretion over the spinners (U∞=15 m/s, T∞= -5 ºC, LWC=2.4 g/m , J=1.8)

In addition, during the ice accretion process under this condition, ice shedding events of conical and coniptical spinners are recorded by the high-speed image system. Figure 5 demonstrates one example of the ice shedding event for the aero-engine fan model with coniptical spinners. From the videos, it founds that a pieces of icicles sheds from the area near the ice accretion limitation on coniptical spinner, highlighted by the red arrows at Fig. 5.

3 Figure 5. Ice shedding images pair of the coniptical spinners (U∞=15 m/s, T∞= -5 ºC, LWC=2.4 g/m , J=1.8) After 135 seconds of icing, the ice buildup on the aero-engine fan model with different spinners are shown in Fig. 6. As shown in Fig. 6, not only the ice accretion on the spinner, but also the ice accumulation over the blades are clearly revealed. For the spinners, the buildup area of feather-like icicles are varied with the shape change of spinners, which are consistent with the time sequences shown in Fig. 6. And more notably, much ice accretes over the leading edge area of blades and the ice grow along the radius direction even beyond the tip of blades, which is called lobster- tail ice.17

Downloaded by IOWA STATE UNIVERSITY on February 13, 2017 | http://arc.aiaa.org DOI: 10.2514/6.2017-0551

(a) Conical (b) Coniptical (c) Elliptical

Figure 6. Final ice buildup on the aero-engine fan model with different spinners (U∞=15 m/s, T∞= -5 ºC, LWC=2.4 g/m3, J=1.8) In short, under this relative warmer and more humid condition, glaze ice accretion is observed on both spinner and blades. The feather-like icicles grow on specific area of the spinners and lobster-tail ice accumulate on the blades. As the aerodynamic shape of both spinners and blades is significantly damaged, the raise of turbulence and even a large scale flow separation before and after the aero-engine fan rotor is inevitable. Meanwhile, ice shedding events are also highly expected under this icing condition.

3 Under lower temperature with humid condition, i.e., for the test case of T∞= -15 ºC, LWC=2.0 g/m , as shown in Fig. 7, the ice accretion process is different between the spinners. In this case, for the conical spinner, shown in Fig. 7 (a). When the water droplets impinge the surface, they are frozen immediately and no water film is observed from the images. The semitransparent rough ice accumulates around the whole spinner surface and no feather-like icicles grow on the spinner. In this case, the ice profile is kept as close as the initial aerodynamic shape of the conical spinner.

(a) Conical

(b) Coniptical Downloaded by IOWA STATE UNIVERSITY on February 13, 2017 | http://arc.aiaa.org DOI: 10.2514/6.2017-0551

3 Figure 7. Snapshots of ice accretion over the spinners (U∞=15 m/s, T∞= -15 ºC, LWC=2.0 g/m , J=1.8)

Figure 7 (b) presents the ice accretion process of the coniptical spinner. When the water droplets impinge the surface, a very thin water film is developed, which can be observed from the video. However, this water is frozen quickly and it doesn’t develop to the feather-like icicles during the ice accretion process. At the same time, the semitransparent rough ice grows continuously around the spinner before X=0.4 Ds. In other word, the ice profile is also kept as close as the initial aerodynamic shape of the coniptical spinner. Figure 7 (c) presents the ice accretion process of the elliptical spinner under this lower temperature condition. From the snapshots and icing video, when the water droplets impinge on the surface, a thin water film is developed

and then breaks into rivulets quickly, which is very similar to glaze ice condition. As a result, the feather-like icicles build up on the elliptical spinner surface. These feather-like icicles are accreted from X =0.9 Ds to X=0.6 Ds area shown in the last image of Fig. 7 (c). Compared to the feather-like icicles on case 1, however, these feather-like icicles are much shorter and stronger, while the color of icicles varies from transparent to white during ice accretion process.

(a) Conical (b) Coniptical (c) Elliptical

Figure 8. Final ice buildup on the aero-engine fan model with different spinners (U∞=15 m/s, T∞= -15 ºC, LWC=2.0 g/m3, J=1.8) Similar to case 1, after 163 seconds of icing, the ice buildup on the aero-engine fan model with different spinners are shown in Fig. 8. For the conical and coniptical spinners, no any feather-like icicles grow around the surface. For the elliptical spinner, the shorter and stronger feather-like icicles can be observed clearly. These observations are also consistent with the time sequences shown in Fig. 7. Moreover, the ice accretes over the leading edge area of blades shows the characteristic of mixing ice, i.e., both transparent and white color ice are observed there. Also, the ice grows along the radial direction only beyond the tip of blades a little, which is called lobster-tail ice but relatively shorter than the lobster-tail ice of case 1. In short, under this relative colder and more humid condition, mixing ice accretion is observed on both spinner and blades. The shorter and stronger feather-like icicles only grow on specific area of the elliptical spinners and shorter lobster-tail ice accumulates on the blades. As the aerodynamic shape of the conical and coniptical spinners and blades is slightly damaged, the penalty of aerodynamic performance is lighter in the expectation. But for the elliptical, the raise of turbulence and even flow separation before and after the aero-engine fan rotor is still inevitable. 3 Under moderate temperature with dryer condition, i.e., for the test case of T∞= -10 ºC, LWC=1.1 g/m , as shown in Fig. 9, the ice accretion process is similar for all the spinners. In this case, when the water droplets impinge the surface, they are frozen immediately and no water film is observed from the images. For the conical spinner, the white Downloaded by IOWA STATE UNIVERSITY on February 13, 2017 | http://arc.aiaa.org DOI: 10.2514/6.2017-0551 rime ice accumulates around the whole spinner surface and the ice profile is kept as close as the initial aerodynamic shape of the conical spinner.

(a) Conical

(b) Coniptical

3 Figure 9. Snapshots of ice accretion over the spinners (U∞=15 m/s, T∞= -10 ºC, LWC=1.1 g/m , J=1.8)

Figure 9 (b) and (c) presents the rime ice accretion process of the coniptical and elliptical spinner. Similar to

Downloaded by IOWA STATE UNIVERSITY on February 13, 2017 | http://arc.aiaa.org DOI: 10.2514/6.2017-0551 conical spinner, when the water droplets impinge the surface, the water is frozen quickly and it doesn’t develop to the feather-like icicles during the ice accretion process. At the same time, the white rough ice only grows continuously around the spinner before X=0.4 Ds for coniptical spinner and before X=0.6 Ds for elliptical spinner, shown by the red dash lines in the last image of Fig. 9 (b) and (c). Similarly, after 308 seconds of icing, the ice buildup on the aero-engine fan model with different spinners are shown in Fig. 10. For the all the spinners, no any feather-like icicles grow around the surface. The ice accrete on the surface are white and rough. Meanwhile, the ice accretes over the leading edge area of blades also shows the characteristic of rime ice, i.e., white color ice is observed there. Also, the little lobster-tail are also accumulating on the fan blade tips. Consequently, under this moderate colder and dryer condition, rime ice accretion is observed on both spinner and blades. The aerodynamic shape of all the spinners are not changed obviously, which cannot have a significant impact on the performance of the aero-engine fan

(a) Conical (b) Coniptical (c) Elliptical

Figure 10. Final ice buildup on the aero-engine fan model with different spinners (U∞=15 m/s, T∞= -10 ºC, LWC=1.1 g/m3, J=1.8) B. Images Digitizing of Rime Ice Accretion Although the high-speed video of the ice accretion provides a qualitative view of the icing process, which demonstrate some insight into the physics of ice formation. However, quantitative measurements of the ice features are highly expected to investigate the details of ice accretion. One method is presented in this section to identify the ice profile under rime ice condition (i.e. case 3) and deduce the development of ice thickness at different radius positions from the time-resolved image sequences of icing process. This method can only be applied to the ice accretion without feather-like icicles buildup, which means two prerequisite assumptions are satisfied. One is that ice accretion are both rime ice on the spinner surface without icicles buildup. The other is that ice thickness on the position with same radius are same, which means the ice accretion on the spinners are axisymmetric. Since the icing condition would determine the style of ice accretion, one moderate cold temperature and lower 3 liquid water content condition is studied here, i.e., the test case of T∞= -10ºC, LWC=1.1g/m . Figure 11 shows the surface ice growth as a function of time. The ice boundary (i.e., red solid line shown in Fig. 11) is determined by identifying the first location in front of the spinner with meaningful change in the pixel count compared to a threshold based on the background noise.10 For each Y position on the spinner, ii= first( () x > ε ) XILE (3) The value of ε is determined by the background noise of the acquired images. Practically, in present study, ε is chosen as the 16 times standard deviations of the typical image noise. This image noise is extracted by calculating the root-mean-square pixel intensity fluctuations of the 20 images’ background area before the beginning of icing. The Downloaded by IOWA STATE UNIVERSITY on February 13, 2017 | http://arc.aiaa.org DOI: 10.2514/6.2017-0551 average standard deviation of pixel intensity was about 0.0025 count, therefore, ε=0.04.

3 Figure 11. The ice profile time sequences of conical spinner (U∞=15 m/s, T∞= -10ºC, LWC=1.1g/m , J=1.8)

(a) Conical (a) Conical

(b) Coniptical (b) Coniptical Downloaded by IOWA STATE UNIVERSITY on February 13, 2017 | http://arc.aiaa.org DOI: 10.2514/6.2017-0551

Figure 12. The quantitative ice profiles of different Figure 13. Development of the ice thickness at spinners under rime ice condition (U∞=15 m/s, T∞= different radius position of different spinners 3 3 -10 ºC, LWC=1.1 g/m , J=1.8) (U∞=15 m/s, T∞= -10 ºC, LWC=1.1 g/m , J=1.8)

Based on the intensity difference image processing techniques shown in Fig. 11, the change of the ice profiles along the spinners are shown in Fig. 12. It is shown that, for the conical spinner, the ice thickness along the bus line is more uniform. The ice thickness at the tip area i.e., Y=0 Ds is close to the ice thickness at the root area i.e., Y=0.4 Ds. However, for the coniptical and elliptical spinners, the ice thickness along the bus line decreases proportionally from the tip to root. At the tip, the ice thickness is the largest and it decreases gradually from Y=0.0 Ds to Y=0.4 Ds. From Y=0.4 Ds to Y=0.5 Ds range, the profiles change a little, which means there is an ice accretion limitation on these two spinners surface. Figure 13 presents the long-time growth of ice thickness at different radius positions of these three types of spinners, which are extracted from the ice profiles shown in Fig. 12. As shown in Fig. 13, the accumulated ice thickness of these spinners grows linearly with time. Practically, for the conical spinner, the average ice growth rate at the tip is the highest, which is about 0.0147 mm/s. With the increase of radius, the average ice growth rates change a little. For the coniptical spinner, the average ice growth rate at the tip is still the highest, i.e., 0.0123 mm/s. However, with the increase of radius, the average ice growth rate decrease rapidly, i.e., at Y=0.4 Ds, its value is only 0.005 mm/s. For the elliptical spinner, the average ice growth rate at the tip is about 0.009 mm/s. From Y=0 Ds to Y=0.3 Ds, there is no apparent change of the average ice growth rate. But at Y=0.4 Ds, the average ice growth rate drops sharply, which is only about 0.0053 mm/s. C. The Measurement of Input Power of Driving Motor In present study, the engine fan model is driven by a brush-less motor, which uses electricity as energy source. It is known that the power of the motor equals the torque multiply rotation speed, shown in Eq. (4). As the mechanical efficiency of the brushless motor and rotation speed are constant in present study, the input power and the torque of the motor are positively correlated. In other words, the change of input power would be caused by the ice accretion over the model and drag increase of the aero-engine fan model. Pη= τω (4) where P is the input power, η is the efficiency of the motor, τ is the torque and ω is the angular rotation speed. Figure 14 presents the input power ratio of the aero-engine fan model during the icing process under different icing conditions. Here, input power ratio is a non-dimensional ratio of instantaneous input power of the motor over average input power of the motor without icing buildup, given in Eq. (5):

= P ins P ratio (5) P no− ice

If Pratio is equals to one that means the situation of the model doesn’t change compared to no ice situation. In other words, the ice accretion influence can be ignored. From Fig. 14 (a), it is found that under glaze ice condition, the effect of ice accretion on input power of the motor is significant. For the aero-engine fan model with conical spinner, after 135 seconds of icing, the input power increases about 237%. For the aero-engine fan model with coniptical spinner, after 135 seconds of icing, the input power increases about 250%, which is littler higher than the conical ones. Most seriously, the input power of the aero-engine fan model with elliptical spinner increase drastically up to 336%, which is far larger than the conical and coniptical ones. In this case, the dramatic change of input power demonstrates that most of the energy is used to compensate the drag and weight increase caused by ice accretion instead of generating thrust and compressing air flow through the fan. Consequently, the performance of the aero- engine fan model degrades a lot during this icing process.

Downloaded by IOWA STATE UNIVERSITY on February 13, 2017 | http://arc.aiaa.org DOI: 10.2514/6.2017-0551 Similarly, Fig. 14 (b) presents the input power ratio of aero-engine fan model under relatively colder and more humid condition. Different with previous case, the changes of input power ratio are very slight. E.g., the input power ratio curve of conical spinner remains stable and it is almost equals to one. For the coniptical and elliptical spinners, both of them show a slight increase. The increase of input power is 6.8% and 11.6% relatively. Still, the input power demand of aero-engine fan model with elliptical spinner is little higher than the aero-engine fan model with coniptical spinner. For the rime ice condition shown in Fig. 14 (c), there is no significant change of the input power ratio of aero- engine fan model. For the conical spinner, the curve keeps stable near one. For the coniptical and elliptical spinners, there are tiny increase of the input power ratio, which is 5.3% and 8.3% relatively.

(a) Case 1 (b) Case 2

VI. Conclusion In the present study, a comprehensive experimental study was conducted to investigate the transient ice accretion process over the spinner and blade surfaces of a rotating aero-engine fan. The experiments were performed in ISU- IRT with a scaled aero-engine fan model operated under a variety of icing conditions (i.e., ranged from rime to glaze ice conditions). A high-speed imaging technique was used to resolve the transient details of ice accretion over the rotating fan model with three typical spinner profiles (i.e., conical, coniptical, and elliptical). The input power required for driving the constant rate of rotation was measured simultaneously with the ice accretion process. Based on such temporally-synchronized-and-resolved measurements, the detailed ice accretion profiles were correlated with the dynamic input power data, which provided further insight into the effects of spinner profiles on the aero-engine fan

Downloaded by IOWA STATE UNIVERSITY on February 13, 2017 | http://arc.aiaa.org DOI: 10.2514/6.2017-0551 performance in such icing conditions. The time-resolved image sequences of the icing processes were digitized to extract the ice features over both the spinners and the blades of the aero-engine fan model. It is found that in glaze ice conditions, location and extent of the feather-like icicles are strongly dependent on the spinner geometry, while the long and irregular lobster-tail ice features on the fan blades are relatively unaffected by the spinner geometry. In rime ice conditions, the conical spinner exhibits a uniform thickness ice accretion, while the ice thickness distributions on the coniptical and elliptical spinners are found to decrease proportionally from the spinner tip to root. By comparing the input power changes during the ice accretion process, the elliptical spinner demonstrates the most dramatic increase in power consumption while feathers-like icicles accumulate closer to the tip area of the spinner.

Acknowledgements The authors want to thank Dr. Rye Waldman, Dr. Yang Liu, Miss. Linyue Gao, Mr. James Benson and Mr. Andrew Jordan of Iowa State University for their help in operating ISU Icing Research Tunnel (ISU-IRT) Facility.The research work is partially supported by Iowa Space Grant Consortium (ISGC) Base Program for Aircraft Icing Studies. The support of National Science Foundation (NSF) under award numbers of CBET- 1064196 and CBET-1435590 is also gratefully acknowledged.

References 1Andreas Linke-Diesinger, Systems of Commercial Turbofan Engines, Springer, 2008, Chaps. 9, DOI: 10.1007/978-3-540- 73619-6 2“Research Aircraft at NASA Glenn Research Center (1943-1958)”, NASA, http://www.nasa.gov/externalflash/NACA/. 3Bidwell, C.S., “Particle trajectory and icing analysis of the E3 turbofan engine using LEWICE3D version 3”, NASA TM 2012-217696, 2012. 4Dong, W., Zhu, J. J., Zheng, M., and Wang, R., “Numerical Study of Ice Accretion on Rotating Aero-Engine Cone,” Proceedings of the 29th Congress of the International Council of the Aeronautical Sciences, International Council of the Aeronautical Sciences Paper ICAS 2014-0035, Stockholm, 2014. 5Wang, J., Hu, Y.P., Ji, H.H., Chen, N.L., Cao, G.Z., “Experiment of ice accretion and shedding on rotating spinner”, Journal of Aerospace Power, Vol.29, No.6, 2014, pp. 1352-1357. (In Chinese) 6Dong, W., Zhu, J. J., Zheng, M., and Chen, Y., “Thermal Analysis and Testing of Nonrotating Cone with Hot-Air Anti-Icing System,” Journal of Propulsion and Power, Vol. 31, No. 3, 2015, pp. 896–903. 7Zhu, J., Dong, W., Zheng, M., and Lei, G. “Numerical Investigation of Heat and Mass Transfer on an Anti-Icing Inlet Cone”, Journal of Propulsion and Power, Technical Notes, 2016, DOI: 10.2514/1.B35875 8Gilchrist, S., Ewing, D., Ching, C.Y., “On the Design of an Aero-Engine Nose Cone Anti-Icing System Using a Rotating Heat Pipe”, Journal of Thermal Science and Engineering Applications, Vo l . 1, 2009. 9Hu, Y.P., Ji, H.H., Wang, J., Chen, N.L., Cao, G.Z., Tong, H., Pang, L.G, “Experiment on effect of cone angle on ice accretion of rotating spinner”, Journal of Aerospace Power, Vol.29, No.3, 2014, pp. 495-503. (In Chinese) 10Waldman, R., Hu, H., “High-Speed Imaging to Quantify the Transient Ice Accretion Process on a NACA 0012 Airfoil”, Journal of Aircraft, Vol. 53, No. 2, 2016, pp. 369-377. 11Ganz, U.W., Joppa, P.D., Patten, T.J., Scharpf, D.F., “Boeing 18-Inch Fan Rig Broadband Noise Test”, NASA CR-1998- 208704, 1998. 12“CFR Title 14 FAA Part 25”, U.S. Government Publishing Office, http://www.ecfr.gov/cgi-bin/text- idx?SID=954e65f17f729e85619d7f83817ff169&mc=true&node=pt14.1.25&rgn=div5#ap14.1.25_11801.o [cited 01 December 2016] 13“CFR Title 14 FAA Part 33”, U.S. Government Publishing Office, http://www.ecfr.gov/cgi-bin/text- idx?SID=18d4325f3a00f7f8d67fda954f057976&mc=true&node=pt14.1.33&rgn=div5#se14.1.33_168 [cited 01 December 2016] 14Yang, B. and Zhou, Y., “Icing Certification of Civil Aircraft Engines”, Procedia Engineering, Vol.17, 2011, pp. 603-615. 15Jeck, R.K., “Icing Design Envelopes (14 CFR Parts 25 and 29, Appendix C) Converted to a Distance-Based Format”, DPT/FAA/AR-00/30, Office of Aviation Research Washington, D.C., April 2002. 16“CFM56 Turbofan Engine”, https://en.wikipedia.org/wiki/CFM_International_CFM56 17Han, Y., “Theoretical and Experimental Study of Scaling Methods for Rotor Blade Ice Accretion Testing”, M.S. Dissertation, College of Engineering, The Pennsylvania State University, University Park, PA, 2011.

Downloaded by IOWA STATE UNIVERSITY on February 13, 2017 | http://arc.aiaa.org DOI: 10.2514/6.2017-0551