APPENDIX R. ICE SHAPES (CONTINUED) The comparison of rime and glaze ice shapes that have about the same mass is another example of how important it is to understand the ice accretion shape’s influence on the boundary-layer behavior when determining critical ice shapes. A rime ice leading edge ice shape that conforms with the profile of the surface may effectively improve the surface’s camber and aerodynamic performance. The rime may improve the local boundary-layer conditions and increase the lifting chord length. Conversely, the rime ice accretion may also thicken and weaken the boundary layer because of the rime ice roughness. This causes the boundary layer to become turbulent or separated earlier than a contamination-free boundary layer. However, the resulting increased drag and reduced maximum lift will usually be much less than that of a glaze ice shape of similar mass. The glaze ice is usually characterized by upper and lower horns that the airflow and boundary layer must negotiate. In fact, complete separation of the flow often occurs behind the horns. Also, the character of the boundary layer and its wake (for example, turbulence intensity, tendency to remain turbulent, tendency to remain attached or to separate from the surface, and the periodicity of separation from the surface) may vary. These viscous flow effects will vary, depending on:
x Reynolds number;
x The size, shape, and surface roughness of the ice accretion;
x The heights of the ice accretion protuberance and its surface roughness when compared with the surface’s characteristic length;
x The boundary-layer characteristics of the uncontaminated surface (that depend on the surface’s camber, thickness distribution, surface finish, and aoa); and x The location of the ice shape.
The boundary-layer effects resulting from the ice accretion typically reveal themselves in a small loss of lift within the normal operating envelope of the aircraft. However, there is a significant loss in maximum lift and a significant decrease in the AOA at which maximum lift occurs (figure R-8). The loss in maximum lift may affect stall-speed margins, maneuverability margins to buffet, and stall warning margins unless maneuvering airspeeds and the stall protection AOA schedules are adjusted. Significant drag increase will occur at operational angles-of-attack. The drag increase will increase with increasing AOA as the aircraft approaches maximum lift. The boundary-layer effects will adversely affect rotary wings and propeller thrusting efficiency. Changes in the contaminated airfoil’s pressure loading usually increase the airfoil’s nose-down pitching moments (more negative pitch stability). These effects are obvious in figure R-8. The increased negative stability may adversely affect airplane stability and control, stall characteristics, and trim. The boundary layer effects may also adversely affect rotary wing and propeller performance.
Figure R-15 shows the effects of Reynolds and Mach numbers on the performance of an airfoil with an ice accretion for a 36-inch NACA 23012 airfoil with and without intercycle ice. For the clean airfoil, increasing the Reynolds number favorably influences the behavior of the boundary layer. The boundary layer remains attached to the airfoil at higher angles of R-24
APPENDIX R. ICE SHAPES (CONTINUED) attack. Also, for the uncontaminated airfoil, the improved boundary-layer behavior reduces drag at higher Reynolds numbers. However, for the airfoil with a simulated ice shape, the Reynolds number effects are much less significant. Figure R-15 shows the effect of Mach number on airfoils with and without the simulated ice shape. Aerodynamic characteristics of contaminated surfaces do not always show large variation with Reynolds number, as shown in figure R-15. Applicants should provide an evaluation of Reynolds number effects when using low Reynolds number testing for the analysis of the aerodynamic effects of ice buildups.
Consider the effects of Mach and Reynolds numbers when determining the critical ice shape for a specific flight safety concern. For example, when comparing the coefficients of lift and drag for clean and contaminated airfoils, Reynolds and Mach numbers significantly affect the aerodynamic differences.
Critical ice shapes may result in reduced control surface authority, hinge moment and control force anomalies, uncommanded control surface movement, and may affect downstream control surface pressure loading. The change in surface pressure loading and periodic separation of the surface airflow may also affect surface vibration and tendencies for surface flutter.
Clean NACA 23012 airfoil NACA 23012 airfoil with Ice Shape 290 (figure R-5).
Figure R-15(a). Reynolds number effect on lift for a NACA 23012 airfoil with and without a simulated ice shape.
R-25
APPENDIX R. ICE SHAPES (CONTINUED) Re = 7.5 × 108 Ma = 0.12 Re = 3.5 × 10° Re = 10.5 x 1ởi Ma = 0.12 °, Ma = 0.12
0. 022- "0.080-
0.020-
0.070
0.018-
0.016-
0.014- C,0.012
0.010-
0.008
0.006-
0.010
0.000- 2 0 2 8 10 12 14 16 18 20 0.000- • 10 12 14 16 a (deg) a (deg) Clean NACA 23012 airfoil NACA 23012 airfoil with Ice Shape 290 (figure R-5) Figure R-15(b). Reynolds number effect on drag for a NACA 23012 airfoil with and without a simulated ice shape.
Re = 7.5 x 10°, Ma = 0.12 Re = 7.5 x 105 , Ma = 0.21 Re = 7.5 × 10 , Ma = 0.28
2.0-
0.0 -0.2- -0.4-
0.67016 4-20 2 4 6 8 10 12 14 16 18 20 5 8 16 12 14 16 a (deg) a (deg) Clean NACA 23012 airfoil NACA 23012 airfoil with Ice Shape 290 (figure R-5).
Figure R-15(c). Mach number effect on lift for a NACA 23012 airfoil with and without a simulate ice shape.
R-26
APPENDIX R. ICE SHAPES (CONTINUED) Clean NACA 23012 airfoil NACA 23012 airfoil with Ice Shape 290 (figure R-5) Figure R-15(d). Mach number effect on drag for a NAC A 23012 airfoil with and without a simulate ice shape.
Figure R-15. Reynolds and Mach Numbers Effects on Lift an d Drag for a NACA 23012 Airfoil With and Without Simulated Ice Shape 290 (Reference R2).
The mass of the critical ice accretion on aerodynamically balanced control surfaces (for example, elevator and aileron horns) may change the effectiveness of the surface’s aerodynamic balance and may adversely affect hinge moments and control forces. Ice accretion mass on surfaces may also affect the control surface’s tendency to vibrate or flutter.
Empirical aerodynamic data are typically used to determine the effects of critical ice accretions.
When used carefully, Navier-Stokes CFD computer codes may provide useful information.
The empirical aerodynamic data used should apply to the surface of interest since the aerodynamic effects of the ice accretion depend on the boundary layer’s behavior over the surface and ice accretion. Empirical data are usually recorded for specific surfaces during flight with simulated ice accretions, or in natural icing conditions. You may also use scaled model data from dry air wind tunnels at the proper Reynolds and Mach numbers. You should use these data carefully since reproduction of ice shapes at small scales is difficult and since data corrections beyond test Reynolds and Mach numbers may be required to get results applicable to flight. If you choose to use scale models, you must verify the wind tunnel model’s performance adequately matches the performance that would occur at full-scale.
You may also have applicable empirical ice accretion effects data from similar or predecessor aircraft designs. Information available in the technical literature may be applicable. However, R-27
APPENDIX R. ICE SHAPES (CONTINUED) data examination is necessary to show the data applies to the aircraft design under consideration.
Limited flight Reynolds number data is available. A survey of ice accretion effects, drawn from
the open literature, and discussions of critical ice shapes are in Reference R6. Also, Reference R7 provides added information on the aerodynamic effects of ice accretions. You may consider the ranking of ice shape features for their adverse aerodynamic effects provided in section R.6 of this appendix when determining critical ice shapes.
You may use verified CFD codes to determine the aerodynamic characteristics of critical ice accretions. Because the air’s viscosity highly affects the airflow behavior over the complex ice accretion shapes and roughness, CFD codes based on estimated solutions of the viscous flow Navier-Stokes equations, if carefully used, may provide useful information. Take care to ensure that the gridding of the surface geometry and flow field is acceptable and the wake turbulence models used produce acceptable estimates of the boundary layer behavior and its wake. The behavior of the boundary layer and its wake strongly influence the code’s ability to predict maximum lift and drag levels accurately. Viscous flow CFD codes based on Reynolds-Averaged Navier-Stokes (RANS) equations (such as NSU2D and WIND), with enough attention given to gridding of the ice accretion shape and roughness and the flow field, provide good information when there is little or no flow separation on the aircraft (at low angles-of-attack).
However, at attitudes where airflow separation occurs (such as near maximum lift) the RANS codes may not robustly and accurately predict lift, drag, and pitching moments. Various causes contribute to this failure. One of those causes is the inability of RANS codes to capture the vortex shedding physics of airflow separation. This flow separation is of particular interest for complex ice shapes and roughness. Figure R-16 shows this RANS codes limit for an NLF 0414 airfoil with a simulated upper surface ice shape horn. An ice shape that displays insignificant vortex shedding at maximum lift may produce good correlation with empirical data using an RANS code. The code may provide misleading information for other ice shapes that display significant vortex shedding. Figure R-17 shows the periodic vortex shedding predicted by a Detached Eddy Simulation (DES) code for the NLF airfoil. (Reference R8 contains added information about DES codes.) Also, because of the vortex shedding and its effects on the surface pressures, values of lift, drag, and pitching moment may become time-dependent. The code may fail to converge on a solution that provides acceptable time-averaged aerodynamic coefficients.
R-28
APPENDIX R. ICE SHAPES (CONTINUED) Steady 2D RANS velocity
contours; note nonphysical “converged” separation bubble Instantaneous DES velocity
contours; note unsteady large eddy shedding Figure R-16. Comparison of Computed Flow Fi elds Using Reynolds-Averaged Navier-Stokes (RANS) and Detached Eddy Simulation (DES) Navier-Stokes CFD Codes (NLF 0414 airfoil at an AOA of 7˚ and a Reynolds number of 1.8x106 with a 0.034 k/c simulated ice shape horn) (Reference R8).
Figure R-17. Time-Dependent Vortex Shedding (Vorticity Contou rs) From an NLF 0414 Airfoil With a
0.034 k/c Simulated Ice Shape Horn Calculated Using a DES Navier-Stokes CFD Code (Reference R8).
R-29
APPENDIX R. ICE SHAPES (CONTINUED) Flow separation is 3D, with lateral flow movement. This is especially true for surfaces that present significant cross-flow, such as on swept finite wings. The 3D character of flow separation raises questions about the accuracy of 2D Navier-Stokes flow solutions when significant boundary-layer separation exists.
Figure R-18 shows the differences in upper surface flow separation and vortex shedding at the same time instant between 2D and 3D DES Navier-Stokes solutions. This solution is for an 18-inch chord NACA 23012m model with a 0 .25-inch, quarter-round simulated ice shape. The ice shape is attached at ten percent of the airfoil’s chord. To predict aerodynamic performance confidently near and at maximum lift, you may need to use Direct Numerical Simulation (DNS ) Navier-Stokes CFD codes that fully solve the Navier-Stokes equations or Large Eddy Simulation (LES) codes. Reference R9 contains definitions and further discussions on RANS, LES, and DNS Navier-Stokes CFD codes. The cost of carrying out LES or DNS solutions for complex geometry and the need to consider 3D flow effects are high. Perhaps this restricts their use when you are investigating several candidate critical ice shapes. Using DES Navier-Stokes CFD codes (a hybrid between RANS and LES codes) may offer good, cost-effective information, however , these codes have not been fully evaluated for predicting flow behavior for surfaces with ice accretion shapes and roughness (Reference R9).
Figure R-18. Vorticit y Contours for 2D and 3D DES Navier-Stokes Computational Fluid Dynamics Code Analysis of the Flow Field Around a NACA 23012m Airfoil With a k/c = 0.0139 Forward- Facing, Quarter-Round Simulated Ice Shape at the Same Time Instant (Re = 1.8x106) (Reference R8).
R-30
APPENDIX R. ICE SHAPES (CONTINUED) R.5 METHODS FOR DETERMINING (PREDICTING) ICE SHAPES.
R.5.1 Natural Icing Flight Tests.
Applicants may determine critical ice shapes during flight tests in natural icing conditions. You should perform tests for showing that the aircraft can fly safely with natural icing critical ice shapes. This should occur unless means are available to document the natural icing flight tests ice shapes accurately for fabrication and installation of simulated ice shapes. You can then use the aircraft with the simulated ice shapes to assess flying qualities. Issues associated with showing safe flight with natural ice shapes include availability of the icing conditions that produce the critical ice shapes and partial shedding, melting, and sublimation of the ice accretion during flying qualities flight tests.
R.5.2 Use of Icing Tankers to Predict Unprotected Surfaces Ice Shapes.
Applicants may use icing tankers to evaluate critical ice shapes. Icing tankers are useful for complex, 3D surfaces whose ice shapes are difficult to predict with computer codes or other analytical methods. Using icing tankers allows evaluation of ice shapes at full-scale, and without concerns for ice accretion code or icing tunnel limits. The larger drops of tanker icing cloud plumes are considered conservative for applications where a larger and more extensive ice shape is conservative.
You should ensure that the icing tanker, associated instrumentation, and icing plume are calibrated. You should consider the commonly accepted practices, described in SAE ARP5904 (Reference R10), when judging the acceptability of ice shapes from icing tanker flight tests.
However, consistency of ice shapes produced by the different icing tankers for a given test article and similar simulated icing conditions is unknown.
R.5.3 Use of Icing Wind Tunnels to Predict Unprotected Surfaces Ice Shapes.
Applicants may use icing wind tunnels to predict critical ice shapes, especially for complex 3D surfaces. Also, you may use icing wind tunnel ice shapes for simple, 2D surfaces to verify ice shapes from computer codes or other analytical methods.
Test full-scale models of aircraft surfaces when possible. Consider immersion of the model in the calibrated test volume of the tunnel, tunnel blockage effects, and tunnel wall effects. The model’s geometry should conform to type design drawings. When the full-scale model of the aircraft surface is too large for the icing wind tunnel test section, consider hybrid or scaled models.
You may use the hybrid scaling method for airfoil testing. Hybrid models reproduce the fullscale airfoil or the airfoil’s leading edge. The models use a trailing-edge flap to produce the fullscale flow conditions at the model’s leading edge (References R11, R12, and R13). This method allows the use of a truncated model that produces less tunnel blockage and the use of full-scale ambient temperature, airspeed, liquid water content, drop size, and icing times. Limits of the R-31
APPENDIX R. ICE SHAPES (CONTINUED) method include truncation of the model and the need to build several models if a large angle-of attack (AOA) range is of interest.
You may use the basic Ruff method, or equivalent, for design of scaled icing wind tunnel models (References R14 and R15). This method requires matching:
x The modified icing cloud water drop inertia parameter, K0;
x The accumulation parameter, Ac; and x The two energy-balance terms: the freezing fraction, n and, either the water-energy transfer parameter, ij, or the air-energy-transfer parameter, ș, for icing wind tunnels with atmospheric total-pressure test sections.
For pressurized icing wind tunnels, the Ruff method requires you to match all the similarity variables. You can determine the scaled tunnel airspeed by matching either the Weber number (We) or the Reynolds number (Re). Determining the scaled airspeed by holding model-scale and full-scale Weber number (based on leading edge water film thickness) constant and holding the freestream temperature the same have produced good results (References R16 and R17).
The drop sizes involved should not be subject to splashing. You should verify:
x Calibration of the icing tunnel and associated instrumentation;
x Cleaning of spray nozzles using the facility’s procedures;
x Consideration of model-mounting effects; and x Installation of the model within the calibrated test volume of the icing wind tunnel.
(Reference R18 contains information on calibration of icing wind tunnels.) The icing wind tunnel model should conform to the aircraft’s TC design. The IPS pressures, temperatures, and electrical current or voltage should account for tolerances.
You may adjust spray times to reproduce the water catch (or the same scaling accumulation parameter Ac) for 14 CFR part 25, Appendix C cloud lengths. A common practice of adjusting spray time to get the same water catch when the target LWC cannot be achieved because of the icing wind tunnel’s LWC limits is inappropriate. Varying only LWC results in different values of the scaling freezing fraction parameter, n0. Scaling of obtainable test icing conditions to get ice shapes for target icing conditions should follow the recommended Ruff method (Reference R14). Evaluate spray times up to 45 minutes. You may have to adjust spray times for the time required for spray bar stabilization. The facility should document this adjustment.
Figure R-19 correlates ice shapes for a 53-cm reference NACA 0012 airfoil model with those for a 27-cm half-scale model. For this test, the icing condition variables of the half model were scaled by holding the full model K0, Ac, n, I, and Weh constant. The data, produced in the NASA Glenn Research Center Icing Research Tunnel (IRT), are shown for freezing fractions, n, ranging from 1.0 to 0.28. They show the robustness of the scaling for freezing conditions ranging from rime ice to near-glaze ice.
R-32
APPENDIX R. ICE SHAPES (CONTINUED)
(b)
0.05 0.05
• Reference - Scale · 0.00 0.00 - -0.05 -0.05 -0.05 0.00 0.05 0.10 0.1> 0.05 0.00 0.05 0.10 0.15 ·/c x/e n = 1.0 n = 0.77
(a)
0.05
0.00 - -0.05 :
-0.05 0.00 0.05 0.10 0.15 x/e x/c x/c n = 0.52 n = 0.40 n = 0.28 Figure R-19. Correlation of Full-Model (55 cm NACA 0012 Airfoil) and Half-Model Ice Shapes by Holding the Full Model Ko, Ac, n, ø, and Wen Constant (Reference R17).
Comparisons between icing wind tunnel and natural-ice simulated ice shapes are limited.
However, visual comparisons of the detailed characteristics of 3D wind tunnel (NASA IRT) and natural-ice ice accretions on a 30° swept wing model, shown in figure R-20, provide confidence in using IRT ice shapes for the icing conditions evaluated (Reference R19). You should consider conformance with SAE ARP5905 (Reference R18) when judging the acceptability of icing wind tunnel ice shapes. The SAE AC-9C Subcommittee is evaluating the consistency of ice shapes produced by different icing wind tunnels for a given test article.
Hatch Freezer Airfoil stand to extend the airfoil Airfoil base for through the hatch pivoting airfoil Figure R-20(a). View of the NACA 0012 swept wing model inside the NASA Glenn Research Center Icing Research Aircraft (DHC-6 Twin Otter).
R-33
APPENDIX R. ICE SHAPES (CONTINUED) Figure R-20(b). Natural-ice ice accretion (scalloped) obtained with the NASA-GRC Twin Otter ( ȁ=30º, V=144 mph, T=21ºF, LWC=0.45 g/m3, MVD=13 ȝm, t=8 min.
Figure R-20(c). Artificial-ice accretion (scalloped) obtained in the NASA-GRC IRT (ȁ=30º, V=150 mph, T=25ºF, LWC=0.5 g/m3, MVD=20 ȝm, t=10 min.
Figure R-20. Comparison of Natural and Icing Wind Tunnel Ice Shapes (Reference R19).
R-34
APPENDIX R. ICE SHAPES (CONTINUED) R.5.4 Drop Impingement and Ice Accretion Computer Codes and Other Analytical Methods.
Applicants may use verified computer codes and analytical methods to predict ice shapes. The predicted ice shape is the result of calculations that define the flowfield, the drop trajectories, water loading, and the ice accretion physics. Many icing codes are available. They differ to varying degrees in their manner of modeling the ice accretion process. SAE ARP5903 (Reference R20) provides information that describes several available drop impingement and ice accretion codes.
Confidence in 2D icing codes to predict ice shapes accurately is mixed when their results are compared with icing wind tunnel ice. Figures R-21 through R-24 show reasons for this mixed confidence. The figures compare ice shapes predicted by several icing codes with those produced in two icing wind tunnels. The data from Reference R21 show that even at colder, rime ice conditions, where confidence in predicted ice shapes is highest, there are significant differences between predicted and empirical ice shapes.
Figure R-21. Comparison of Drop Impingement and Ice Accretion Code Results With Experimental Ice Accretion Produced in the NASA IRT (V = 135.8 kts, TS = -15.8°C, LWC = 1.16 g/m3, MVD = 50.0 ȝm, icing duration = 517.1 s, GLC305-836-23 airfoil model chord = 0.9144 m) (Reference R21).
R-35