section which in turn led to considerably thicker natural layers of
2.044 0.572 3.272 0.916 81 -18 1 N section which in turn led to considerably thicker natural layers of frost without separate added humidity.
2.250 1.222 3.600 1.955 (82) -18 0 AH The mean frost thicknesses measured before each test are collected in not possible to draw any definite conclusions regarding the effect of Table 3 and Table 4 . It was difficult to achieve a uniform frost layer relative humidity or outdoor air temperature from these data. An as seen from high standard deviation values, which is explained by important outcome of this investigation is that more precise control of the wing ribs and other under structure [Fig. 13]. The frost densities the environmental conditional is needed in order to make a or liquid water contents of the frost were not measured in this study.
quantitative study on the factors influencing the growth of Cold The frost thicknesses on the DLR-F-15 model were measured Soaked Fuel Frost. These experiments were designed, in part, to approximately one hour after the model was brought into the wind- evaluate aerodynamic impacts of real frost which led to some tunnel test section. The frost thicknesses on the HL-CRM Mod compromises in the ability to control the environmental conditions.
model were measured after a longer period of time of several hours.
As shown in Table 4, there is a large range of thickness values. Since In spite of the challenges associated with precisely controlling all of many environmental factors can contribute to the frost growth it is the environmental conditions, the frost formations were considered to be representative of natural CSFF. The frost characteristics should be Table 3. Frost thicknesses generated on the coolant tank upper surface before considered to be “full scale” despite the frost being grown on a the tests for DLR F15 wing section. k is the mean frost thickness in mm.
SD_k is the standard deviation of k. k/c is the frost thickness as a fraction of subscale wing. The environmental conditions were not scaled in any wing chord. SD_k/c is the standard deviation of k/c. RH is the Relative way thus leading to frost thicknesses and morphology that should be Humidity in the test section, TT is the coolant tank temperature just before the consistent with frost growth on large-scale surfaces at similar wind tunnel run, OAT is the wind tunnel air temperature. N indicates natural environmental conditions. Therefore, the normalized frost type of frost and AH added humidity type of frost.
thicknesses reported in Tables 3 and 4 were expected to be much larger than typical values for full-scale wings. There are no data RH TT OAT known to the authors in the technical literature regarding frost k [mm] SD_k [mm] 1000*k/c 1000*(SD_k/c) Type thicknesses for CSFF. There are, however, some data reported for [%] [˚C] [˚C] hoar frost that typically forms overnight in cold conditions. Bragg et al. [5] provide a reasonable summary of work up to and including the 0.093 0.022 0.142 0.035 78 -14 -5 N early 1990’s. They suggested a k/c value equal to 0.00011 for 0.525 0.375 0.808 0.577 -13 -5 AH “typical hoar frost” increasing to 0.00040 for “large frost.” Using the 650 mm chord length for the DLR-F15 model, the equivalent range 0.375 0.225 0.577 0.346 -11 -5 AH of dimensional frost thickness is 0.07 to 0.26 mm. Bragg et al. [Ref.
5] do point out, however, that as frost grows it tends to build itself in 0.133 0.093 0.205 0.143 76 -15 -1 N layers which leads to an important difference between the total thickness and the roughness level. This is an important aerodynamic 0.650 0.187 1.000 0.288 -13 -1 AH distinction because the overall thickness slightly modifies the airfoil or wing contour while the roughness level affects the aerodynamic 0.550 0.153 0.846 0.236 -10 -1 AH penalties. Thus, it is important to quantify both the frost thickness and roughness level. That was the original intent of the 3D scanning 0.154 0.151 0.237 0.233 86 -15 4 N and photogrammetry measurements. However, these methods were not fully implemented because of technical and programmatic 0.475 0.263 0.731 0.405 -13 4 AH challenges.
Due to structural differences in the two models, the frost layer Table 4. Frost thicknesses generated on the coolant tank upper surface before appeared to concentrate on the DLR-F15 model strictly to the tank the tests for HL-CRM Mod wing section. For symbols and abbreviations see area (Fig. 13) whereas on the HL-CRM Mod model the whole main Table 3. *Wing configuration flaps and slats retracted.
element of the model was covered by frost (Fig. 14). Only the composite slat and flap were clear of frost. On both models the lower RH TT OAT side of the wing section was covered with even thicker frost layer.
k SD_k 1000*k/c 1000*(SD_k/c) Type [mm] [mm] However, the lower side skin contamination does not affect the lift [%] [˚C] [˚C] degradation at relevant angles of attack based on results by Bragg et al. [5].
0.253 0.052 0.405 0.084 (72) -18 7 AH 0.342 0.063 0.544 0.101 87 -18 2 N 0.500 0.175 0.800 0.280 74 -18 -3 N 1.200 0.275 1,920 0.440 78 -18 3 N * 1.206 0.410 1.929 0.656 65 -18 8 N 1.389 0.176 2.222 0.282 65 -18 7 N Fig. 13. Frost on DLR-F15 wing section model 1.500 0,176 2.400 0.282 78 -18 3 N Page 8 of 12 10/19/2016 Wing Ribs Fig. 14. Frost on HL-CRM-Mod wing section model.
Lift Degradation after Rotation and Visual Observations Figure 15. Lift coefficient degradation variation with time after rotation for temperatures at or below freezing point.
The traditional method to assess the effect of frost on the wing section lift loss has been to simulate the frost by an equivalent “sand paper” type of fixed roughness. However, real frost is not a fixed roughness but reacts to the surrounding air stream. Depending on the air temperature there may be sublimation, melting and even self- shedding of frost particles. Part of these processes are present already during the take-off run on the ground but after the rotation they are clearly amplified. To figure out these processes and to assess the transiency of the effect of frost the lift losses were recorded not only at the point of lift off but after the rotation too. This gives a better possibility to compare the real frost and de/anti-icing treatment effects on lift loss after the lift off. Naturally the most relevant lift loss value is the one at the lift off.
Some samples of the lift degradation variations in time are collected into Figure 15 and Figure 16. Lift degradation results are presented as a percentage (%) of the uncontaminated wing lift coefficient while the time after rotation is measured from the point where the rotation Figure 16. Lift coefficient degradation variation with time after rotation for has stopped (see Table 2 ).
four different frost (CSFF) thicknesses at a wind tunnel air temperature above freezing point.
Figure 15 presents the lift degradation variations in time for both wing section models at temperatures close to or below freezing point.
The slopes of these curves are quite gentle and almost linear, however clearly negative. At these temperatures there is practically no melting or sublimation though. This can be seen also from the video snapshots immediately after and 60 s after rotation – see Figure 17. There is only very slight darkening on the trailing edge area which implies some decrease in frost thickness. However, no signs of melting are present. The alteration in lift degradation and the change in the frost color shade may be due to shedding of the frost layer in the airstream.
Figure 17. Video snapshot from the upper side of HL-CRM Mod wing section model just after rotation (left) and 60 s after rotation (right). Test Figure 16 presents lift degradation variations in time for temperatures section air temperature is -0.6 ºC and initial frost thickness k/c = 3.27 10 .
clearly above the freezing point. The negative slopes are steeper and There is slight darkening observable at the trailing edge area.
have a concave shape. The three lowest curves are for the first 30 s close each other for both wing section models. For DLR-F15 there The frost melting and sublimation process at an ambient temperature are no measurements beyond 30 s. In HL- CRM Mod tests it was of 7.2 ºC is illustrated in video snapshot collection of Figure 18. It is revealed that the frost began to melt first behind the slat upper difficult to distinguish melting from sublimation when the frost layer opening. However, the melted water froze back to ice at the point just turns darker in the video, but as the water starts to run on the where the front spar is located. Obviously, the front spar was a cold surface it is obvious that melting is the dominant process in phase spot due to its mass. The runback ice ridge was chord wise transition. When melting becomes dominant to sublimation the frost approximately 5 mm wide ranging over the whole span. The ice ridge removal accelerates due to running water that sweeps over the frost.
remained on the surface about 50 to 60 s after which it shed from the surface along the span within a couple of seconds. Most probably it is The melting is visible in the third photograph from left in Figure 18.
this event that is observable in HL-CRM MOD – curves in Figure 16.
The edge of the frost layer bends from a straight line into a round There is a clear abrupt drop in lift degradation values at a point of shape as the melted water begins to run on the surface. Note that at time that matches well with the shed ice ridge event observed in the this high temperature at least the sublimation begins already quite videotape.
Page 9 of 12 10/19/2016
section air temperature is 7.2 ˚C and initial frost thickness k/c = 0.41 10 .
AoA=0º / U=38 m/s AoA=0 º / U=57 m/s AoA=9.2º / U=60 m/s 30 s after rotation Figure 18. Melting and sublimation process during a simulated take-off. Test -3 section air temperature is 7.2 ˚C and initial frost thickness k/c = 0.41 10 .
AoA is angle of attack and U wind tunnel speed. The third photograph from left is taken immediately after rotation. In this photograph the gradually increasing melting is revealed by the round shape of the downstream Figure 19. Frost induced initial lift degradation results of the present study borderlines of the frost clear areas in the middle of the tank area.
compared with the results by Brumby [28]. The dotted line is created from results with localized spanwise roughness distribution at distance of 15 % early in the acceleration (take-off roll) phase. The melting process chord from the leading edge. The continuous line describes the results for a single roughness element at 5 % chord distance from leading edge. The fuel will however clearly accelerate after the rotation. When 30 s has tank area extends from 21.5 % to 61.5 % of chord for DLR-F15 and from 12 elapsed from the rotation there are a few stripes of frost left located at % to 65% of chord for HL-CRM MOD wing section.
the wing ribs area. The front spar ice-ridge separation is observable in a moving video, however it is difficult to catch it in still snapshot photographs.
Frost as a Surface Roughness As stated earlier there are no previous studies considering the CSFF which is located in the fuel tank area only. There are however lots of studies on the effects of ice and frost type of surface roughness on wing aerodynamic properties when the roughness is either distributed over the whole wing area or alternatively concentrated to a short area on the leading edge (ice accretion) or slightly behind the leading edge area (runback type of ice). Lynch and Khodadoust [27] have provided Figure 20. Comparison between the lift degradation caused by Cold Soaked an exhaustive review of the effect of ice and frost accretion on Fuel Frost and anti-icing treatment at OAT +4°C for DLR-F15 wing section.
aircraft aerodynamics. In their review it becomes evident that the effect of roughness on lift degradation is more dominated by the is not considered as an alternative for Cold Soaked Fuel Frost it is position of the roughness than the total area of it. The well-known operationally always considered as a safe option while the CSFF is plot of Brumby [28] illustrates this behavior.
according to present aviation safety practices normally to be removed.
To assess the lift degradation due to the frost generated roughness the present study data have been collected in Figure 19 along with two of In Figure 21 the comparison of CSFF and the two-step de-icing the plot lines of Brumby. The lift degradation values of this study are process is illustrated. These two are operationally realistic not directly comparable to the Brumby lines as the data from them alternatives as the de-iced wing section was initially frost covered.
are mostly collected from maximum lift coefficient loss.
Frost was de-iced by 50°C diluted (30%) Type I followed by Type IV Furthermore, the Brumby lines are collected from tests with a anti-icing treatment. The other curve presents an untreated CSFF localized spanwise disturbance while the roughness in present study frost. Figure 21 suggests that within certain conditions (OAT above is distributed. However as emphasized above the distance of freezing point and frost layer relatively thin) a two – step de-icing roughness from the leading edge is crucial.
may result in worse lift degradation than an untreated CSFF on the wing.
There is some uncertainty in the frost thickness data of Figure 19 as at least in part of the wind tunnel runs frost thickness has already
Summary/Conclusions
decreased somewhat during the acceleration phase before the rotation either by wear or by melting and sublimating. This applies especially at temperatures well above freezing point.
The present study has compared the effects of anti-icing treatment and CSFF on lift degradation of two different 2D wing section models. Considering anti-icing treatment, the main findings were: Comparison of Fluids and Frost Results • Diluted Type IV fluid (75% TIV) gave practically equal lift degradation with the neat Type IV during the first 10 s after An anti-icing treatment is seldom an alternative for Cold Soaked Fuel rotation which after the diluted fluid caused the lift Frost on the wing upper side, as this side is usually merely de-iced degradation to stay clearly higher than the neat fluid. The without any anti-icing treatment. However, comparing these two is residual fluid layer was also found to be considerably not merely of academic interest. Especially a diluted Type IV fluid thicker in the diluted fluid test.
may be used to prevent the frost effects. Figure 20 illustrates the • Acceleration time has a straightforward time shifting effect similar effects of certain amount of frost compared to anti-icing on the lift degradation. The lift degradation is delayed as treatment for DLR-F15 wing section. Even if anti-icing treatment Page 10 of 12 10/19/2016 fluids being introduced into the environment and with the associated costs.
References
[1] Terms of Reference for rulemaking task RMT.0118, Analysis of on-ground wing contamination effect on take-off performance degradation, European Aviation SAfety Agency, Issue 1, 21 Mar 2017, 6 p.
[2] Koivisto P., A Literature Review on Cold Soaked Fuel Frost Formation and Sublimation, Finnish Transport Safety Agency, Trafi Research Reports 6-2015, Helsinki, Finland, 2015.
Figure 21. Comparison between the lift degradation caused by Cold Soaked [3] Ljungström B.L.G., Windtunnel investigation of Fuel Frost and two step de- icing treatment at OAT 4.5 ˚C (Frost) and 7.2 ˚C simulated hoar frost on a two-dimensional wing section (30 %, 50 ˚C Type I + Type IV fluids) for HL-CRM Mod wing section.
with an without high lift devices, Report FFA-AU-902, Stockholm, Sweden, April 1972.
many seconds as the time is shortened to reach wind tunnel [4] Kind R.J., Lawrysyn M.A., Aerodynamic Characteristics speed of 60 m/s. This finding is consistent with results of of Hoar Frost Roughness, AIAA Journal, Vol. 30, No. 7, Ref. 26.
July 1992..
Considering the Cold Soaked Fuel Frost tests the frost effect on lift [5] Bragg M.B., Heinrich D.C., Valarezo W.O, McGhee R.J., degradation were demonstrated to be at least somewhat transient in Effect of Underwing Frost on a Transport Aircraft Airfoil nature even at below freezing temperatures. The effect of frost at Flight Reynolds Number, Journal of Aircraft Vol. 31, reduced with time and temperature after rotation. At air temperatures No. 6, 1994, pp. 1372, 1379..
above freezing point part of the frost effect vanished already during [6] Khodadoust A., Dominik C., Shin J., Miller D., Effect of take-off roll (acceleration) phase and the rest of the frost within 60 s In-flight Ice Accretion on the Performance of a Multi- after the rotation. The recovery from the Cold Soaked Fuel Frost induced lift degradation at temperatures above freezing point is element airfoil, AHS/SAE international Icing caused by frost sublimation and melting. At subfreezing point there is Symposium, 1995, 10 p.
probably also present some shedding of frost particles that reduces [7] Valarezo W.O., Lynch F.T., McGhee R.J., Aerodynamic the frost effect on lift degradation. The lift recovery improves clearly Performance Effects due to Small Leading-Edge Ice with increasing air temperature.
(Roughness) on Wings and Tails, Journal of Aircraft, Vol. 30, No. 6, Nov.-Dec. 1993, PP. 807-812.
An initial set of full-scale frost thickness measurements were reported for the two different 2D wing sections that is representative [8] Boer J.N., van Hengst J., Aerodynamic Degradation due of full-scale Cold Soaked Fuel Frost. While this initial data set may to Distributed Roughness on High Lift Configuration, be representative of frost formations in natural conditions, there were AIAA Paper 93-0028, 1993, 9 p.
limitations regarding the ability to precisely control the [9] Koivisto P., Effects of Cold Soaked Fuel Frost on Lift environmental conditions in these experiments.
Degradation during Simulated Take-off, Finnish This study may be considered as a motivation for further research on Transport Safety Agency, Trafi Research Reports 4- this topic and the following issues should be addressed in the future: 2015, 2015, 24 p..
[10] Koivisto P., Preliminary Cold Soaked Fuel Frost Studies • Frost density or liquid content and frost microstructure with CRM Wing Model, Finnish Transport Safety effect on lift degradation – in this study only the frost layer Agency, Trafi Research Reports 12-2016, Helsinki, thickness was determined Finland, 2016, 11 p.
• Improvements to the control and measurement of the environment variables affecting the frost growth.
• The fuel tank temperature effect on lift degradation. In this [11] Lacy, D.S., Sclafani, A. J.: Development of the High Lift study temperatures between - 10 ˚C to - 18 ˚C were Common Research Model (HL-CRM): A Representative considered and there was no control for the fluid High Lift Configuration for Transonic Transports, 54th temperature once wind tunnel run started AIAA Aerospace Sciences Meeting 4-8 January 2016, • Theoretical background for frost sublimation and melting San Diego, California, USA on a wing [12] Wild, J.: “Experimental investigation of Mach- and When comparing the lift degradation due to CSFF and anti-icing Reynolds-number dependencies of the stall behavior of fluids it is obvious that there may occur well defined ambient 2-element and 3- element high-lift wing sections”, AIAA conditions in normal airliner operations where a strict “clean wing 2012-0108, 50th AIAA Aerospace Sciences Meeting concept” does not really enhance safety but only results in additional Page 11 of 12 10/19/2016 including the New Horizons Forum and Aerospace [26] Koivisto, P. “Anti-icing Fluid Flow off on a Wing Section Exposition 09 - 12 January 2012, Nashville, Tennessee During Simulated Taxi and Take-off”, AIAA 2013-2932, 5th atmospheric and Environments Conference, June 24- [13] Code of Federal Regulations, Title 14 Aeronautics and 27. 2013, San Diego CA.
Space, Volume 3, Chapter I, Subchapter G, Part 121, Federal Aviation Authority, Department of [27] Lynch F.T., Khodadoust A., Effecs of Ice Accretio on Transportation, Operating Requirements, Domestic, Aircraft Aerodynamics, Progress in Aerospace Sciences Flag, and Supplemental Operations, August 25, 2017. 37 (2001) pp. 669-767,.
[14] Commission Regulation (EU) No 965/2012, 5 October [28] Brumby RE. The effect of wing ice contamination on 2012, 148 p.. essential flight characteristics. AGARD CP-496,.
[15] Koivisto P., Soinne E., Kivekäs J., Anti-Icing Fluid Secondary Wave and Its Role in Lift Loss, Journal of Aircraft, published online 29 May2018, 9 p.
Contact Information
[16] Soinne E., Rosnell T., Aerodynamic, Laser scanning and The lead author email address is pekka.koivisto@aalto.fi Photogrammetric Measurements on Cold Soaked Fuel Frost, AIAA Paper 2018-3830, 2018, 12 p.
Acknowledgments
[17] Lee, S., Broeren, A. P., Kreeger, R. E., Potapczuk, M.
G., and Utt, L.,“Implementation and Validation of 3-D The wind tunnel measurements were performed by Arteform Ltd.
Ice Accretion Measurement Methodology,” AIAA Paper Laser scanning and analysis of the results was performed by NASA.
2014-2613, June 2014.
Support from the U.S. Federal Aviation Administration and Finnish Transport Safety Agency is gratefully acknowledged.
[18] Broeren, A.P., Addy, Jr., H.E., Lee, S., Monastero, M.C., McClain, S.T., “Three-Dimensional Ice-Accretion
Definitions/Abbreviations
Measurement Methodolgy for Experimental Aerodynamic Simulation,” Journal of Aircraft, Vol. 55, c wing model chord No. 2, Mar.-Apr. 2018.
wing lift coefficient C L [19] McClain, S.T., Vargas, M., Tsao, J.-C., and Broeren, A.P., “Ice Roughness and Thickness Evolution on a C LClean clean wing lift coefficient Business Jet Airfoil,” AIAA Paper 2018-3014, June k frost roughness 2018.
frost thickness t [20] Certification Specifications and Acceptable Means of Compliance for Large Aeroplanes, CS-25 Amendment U wind tunnel velocity 13, 10 June 2013.
[21] Hill E.G., Zierten T.A., Aerodynamic Effects of Aircraft V rotation speed R Ground Deicing/Anti-Icing Fluids, Journal of Aircraft, Vol. 30, No.1 Jan-Feb 1993, page 24-34.
take-off safety speed V 2 [22] Broeren A.P., Riley J.T., Review of the Aerodynamic angle of attack α Acceptance Test and Application to Anti-Icing Fluids Testing in the NRC Propulsion and Icing Wind Tunnel, Aerodynamic Acceptance Test AAT NASA/TM-2012-216014, 2012, 34 p..
[23] Koivisto P., Effects of Anti-icing Treatment on Lift AGL Above Ground Level Degradation during Simulated Take-off, Finnish Transport Safety Agency, Trafi Publications 25-2013, Cold Soaked Fuel Frost CSFF Helsinki, Finland, 2013, 22 p..
[24] Standard Test Method for Aerodynamic Acceptance of European Aviation Safety Agency EASA SAE AMS 1424 and SAE AMS 1428 Aircraft Deicing/ Anti-icing Fluids”, AS5900 Rev.B , SAE International, OAT Outside Air Temperature 26 July 2007.
One Engine Inoperative OEI [25] Runyan, L.J., Zierten, T.A., Hill, E.G., Addy, H.E., “Lewis Icing Research Tunnel Test of the Aerodynamic Effects of Aircraft Ground Deicing/Anti-icing Fluids”, NASA-TP 3238, 1992.
Page 12 of 12 10/19/2016