NASA Technical Reports Server (NTRS) 19990018601: An Experimental and Numerical Study of Icing Effects on the Performance and Controllability of a Twin Engine Aircraft

20 Pages·0.95 MB·English

by  NASA Technical Reports Server (NTRS)

ICOMP-99-02
NASA / TMn1999-208896
AIAA-99-0374
I_lICOMP I_,,
An Experimental and Numerical Study of Icing
Effects on the Performance and Controllability
of a Twin Engine Aircraft
A. Reehorst
Lewis Research Center, Cleveland, Ohio
J. Chung
Institute for Computational Mechanics in Propulsion, Cleveland, Ohio
M. Potapczuk and Y. Choo
Lewis Research Center, Cleveland, Ohio
W. Wright and T. Langhals
Dynacs Engineering Company, Inc., Brook Park, Ohio
Prepared for the
37th Aerospace Sciences Meeting & Exhibit
sponsored by the American Institute of Aeronautics and Astronautics
Reno, Nevada, January 11-14, 1999
National Aeronautics and
Space Administration
Lewis Research Center
January 1999
Available from
NASA Center for Aerospace Information National Technical Information Service
7121 Standard Drive
5285 Port Royal Road
Hanover, MD 21076 Springfield, VA 22100
Price Code: A03 Price Code: A03
AIAA 99-0374
AN EXPERIMENTAL AND NUMERICAL STUDY
OF ICING EFFECTS ON THE PERFORMANCE AND CONTROLLABILITY
OF A TWIN ENGINE AIRCRAFT
A. Reehorst
Icing Branch
National Aeronautics and Space Administration
Lewis Research Center
Cleveland, Ohio
J. Chung
Institute for Computational Mechanics in Propulsion
NASA Lewis Research Center
Cleveland, Ohio
M. Potapczuk and Y. Choo
Icing Branch
National Aeronautics and Space Administration
Lewis Research Center
Cleveland, Ohio
W. Wright and T. l_zmghals
Dvnacs Engineering Company, Inc.
Brook Park, Ohio
Abstract Using a two-dimensional Navier-Stokes code,
the flow field and resultant lilt and drag were
In September 1997 the National Transportation calculated for the wing section with various ice
Safety Board (NTSB) requested assistance from shapes accreted in the IRT test. Before the final
the NASA Lewis Research Center (LeRC) Icing calculations could be performed extensive
Branch in the investigation of an aircraft accident examinations of geometry smoothing and
that was suspected of being caused by ice turbulence modeling were conducted.
contamination. In response to the request NASA
agreed to perform an experimental and The most significant finding of this effort is that
computational study. several of the five-minute ice accretions
generated in the IRT were found by the Navier-
The main activities that NASA performed were Stokes analysis to produce severe lift and drag
LeRC Icing Research Tunnel (IRT) testing to degradation. The inlormation generated by this
define ice shapes and 2-D Navier-Stokes analysis study suggests a possible scenario for the kind of
to determine the performance degradation that control upset recorded in the accident.
those ice shapes would have caused. Secondary findings were that the ice shapes
accreted in the IRT were mostly limited to the
An IRT test was conducted in January 1998. protected pneumatic boot region of the wing and
Most conditions for the test were based upon raw that during testing, activation of the pneumatic
and derived data from the Flight Data Recorder boots cleared most of the ice.
(FDR) recovered from the accident and upon the
current understanding of the meteorological
conditions near the accident.
"'Copyright © 1999bythe American Institute of Aeronautics andAstronautics, Inc.No copyright isasserted inthe United States under
Title 17,U.S. Code. The U.S.Government has aroyalty-free license toexercise allrights under the copyright claimed herein for
Governmental Purposes. Allother fights are reserved bythe copyright owner."
I
American Institute of Aeronautics and Astronautics
Background The c(,ntaminated airfoil geometries measured
during the IRT test were then analyzed with a
In early 1997, a twin-engine commuter aircraft two-dimensional (2-D) Navier-Stokes code to
(Figure 1)crashed during a rapid descent after an predict their degraded aeroperformance. Due to
uncommanded roll excursion. There were no the time constraints of this project, the icing
survivors. The ground impact forces and post wind tunnel test concentrated on measuring the
accident fire destroyed the aircraft. Instrument ice accretion shapes, with all aerodynamic
meteorological conditions prevailed at the time analysis performed computationally.
of the accident 1. During the investigation, the
National Transportation Safety Board (NTSB) To verify that the 2-D analysis was producing
contacted the Icing Branch at NASA Lewis realistc results (that no significant thrce-
Research Center. The NTSB requested NASA's dimenfional (3-D) flow effects were present), a
participation since ice contamination was limited 3-D Navier-Stokes analysis was
suspected as being a factor. NASA offered to conducted. The 3-D analysis was focused on
pursuc a study that included experimental and assuring that no significant span-wise lqow was
computational elements to identify the broad present that would invalidate the chord-wise
nature of possible ice contamination and the flow assumption required by the 2-D Navier-
likely associated performance degradation. Stokes analysis.
Project Methodology IRT Test
The primary goal of the NASA activity The Icing Research Tunnel (Figure 2) is a
supporting the NTSB investigation was to closed-loop atmospheric tunnel with a test
attempt to identify a mechanism explaining the section that is 6ft high, 9ft wide, and 20 fl long.
stability and control upset that preceded the It ise(iuipped to support testing at airspeeds from
accident. The methodology that was adopted to 50 to z,30 mph in a temperature and water-
satisfy this goal was a multistep process that drople environment that simulates natural icing
included both computational and experimental conditions 2. Temperature is controllable
tasks. betwet;n -20 and +33"F. Clouds can be
produced in a controlled manner with droplet
The first step was to perform initial ice accretion mean volumetric diameter (MVD) between 9.5
predictions with the NASA LEWICE program to and 270 lam.
define the impingement limits and total mass of
ice to be expected. Since the specific For this test an existing model was utilized. The
mctcorological conditions at the time of the model was a modified piece of wing from the
accident are unknown, the LEWlCE calculations subjec aircraft centered on the inboard region of
allowcd the generation of ice shapes over the the aiL'ron. The model had a 6-foot span and
range of possible conditions. was m3unted vertically in the IRT (Figure 3).
The wing had a modified trailing edge and
Thc LEWICE calculations were followed by a pneumatic boot. The trailing edge was faired
test in the NASA Lewis Research Center's Icing behind the aft spar to avoid the mechanical
Research Tunnel (IRT). The IRT testing was compl,;xity of the aileron. The pneumatic boot
considered necessary since LEWICE, as with all was m)dified for its original test to allow the
ice accretion codes, does not model all of the incorp )ration of thermal measurements. This
physics of the accretion process. The complex featurt was not utilized for this subject testing,
movement of liquid water on the airfoil prior to howe'ver, the modification resulted inthe boot
its freezing is currently modeled by simplified extending further aft on the lower surface than a
relationships inall ice accretion codes. Due to production boot for this wing. This extended
this limitation and other assumptions required by boot surface acted as an enhanced ice accretion
ice accretion codes, the ice shapes predicted at Iocaticn that isevident when reviewing the test
temperatures near freezing are not always as ice sh_pe tracings. Except for the cases when the
accurate as those predicted for lower pneumatic boot was activated, the manufacturers
temperatures. recom nended level of vacuum was applied to
2
American Institute of Aeronautics and Astronautics
thebootthroughotuhtetestingtoprevenatuto- uncertainties in the FDR measured values of
inflationT.oensurtehattheairtemperatuinre
angle-of-attack and temperature, additional series
thetesstectiownasaccuraatetthedesirendear-
were added tothe matrix that examined higher
freezincgonditiona,nauxiliartyemperature
and lower angle-of-attack than the baseline value
probewasmountendeatrhemodealndisvisible
and higher and lower temperatures than that
inFigure3.
baseline value. An additional series with longer
accretion time was also added since the ice
accretion time was inferred information and not
Test Environment Conditigns directly measured. A series of supercooled large
droplet (SLD) conditions was added during the
Based upon the findings of the investigation's testing at the request of the NTSB to ensure the
meteorological team 3,the accretion time, the completeness of the database. To add to the
maximum liquid water content (LWC), and the level of confidence in the IRT results, all primary
cloud droplet's median volumetric diameter conditions were repeated at least once.
(MVD) were estimated fbr this test. The
accretion time was estimated at 5minutes. This
was the length of time required by the accident Test Procedure
aircraft to transit an area of low reflectivity
evident on regional weather radar. This time The IRT test was conducted over two weeks in
also agrees with estimates based upon a January 1998. The conditions described in Table
measurable drag increase calculated by the I were used for this test and the most significant
investigation's performance team from flight conditions were repeated at least once, as seen in
data recorder (FDR) information 4. The Table 2. The normal process used during this
maximum LWC was estimated to be 0.8 g/m_ test was to cool the IRT to the desired
based upon microwave radiometer measurements temperature and set the airspeed. When the
of the same airmass following the accident. The temperature and airspeed were stable at the
maximum MVD of the cloud droplets was desired test conditions, the spray was activated
estimated to be no greater than 70 to 80 lain. for the length of time at air and water pressures
This droplet-sizing estimate was considered defined by the test matrix to simulate a natural
reasonable since the aircraft was known to have icing cloud. When the spray was shut off,
been in the clouds, and droplets larger than that several wake survey probe traverses were made
are normally associated with below-cloud behind the model. These wake probe traverses
drizzle 5. were used to calculate an uncorrected tunnel
centerline section drag. Due to project time
The remaining test parameters were defined constraints, further correction, analysis, and
based upon a combination of the accident comparisons of this data have not been
meteorological data and the FDR inlormation. completed. Following the wake probe traverses,
The times that defined the assumed icing period the tunnel fan was stopped so that test personnel
from the meteorological report were used to could enter the test section. Upon entering the
access the pertinent information from the FDR. test section, the test engineers would first
Representative airspeed, aircraft angle-of-attack photograph the model. Both digital and film
and air temperature values were calculated by cameras were utilized for this test, with the film
averaging over the assumed ice accretion period photos typically being used lor close-up detailed
(total air temperature was derived from static air shots and the digital photos tor overview shots
temperature measurements and corrected for (Figures 4 and 5). When the photographs were
altitude and airspeed, and wing angle-of-attack completed, the leading edge ice accretion was
was calculated by adding the wing incident angle sliced with a hot knife, then traced onto a
to the aircraft body angle-of-attack). cardboard template (Figure 6). These tracings
were made at 30, 36, and 42 inches from the
The combination of the assumed cloud wind tunnel floor (the 36 inch location is the
conditions and the measured flight environment vertical center of the test section). The ice
data was used to define the IRT test matrix shapes originally traced onto cardboard are then
(Table I). Four conditions were considered the later digitized and stored electronically. All
baseline series inthe matrix. Due to tracings presented here are digitized data
3
American Institute of Aeronautics and Astronautics
(Figur7e).Threechanneolsfvideowerealso
The lower temperature series (at 28 degrees
recordelodreachiceaccretionT.hevideo
F threnheit total temperature instead of 30
cameracsaptureudppearndlowerleadinegdge
degrees) produced slightly more total ice
viewsalongwithabroadmodevliew.Thevideo
v,._lume than the baselines series and more
signalwsererecordetodSVHSlormattapes.
importantly started to exhibit some ridge
fc,rmation for the 40 and 70 lainMVD cases
(Figure 15and 16).
Resultant Ice Accretions
Tile second lower temperature series (at 26
Repeatability was good for all cases, with the degrees Fahrenheit total temperature instead
volume of ice, impingement limits, and horn or of 30 degrees) exhibited an ice ridge for all
ridge location closely matched for repeat cases. femurconditions (Figures 17to 20). The
The majority of the ice shapes accreted during ri_tges apparent in these accretions are
this test fell into two categories. The first was a different from a standard horn formation in
thin ice shape with arough surface made up of that they are surrounded by only a thin layer
small bumps of varying size and sharpness ot ice, where a standard horn is part of a
(Figure 8). The second category of ice shape had more substantial ice accretion. The ridge
the same general volume of ice with similar w asmost prominent for the two cases with
roughness elements as the first category, but also LWC=0.8 g/m _(matrix numbers 26 and 27,
a prominent ridge (Figure 9). This ridge was Figure 17and 18). For the two cases with
approximately 0.5 inches high. c_mstant LWC, the ridge moved aft with
increasing MVD. For the two cases with
The following is a description of the trends cc_nstant MVD, the ridge moved all with
apparent in the test data: increasing LWC.
The baseline series (see Table I) resulted in In the longer time series (with an accretion
thin ice accretions with distributed time of 10minutes instead of 5), self-
roughness and no ridge Ibrmation lbr all shedding occurred in matrix number 32
four of the series conditions (Figures 10 to (Figure 21). For the remaining cases in this
13). series the ice mass was greater than in the
baseline series, as would be expected.
Compared to the baseline series, the higher These shapes also exhibited aridge
angle-of-attack series (at 7degrees instead lo'mation (Figures 22 to 24). However
of 5degrees) produced ice shapes that were th:;sc ridge formations appeared to be much
shifted towards the lower surface, as would m_)re like a standard horn lbrmation. These
be expected with this change of angle-of- ice shapes were in general larger and
attack (Figure 14). Since the upper rougher than those seen in the baseline
impingement limit moved forward, the series.
overall ice thickness of these shapes was
somewhat greater than that seen in the Tt_e additional supercooled large droplet
baseline series.
se ies (which extended the MVD up to 270
I.ttifrom the baseline maximum of 70)
The lower angle-of-attack series cases (at 3 sh _wed no significant difference in the
degrees instead of 5degrees) were not leading edge ice accretion over those
significantly different than the baseline exaibited in the baseline series, except that
series. the 175and 270 lam cases self-shed.
The higher temperature series cases (31 Tte pneumatic boot was operated for several
degrees Fahrenheit total temperature instead ca ;es in the de-ice series. It was operated for
of 30 degrees_ were not significantly th, matrix number 3, 26, and 27 conditions.
different than the baseline series, but may Depending on the case, the pneumatic boot
have had a very slight decrease in the total w_s either activated at spray-off only
volume of ice accreted. (Figure 25), at two minutes into a five-
minute spray (Figures 26 and 27), or at
4
American Institute of Aeronautics an, IAstronautics
twominuteasndatspray-o(fFf igure2s8 exhibited a thin ice accretion with many small
and29)•Inallcasewshenthepneumatic
roughness elements. Matrix number 3was also
boowt asoperateadttheendoftheicing selected since itexhibited larger, sharper
sprayth, eleadinegdgeiceshedverywell. roughness elements than did matrix number 2.
Forthetwocasewshentheboowt as
Matrix number 9was selected because it
activateadttwominuteisntoafive-minute
exhibited the roughness of matrix number 2, but
sprayin, onecaseitpreventethde also had a larger accretion on the upper surface.
developmoefnatridgeformatioann,dinthe
Finally, the resultant ice shape lorm matrix
othecraseitdidnot.Whethethreridgewas number 26 was selected since itexhibited a
suppressoerdnotwasprobabdlyependant significant ridge formation.
onhowwellthebooctlearetdheiceatthe
two-minuatectivationIt.shoulbdenoted
For the 2-D calculations, a significant effort was
thatthepneumabticootusedduringthistest undertaken to determine the best level of ice
wasnotaproductiobnooat ndithadnever
tracing point smoothing and optimal grid point
experienctehdeoperationeanlvironment. resolution 7s. The calculated lift curves of the
Howevenro,speciasltepwseretakento
four ice shapes examined compared to that of the
enhanctheeboopt erformanscuec,hasthe
clean airlbil are presented in Figure 30. As
applicatioonfIce-Xa, nditwasoperateadt
expected, when ice was added to the airfoil, the
thesamepressuraessaflightsystemT•he
calculated lilt was degraded. The matrix 2shape
boopt erformanince
the deicing tests is had the least degradation since it had the lowest
therefore considered to be broadly level of roughness of the shapes examined,
representative of that of a flight system. followed by the matrix 3shape, which had
larger, rougher roughness elements. When the
The physics involved in the development of the ice thickness increased, as in matrix number 9,
ice ridges observed during this test are not well the lift degradation was further increased. The
understood. It isvery possible that the ridges are surprising result was the level of degradation
a result of the damming of surface runback water when aridged ice shape was examined. The
by ice leathers. The ridges appear to have major difference between the ice shape from
formed at the aft edge of the primary ice matrix number 2 and that from matrix number 26
accretion where ice leathers begin. Ice damming was the ice ridge. The roughness levels observed
by ice feathers has been observed by close-up in the two cases were very close. When the ice
videography% and may explain this form of ice ridge was present, a45 percent reduction of
growth. More research is required into the maximum lift coefficient (C_.... )was predicted
lbrmation of the ice ridges seen in this test and, in comparison to the clean airfoil. Since the
more generally, into the movement of liquid ridged ice shape caused the greatest lift
water on developing ice accretions. degradation, it was selected for all further
computational analysis.
Computational Analysis Of equal significance compared to the lift
degradation caused by the ice contamination is
The computational analysis performed as part of the effect of aileron deflection for contaminated
this eflbrt isdescribed in two AIAA reports airfoils. Navier-Stokes calculations were
TM.
The computational analysis consisted of both 2- perlbrmed for the wing in the configuration at
D and 3-D NavJe•r-Stokes 9computations. As the moment of the control upset. The moment of
described above, the performance of the ice- control upset was defined for this activity as the
contaminated wing was calculated via point when the aircraft's roll no longer tracked
computational techniques instead of by the aileron input. The roll angle and aileron
experiment inorder to complete the analysis in a (control wheel) position were recorded on the
timely manner Ior the NTSB investigation. FDR (Figure 31.). The configuration of the
aircraft with regards to bank and aileron position
Four cases of the IRT ice shapes were selected is shown in Figure 32. It should be noted that
lbr computational analysis based upon their the aircraft flight variables of interest here
varying nature• The ice shape from test matrix (airspeed, angle-of-attack, altitude, and
condition number 2 was selected since it temperature) are different from those used to
5
American Institute of Aeronautics and Astronautics
simulattehepossiblieceshapeinstheIRT,since Illinoi: at Urbana/Champaign _'_.While the
theaccretioonficeandthecontroulpset
airfoil used for the experiment was different than
occurreadtdifferenttimesandspatiaplosition.
that used for this computational activity, the ice
ridged shape was similar. The experimental stall
Whentheailerondeflectioenffecwt asaddedto
angle is greater than that predicted by the NASA
thedegradaticoanusebdytheridgediceshape 2-D effort, but occurs due to trailing edge flow
theliftcurveisnFigure33resultedT.heclean
separation. The nature of the flow results appear
leftwing(CleaLnW)linerepresenthtselift to be similar to the NASA results, but as stated
curvefortheairfoiwl iththeailerodneflected
before more research will be required before the
downat2.56degreesT.hecleanrightwing compt:tational results can be utilized
(CleanRW)linerepresenthtseliftcurvelbrthe independently with complete confidence.
aidbilwiththeailerondeflecteudpat2.74
degreeTs.heleftwinggeneratmesoreliftthan
theright:therelbrtheeaircrafwtillrollright. Summa_ of Significant Experimental and
Thiswastheintentioonfthepilotwhenthese Computational Findings
controilnputwseremadeH. owevewr,iththe
ridgediceshapceontaminatioonnthewings, The ice shapes accreted during the IRT test
boththeIcedLWandIcedRWlinesshowhow
largel_ fell into two categories. The first was a
theleftandrightairfoilliftcurvewsere thin iczshape with arough surface made up of
degradeAd.sangle-of-attaincckreasethse small romps of varying size and sharpness. The
differentialifltdecreasaensdatanangle-of-
second category of ice shape had the same
attacakroundIIdegreethsedifferencinelift
gener_:l volume of ice with similar roughness
betweethnetwowingsgoetsozero.This
eleme_lts as the first category, but also a
degraderodllcontroaluthoritiysapossible
promi_lent ridge. This ridge was approximately
answearstowhytheaircrafctontinuetodrollto
0.5 in_hes high.
theleftwhentheaileroninputwsere
commandianrgightroll. In all ,:ases when the pneumatic boot was
operal_d at the end of the icing spray, the leading
Atthecompletioonfthe2-Danalysaislimited edge i:e shed very well.
3-Danalyswisasundertak8e.nTheleftwing
geometriyn,cludinagilerondeflectiobnutminus Of the four ice shapes analyzed with the Navier-
enginneacellew,asapplietdoa3-Dstructured Stokes code, the degraded C_.maxfrom the ice
grid.Developinaggridthataccurateclyaptured shape with the lowest level of roughness, matrix
thcwinggeometaryndalsoallowetdheNavier- number 2, was about 75 percent of the clean
Stokecsodetoconvergtoeokagreadteaolf airfoil The C_.m_xfrom the ice shape with larger,
effort.Toallowtheconvergence
of the Navier- sharpt r roughness elements, matrix 3, was about
Stokes code in a timely manner the grid 68 pel cent of the clean airloil. The C_.... from
resolution used was relaxed. The resultant 3-D the ro lgh shape with more ice accretion on the
flowfield demonstrates a chord-wise flow over upper surface of matrix number 9 was about 60
the wing. The trailing edge stall mechanism is percent of the clean airfoil. And the C_.... from
also similar to that seen in the 2-D analysis, the ice ridge case, matrix 26, was about 53
however, the stall occurred much more abruptly percent of the clean airfoil.
in 3-D than it did in 2-D. In 3-D the separation
started right belbre the stall, but lor 2-D the With Theaileron effects added to the lift
separation started several degrees before the degra, lation caused by the ridged ice shape, the
stall. The angle-of-attack for stall in the 3-D result:mr lift curves for the left and right wing
analysis is also higher than that for the 2-D airtbil sdisplay that little or no lift difference
analysis. These differences may be caused by the may I-ave been present and therefore roll control
relaxed grid resolution utilized for the 3-D autho 'ity may have been significantly degraded.
calculations. While explainable, the difference
between the 2-D and 3-D results does warrant While 3-D computational and experimental
further investigation. A preliminary result;do not directly confirm the 2-D Navier-
experimental comparison has also been made Stoke ;results, they do confirm the general flow
with results generated by the University Of chara,_teristics, and stall mechanism. And
6
American Institute of Aeronautics a _dAstronautics
enougdhifferenceexsisbtetweethne2-Dand3- guidelines that suggested waiting until
Dgridsandthecomputationaanldexperimental significant accretions existed belore activating
geometriteosaccounfotrthedifferenciens
the pneumatic ice protection systems may have
results. been a fatal error. Due to the limited information
recorded, we will never be completely sure of the
cause of this accident and can surely never know
Concluding Remarks the level of awareness and intentions of the flight
crew. However, based upon a review of the
There are many lessons to be learned by this NASA results from this study along with the
activity. The NASA Icing Branch has gained a group consensus at a recent workshop __,two
significant amount of experience in measuring operational suggestions become apparent:
wind tunnel generated ice shapes and
determining the aerodynamic perlbrmance for a • Even ice accretions that may appear small
contaminated wing. Along with this added and benign can be truly hazardous. Treat all
knowledge have come more questions. What are icing conditions and resultant ice accretions
the physical processes that led to the generation as threats to your aircraft.
of the ridged ice shapes'? How do varying ridge • Unless otherwise instructed by the aircraft
locations and heights influence airfoil and wing flight manual, activate the ice protection
performance? How well do the Navier-Stokes systems at first indication of ice accretion:
codes compare to experimental aeroperformance do not wait until a significant accretion has
data, and how can the agreement be improved'? formed.
And, what is the best way of using the 2-D and
3-D codes together? There isa significant
amount of work that needs to be accomplished References
before these question are answered and before
we have a good understanding and confidence in I.) National Transportation Safety Board
all the methodologies. Aviation Accident Report, In-Flight Icing
Encounter and Uncontrolled Collision with
The authors therefore recommend that the Terrain, Comair Flight 3272, Embraer
following activities be pursued: EMB-120T, N265CA, Monroe, Michigan,
January 9, 1997, NTSB Accident #
• Additional testing be conducted to gain an DCA97MA017.
understanding of the ice ridge formation
process. 2.) Soeder, Ronald H., Sheldon. David W.,
• More analysis and testing be conducted to Andracchio, Charles R., Ide, Robert F.,
understand the criticality of the ice ridge in Spera, David A., and Lalli, Nick M.,"NASA
regards to airfoil performance. This needs Lewis Icing Research Tunnel User Manual",
to include both ridge location and height. NASA TM-107159, June, 1996.
• Compare the computational results from 2-D
and 3-D Navier-Stokes code utilized to other 3.) The Meteorological Group Chairman's
Navier-Stokes codes using the same Factual Report, NTSB Accident #
geometry and conditions. DCA97MA017.
• Aeroperlormance testing and comparison to
the Navier-Stokes solutions. 4.) The Group Chairman's Aircraft Performance
Study, NTSB Accident # DCA97MA017.
The other lessons learned through this activity
relate to the operation of aircraft in icing 5.) Personal communication with Marcia
conditions. Based upon the cockpit voice Politovitch and Ben Bernstein of the
recorder from the accident aircraft, it appears National Center for Atmospheric Research
that the flight crew was unaware of the potential (NCAR), December 1997.
hazard of their flight situation. While the total
ice accretion was probably quite small, the 6.) Hansman, R.J., et al.,'Close-up Analysis of
aerodynamic effects appear to have been severe. Aircraft Ice Accretion", AIAA-93-0029,
For this incident, relying on past experiences and NASA TM- 105952, January 1993.
7
American Institute of Aeronautics and Astronautics
7.) Chung, J., Reehorst, A., Choo, Y., and 10.) Bragg, M.B., Loth, E., Dunn, T., Lee, S.,
Potapczuk, M.,"Effect of Airfoil Ice Shape and Kumar, S.,'EMBI20 airfoil with Ice
Smoothing on the Aerodynamic E_periment and Computation (preliminary
Performances", AIAA 98-3242, July 1998. results)", FAA/NASA briefing, October
1998.
8.) Chung, J., Choo, Y., Reehorst, A., Potapczuk,
M., and Slater, J.,"Navier-Stokes Analysis 113 Joint NASA/FAA Airplane Deicing Boot
of the Flowfield Characteristics of an Ice Ice Bridging Workshop held at NASA
Contaminated Aircraft Wing", AIAA 99- Lewis Research Center, November 1997.
0375, January 1999.
9.) "NPARC Version 3.0 User's Manual",
NPARC Alliance, September 1996.
8
American Institute of Aeronautics a Id Astronautics