NASA/TM-2002-211818

Flutter Analysis of a Transonic Fan

R. Srivastava, M.A. Bakhle, and T.G. Keith, Jr.

University of Toledo, Toledo, Ohio

G.L. Stefko
Glenn Research Center, Cleveland, Ohio

September 2002
 The NASA STI Program Office .. . in Profile

Since its founding, NASA has been dedicated to • CONFERENCE PUBLICATION. Collected
the advancement of aeronautics and space papers from scientific and technical
science. The NASA Scientific and Technical conferences, symposia, seminars, or other
Information (STI) Program Office plays a key part meetings sponsored or cosponsored b y
in helping NASA maintain this important role. NASA.

The NASA STI Program Office is operated b y • SPECIAL PUBLICATION. Scientific,

Langley Research Center, the Lead Center for technical, or historical information from
NASA's scientific and technical information. The NASA programs, projects, and missions,
NASA STI Program Office provides access to the often concerned with subjects having
NASA STI Database, the largest collection of substantial public interest.
aeronautical and space science STI in the world.
The Program Office is also NASA's institutional • TECHNICAL TRANSLATION. English-
mechanism for disseminating the results of its language translations of foreign scientific
research and development activities. These results and technical material pertinent to NASA's
are published by NASA in the NASA STr Report mission.
Series, which includes the following report types:
Specialized services that complement the STI
• TECHNICAL PUBLICATION. Reports of Program Office's diverse offerings include
completed research or a major significant creating custom thesauri, building customized
phase of research that present the results of databases, organizing and publishing research
NASA programs and include extensive data results ... even providing videos.
or theoretical analysis. Includes compilations
of significant scientific and technical data and For more information about the NASA STI
information deemed to be of continuing Program Office, see the following:
reference value. NASA's counterpart of peer-
reviewed formal professional papers but • Access the NASA STI Program Home Page
has less stringent limitations on manuscript at http://www.sti.nasa.gov
length and extent of graphic presentations.
• E-mail your question via the Internet to
• TECHNICAL MEMORANDUM. Scientific hel [email protected]
and technical findings that are preliminary or
of specialized interest, e.g., quick release • Fax your question to the NASA Access
reports, working papers, and bibliographies Help Desk at 301-621-0134
that contain minimal annotation. Does not
contain extensive analysis. • Telephone the NASA Access Help Desk at
301-621-0390
• CONTRACTOR REPORT. Scientific and
technical findings by NASA-sponsored • Write to:
contractors and grantees. NASA Access Help Desk
NASA Center for AeroSpace Information
7121 Standard Drive
Hanover, MD 21076
r-.- - -- -
I

NASA/TM-2002-211818

Flutter Analysis of a Transonic Fan

R. Srivastava, M.A. Bakhle, and T.G. Keith, Jr.

University of Toledo, Toledo, Ohio

C .L. Stefko
GleIm Research Center, Cleveland, Ohio

National Aeronautics and

Space Administration

•
Glenn Research Center

September 2002
 ~ - -~I

Acknowledgments

This work was performed under funding from the Quiet Aircraft Technology Project of NASA
Glenn Research Center. Joseph E. Grady is the project manager. The authors also wish to gratefully
acknowledge the assistance and suggestions provided by Joseph Panovsky of Honeywell Engines,
Systems, & Services, Brian Fite of NASA Glenn and John A. Gazzaniga of QSS Group, Inc.

Available from
NASA Center for Aerospace Information National Technical Information Service
7121 Standard Drive 5285 Port Royal Road
Hanover, MD 21076 Springfield, VA 22100

Available electronically at http: (lgltrs.grc.nasa.gov

 FLUTTER ANALYSIS OF A TRANSONIC FAN

R. Srivastava,* M.A. Bakhle,* and T.G. Keith , Jr.

University of Toledo
Toledo, Ohio 43606

G.L. Stefko
National Aeronautics and Space Adm inistration
Glenn Research Center
Cleveland , Ohio 44135

ABSTRACT turbomachine confi gurations have al 0 been re ported [11 - 13].

This paper describes the calculation of flutter stability Williams, Cho, and Dalton [11] u ed a linear panel method to
characte ristics for a transonic fo rwa rd swept fa n co nfiguration so lve the eigenvalue probl em . Gerol ymos [12] and Sri vastava
using a viscous aeroe lastic analysis program. Unsteady avier- and Redd y [1 3] sol ved the coupl ed ae roelastic equati ons based
Stoke equations are so lved on a dyna mica ll y deforming, body on an inviscid aerodynamic analys is. The majo ri ty of the
fitted , grid to obtain the aeroelastic characteristics using the methods are either analytical, applicable to two-dimensional
energy exchange method. The non-zero inter-bl ade phase geometry and/or ignore the viscous effects.
angle i modeled using phase-lagged boundary conditions. The modem fans being designed to reduce noise are swept
Resul ts obtained show good correlatio n with measurements. It forward as well as leaned in the plane of rotation. These
is fo und that the location of shock and vari ation of shock fea tures coupl ed with transonic fl ow resul t in aeroelastic
strength strongly infl uenced ta bility. Also, outboard stations problems. For turbomachinery blade rows, flu tter i observed
primarily contributed to tability charac teri sti cs. Results primaril y in single natural mode. Single mode fl utter can be
demonstrate that changes in blade shape impact the calcul ated captured using the energy exchange method, w hich calcul ates
aerodyna mic dam ping, indicatin g importance of usin g acc urate the flo w of energy between the vibrating blade and the
bl ade operati ng shape under centrifuga l and steady surrounding fluid . A three-dimensional vi cous aeroelastic
aerodynamic load ing fo r fl utter prediction. It was found that analys is code TU RBO-AE, ba ed on energy exchange method
the calculated aerodynamic da mp ing was relatively insen itive has been developed and reported in [14] . The objective of the
to variation in natural frequency . present stud y is to investi gate the tability of a forward swept
transonic fan geo metry. The transonic fan analyzed , showed
INTRODUCTION flutter at part speed . The T URBO-AE code is applied to
Reducing noise has become a sign ificant criterion in the calculate the flutter observed in the wind tunne l and to
design of modem turbomachinery fa ns. For fa n with understand the fl ow fea tures that caused the instability.
super onic tip speeds, shock waves playa sign ifi can t role in the Modem fa n bl ades, by virtue of having a small thi ckness
generation of noise. Aggressive blade designs with forward ratio coupled with forward sweep could be highl y fl exibl e in
sweep and lean are being used to improve the noi se the tip region . Hi gh rotational speed and large ae rodynamic
characteristics. These geometri c features coupl ed with th e loading result in significant changes in bl ade shape. It is
presence of shock wave can lead to fl utter instability. Flutter important to accurately account for these changes in the
problems, us ually detected during the design phase, result in analysis. To investigate the sensitivity of damping calculations
program delays and cost overruns, adversely impacting the to changes in steady blade defl ection, calcul ation are also
deve lopment and mai ntenance costs . umerical flutter carried out by deforming the blade at the tip to change the bl ade
prediction methods will help in achi eving the design obj ectives tw i t by 0.5 deg such that the incidence angle at the sectio n is
and developing a fl utter free des ign . increased. Al so, results are obtained to investigate the impact of
Several methods have been developed fo r the prediction of time step used in the analysi and sensiti vity to natural
aeroelastic characteristic of turbomachine with va ri ous frequency va ri ation. These studi es are carri ed out to gain an
degrees of fidelity . AeroeJastic analyses based on energy understanding of issues that need to be addressed in fl utter
exchange between vi brating bl ades and the surroundin g fluid , analyses of simil ar fan confi guration.
have been reported for turbomachines using semi-analytical
methods (La ne and Friedman (1] , Smith [2]), lineari zed THE TURBO-AE CODE
•
methods (Hall and Clark [3], Verdon [4]), Eul er methods (He Th e aeroelasti c solver T URBO-AE is briefl y described in
[5] , Gerolymos and Vallet [6], Bakhl e, et al. [7]) and viscous thi s section. The solver can model multipl e bl ade rows
methods (Gile and Haimes [8 ], Siden [9], He and Denton undergoing harmonic oscillations with arbitrary inter-blade
(1 0]). A limited number of coupl ed aeroelasti c analyses of phase ang les ( /BPAs). It is based on a avier-Stokes un steady
ae rodynamic sol ver for interna l fl ow calcul ations of axial fl ow

* ASA Re ident Re earch As ociate at Glenn Research Center.

NASAfTM-2002-21 18 18
turbomachineiy components, TURBO [15, 16]. Viscous effects on the suction surface over the entire span. A normal shock
are modeled using the Reynolds-averaged avier-Stokes extends across the blade passage in the aft section of the blade
equations. The two-equation k-£ turbulence model is used for
closure. The aerodynamic equations are solved using a finite
volume scheme. Flux vector splitting is used to eva luate the
flux lacobians on the left-hand side. The right-hand side fluxes
are discretized using the high order Total Variation
Diminishing (TVD) scheme based on Roe's flux difference
splitting. ewton subiterations are used at each time-step to
maintain accuracy. Symmetric Gauss-Seidel iterations are
applied to the discretized equations for improved convergence.
The aeroelastic characteristics of the rotor are obtained by
calculating the energy exchange between the vibrating blade
and its surrounding fluid . Positive work on blade indicates
instability. T he aeroe lastic analysis is perfonned by first
obtaining the "steady" aerodynamic so lution for a given
operating condition. The blades are then forced into a
prescribed hannonic motion (s pecified mode, frequency, and
IBPA) to calculate the unsteady aerodynamic response and
thereby work-per-cycle W , Eq. 1.
Figure I . Transonic fan

w= ~ f _p .d A . (dX)dt ( I)
surface dt
Where, p is the surface pressure, A is the area of the blade
surface and X is the blade displacement associated with the
forced harmonic vibration. The work W is then used to
calculate the aerodynam ic damping y, using the average kinetic
energy K£ of the blade over one vibration cycle [17], Eq. 2.
w
Y=- - - (2)
8nKe
The blade motion is simulated using a dynamic grid
deformation technique. Phase-lagged boundary conditions are
used to calculate the non-zero IBP A vibrations, which
eliminates the need to model multiple blade passages.

RESULTS AND DISCUSSION

A sca le model of an experimental transonic fan with 22
blades (Fig. 1) was tested in a rig. The design operating Figure 2. First natural mode
condition for the model tested was mass flow of 9S.S9 lbs/s
with a relative tip Mach munber of 1.4. The fan fluttered at part on the suction surface. For the in-board sections, this shock is
speed in the first natural mode [IS] , between the operating line outside the blade passage (in fro nt of blade leading edge) on the
and predicted stall line. The first natural mode is shown in pressure surface side. However, the shock is completely
Fig. 2, hotter colors signifying larger displacements. The mode ingested within the passage in the outboard sections of the
shows a high degree of twist bend coupling, especially in the blade. This shock pattern is typical for a flow condition
outboard section .. referred to as moderate to high loading for blades with a
The results are presented here for 90 percent rotational supersonic leading edge operating in subsoni c axia l flow. As
speed. A grid with approximately 2l3,000 grid points per blade the back pressure is increased, moving the operation point
passage, 46 axial by 47 radial points on each surface, and a towards the sta ll line, the shock mo ves forward and becomes
uniform tip-gap is used in the analysis. The blade shape and stronger. Tllis results in flow separation on the suction surface •
modal properties for 100 percent speed are used. in the middle section of the blade. The separation is large
Figure 3 shows the steady aerodynamic loading on the enough to stall the blade resulting in a break down of the
blade surfaces for 90 percent speed at a condition near the numerical analysis. The observed breakdown of the analysis
operating line. Shock wave structures are clearly evident from due to flow separation co uld not be verified, as velocity profiles
these figures. An ob lique shock appears near the leading edge were not measured during the experiment.

NASAlTM-2002-2 11 SI S 2
 - - - - ----'---,

Figure 5 shows the aerodyna mi c damping calcul ated fo r

110
several different back press ures and IBPAs. For lower back
pressures, the lea t stable IBPA was fo und to be 0 deg.
OS:> However, as the bac k press ure was raised moving the fa n
operating condition to wa rds the stall line, the 32.73 deg IBPA
0.7-4 (2 nodal diameter forward traveling wave) became the least
stable. Flutter was ob erved in the wind tunnel fo r 32 .73 deg
IBPA.
0$

O.:!ll 2.5

2.25
0"" 2
Figure 3. Steady blade surface pressure near operating line for ~
90 percent speed "'"
'0.
.,
E
Dynam ic Analysis c
.2
Experiments showed blade flu tter in the first natural mode, .,
E
Fig. 2, (natural freq uency 351 Hz) for 32.73 deg [BPA (2 odal ">-
"C
0 0 .75
Diameter forward traveling wave) [ 18). To establi sh the lea t 4i
table mode the analy is was carried out fo r several pres ure < 0.5

ratios and IBPAs. 0 .25

Flow separation for higher back press ures was observed in
the hub region of the blade resul ting in unsteadiness of the ..\'9.09 ·32.73 · 16.36 0.00 16.36 32.73 49.09 65.46

mean flow. Because of the mean fl ow un teadiness, one needs Inter-blade phase angle (deg)

to ensure that the blade vi bration amp litud e is large enough so

as to mi ni mize the impact of mean fl ow unsteadiness on the Figure 5. Variatio n of aerodynamic damping with back
unsteadiness due to blade vi brations, yet should be small pressure and IBPA for the transonic fa n
enough to maintain linearity. The calcul ated time history of
mass flow is shown in Fig. 4 for fl ow with and without bl ade
vibration for a back pressure of 16.2 pounds per square inch The calcul ated van atlon of aerodyna mic damping with
(psi). The blade is forced into v ibratio n at 180 deg IBP A and back pressure is shown in Fig. 6 at 32.73 deg IBPA. As may be
351 Hz. It can be seen from Fig. 4 that fo r the vibration een in the figure, as the back pressure i increased, the
amplitude chosen for the analysi , the unsteadiness due to blade aerodynamic damping decreases rapidl y, dropping to
vibration is at least an order of magnitude larger than the mea n approximately 0.2 percent of cri tical damping at a back
fl ow unsteadi ness. It can also be seen that the fl ow conve rges to pressure of 16.4 psi. Increas ing th e back pressure a bove 16.4
periodicity in roughl y 10 oscill ations. psi for 90 percent speed resulted in sta lled flo w with large
eparation on the suction s urface in the mid section of the
89 .00 r---------- -- --------., blade. Thi s separation was fo und to be shock induced and
prevented a steady mean fl ow from which the blade vibration
analysis could be carried out. These results indicate that the
Blade Vibrations started 32.73 deg IBP A is the least stabl e IBPA for first natural mode,
after 23 00 ite rations
as observed in experiments, and that the aerodyna mic damping
rapidly decrease with increasing back pressure.
~
.Q 8S.00 From Fig. 6 it is evident that the characteri stics of va riation
u..
of ae rodynamic damping changes sign ificantly as the back
~
11)
pre sure is vari ed . Initi all y the damp ing is not affected much
:E No Vibration
83.00 by an increase in bac k pressure, at least up to the operating line,

~:.l
which falls cl ose to a back pressure of 15.7 P i. Further
Blade Vibration increase of the back pressure results in a dramatic change in
• damping van atlOn, becoming inversely proporti onal to
81.00 L - - - :S:-:!0"'
0- """'-'
1 0-f:0"'0- - :1C:::S"'"
0:::-
0- -=
2 -:'
00"0''---2"'S-f:0"'
0- -:3'""0"'"
00 ::---"""""'
3-d-
SO' 0
Iteration
vari ation in back pressure. To under ta nd the reason for such a
change, fl ow detai ls were inve tigated fo r back press ure va lues
of 15.2, 15.9 and 16.4 psi. Figure 7 shows the di stribution of
Figure 4. Variation of ma s flow for the transo nic fa n, wi th and wo rk on the bl ade surface . Majori ty of the bl ade shows little
without blade vibratio n

ASA/TM-2002-2 118 18 3
 1 ~ r-------------------------------~ blade positive work is being done on the blade indicating a
destabilizing contribution. These plots, when superimposed
1.5 with relative Mach number plots, indicate that the positive
;t. work regions are centered around the shock location.
~ 1.25
Cl
c::
To investigate this in more detail, the pressure coefficient
C. at 90 percent span, around the shock location on the suction
E
co
c surface, is plotted in Fig. 8 for the three back pressures.
.Y Instantaneous pressure coefficient at five instants of time over
E 0 .75
IV
c::
>-
one period of osci llation is shown. The steady pressure
e~ 0 .5
100 Steps/Cycle
20 0 Steps/Cycle coefficient is also shown to highlight the variation of shock
strength from the mean position. It is clear that as the back
0.25 pressure is increased the shock moves upstream. Also, the
• variation in the pressure change associated with the shock,
changes significantly. It is these changes that contribute to
~3.5 14.5 15.5 16 .5
Back Pre ssure (PSI) positive work on the blade. These figures indicate that the
Figure 6. Variation of aerodynamic damping for 16.4 PSI back location of shock and associated changes in strength are
pressure. Forced vibration in the ftrst natural mode at 351Hz responsible for the change in characteristics of the vari ation of
and 32 deg lBP A dampi.ng and are responsible for flutter.

0.5 G T /5
contribution towards establish ing the flutter characteristics of 2T/5
the blade. Outboard sections of the blade contribute most 0.4 3T/5
significantly. This has a side benefit from an analysis 4T/5
perspective. Resolving the flowfield on the inboard sections is 0.3 T
probably not significant and hence coarse grid may be used in
0.2
this region , concentrating more grids in the outboard sections of
the blade. These figures also show that over certain areas on the c.
U 0.1
I

-0.1

-0 .2

-o·1l.s 0 .6 0.7 0.8 0 .9

X/C
(a) back pressure 15.2 psi

0.4 G T/5
2T/5
0.3 3T/5
V
4T/5
0.2
T
Steady
0.1

152 PSI 15.9 PSI 164 PSI Q.

0
~

Work -0.1

-0.2
•
-0.3

Figure 7. Work distribution on blade surface for various back

-0·~.4 5
pressures 0 55 0.65 0.75 0.85 0.9 5
•
X/C
(b) back pressure 15 .9 psi

ASNTM--2002-2 11818 4

I
I
---~- - - _ _. ~-~~ --.- ~- - _ _ _ _ __ . _ _ _ _ _ _ _ __ _ _ _ __ _ _ _ _ _ --1I
 blade. Structural analysis showed the bl ade twist change at the
0.3 T /5
8 tip to be of this order for the 90 percent speed. T he variation in
2T/5
aerodynamic damping with oscill ation cycles is shown in
02
=v 3115
Fi g. 9. The changes in twist reduce the damping moving it
41 15
0.1 T closer to instabi lity. Earl ier stud ies have shown smaller time
Steady tep to have moderate impact on the damping calculati on, see
Fig. 6. As the calculated damping is clo e to zero, the impact of
Q.
·01 small er time steps was also investigated. Most of the analysis
't
was carried out using 100 steps per osci ll ati on cycle in order to
·02 minimize the computational cost. The comp utational cost
varies linearl y with the number of steps per cycle used in the
·0 3
analysis. Three different time step were used, 100, 200 and
·04 300 teps per oscillation cycle. The impact of time step ize is
also shown in Fig. 9. From 100 teps per cycle to 200 steps per
.Oib5 045 0 55 065 0 75 0.85 cycle there is a measurabl e change in damping. Whereas u ing
X/C 300 steps per cycle the change is small and does not justify the
(C) back pressure 16.4 psi additional computatio nal cost. From this stud y it wa decided
that the 100 step per cycle is good for investigating the trends
Figure 8. Pressure coefficient variation over one period of of the behavior. For cases where a more accurate da mping
osci ll ation cycle for back pressure of 15.2, 15 .9, 16.4 psi va lue is required, 200 steps per cycle should be used. Since the
difference is approximately 0.2 percent of critica l damping, use
of small er time steps wi ll be signifi cant onl y in the
The blade natural vibration frequency for a rotor is neighborhood of zero damping or flutter boundary.
proportional to the rotational speed. Hence, for 90 percent Figure 10 compares the steady pressure coeffic ient and
speed the blade natural frequency would be lower than that at work contributions from the suction and pressure surfaces at
100 percent speed. To investigate the impact of change in the 95 percent span location. T he results for the original grid
frequency, a sens iti vity study was carried out by varying the are (Fig. 10(a» obtained usi ng I 00 steps per cyc le, where as for
natural freq uency. Calculations were performed for 20 percent the deformed grid the res ults (Fig. I O(b») are for 200 teps per
higher as well as 20 percent lo wer natural frequency for 100 cyc le. These comparisons are to under tand the differences in
percent speed at 32.73 deg !BPA. These changes in frequency calculated damping for the deformed shape. Primari ly the
resulted in changes of less than 0.1 percent of critical damping
ratio, indicating a low sensitivity to natural frequency. It
should be noted here that change in rotational speed will also
0 .95
impact the mode shape, however, in the present stud y no
attempt was made to alter the mode shape.
Analysis was a lso performed at the second natural
vi bration mode, which wa stabl e in the experiment. The ~ 8 Deformed, 100 S/cyc
ae rodynamic da mping calculated was much higher than tho e ~ 0 .7 Deformed, 200 Sleye
for the first mode, indicating the econd mode to be more stable 01
c: o Deformed, 300 S/cyc
than the first mode. These re ults show that the anal ysis Q. - - <7- - Undeformed, 100 S/cyc
E
succe sfully calculates the natural mode and IBPA of instability 10
C
identifying the flutter characteristics of the transonic fan. o 0.45 $>---0.
For the back pre sure of 16.4 psi , the experiment sho wed E
10
/ ',
flutter, whereas the analysis predicted the fan to be marginall y c:
>-
stable. The ana lysis correctly predicted the least stable mode,
ho wever, it did not predict a negative ae rodynamic damping. A
¥
~ 0.2
possibl e rea on for that was thought to be the use of an
inacc urate operating blade shape. The analysis at the 90 percent
speed was performed using the des ign speed geometry and
characteristics. To investigate the influence of changes in blade
• shape on the aerodynamic damping, the analysis was carried
out by altering the twist distribution, such that the blade twist
Oscillation Cycle

changed from mid-span to tip varyi ng linearl y from zero

Figure 9. Effect oftim e steps and change in twist di trib ution
change at mid span to 0.5 deg change at the tip . The change in
for a 0.5 deg change in twist at tip of the blade
twi t di trib ution is such that it increases the incidence of the

ASAlTM-2002-2 11 818 5
suction surface contributes to the positive work whereas the contribution associated with this shock has also moved. Only
pressure surface contributes to negative work providing a minor changes in the level of work are observed. A second
stabilizing influence. The area of positive work is centered change, which is more significant from the work perspective, is
around the normal shock on the suction surface. However, there seen on the pressure surface. The area of negative work in the
are two main differences that are observed from these figures . leading edge region is reduced, and the peak is higher but
Fir t, we find that the shock on the suction surface has moved narrower for the deformed shape. This change contributes
upstream by roughly 5 percent. In addition, the work mainly to a reduction in the damping for the deformed sbape.
The primary reason for this difference is the motion of the
shock on the pressure surface. Figure 11 shows the steady
0.0016
pressure contours and Fig. 12 show the instantaneous pressure
0.8
8. C. ' suction contours at five instants of time over the vibration cycle for
o
--{>-
C.' pressure
Wall<, suction steady and deformed geometry at 95 percent span. The steady
0.6 0.0012
- -<3- - Wall<, pressure contours show that a normal shock wave extends from the
Wall<, tolal suction surface across the blade passage. For the original grid,
0.4 0.0008
the shock intersects the pressure surface at the leading edge,
whereas for the defo011ed geometry the shock is just ahead of
the leading edge. Figure 12 shows the shock movement over
"?
.. 0
the oscillation cycle for the two geometries consi dered here .
Instantaneous pressure contours near tbe pressure surface
-0.2
leading edge are shown for five time instants over one cycle of
vibration . For ease of comparison, these contours are plotted
next to each other going from left to right with increasing time.
The pressure surface, leading edge, and time instant
corresponding to the contour are marked on the figures. From
Fig. 12 a it can be seen that the shock stays on the blade surface
for the original geometry, whereas for the deformed geometry
(Fig. 12 b), the shock intersects the blade for part of the
X/C
vibration cycle, moving on and off the blade surface over the
oscil lation cycle. This shock motion on and off the blade
(a) Original Geometry

0 .8 0.0016
Cp, suction
o Cp ' pressu re
0.6 - 1>--- Work, suction 0.0012
- -<3- - Work, pressure
Work, tolal
0.4 0.0008

0.2 0.0004

Q.

<.? 0

-0 .2 -0 .0004

-0.4 -0 .0008

-0 .6 -0 .0012

X/C

(b) Deformed geometry •

Original geometry Deformed geometry
Figure 10. Pressure coefficient and work distribution at
Figure 11. Pressure contours for 16.4 psi at 95 percent span
95 percent span

NASA!TM-2002-211 8 18 6
surface is the reason for differences in work contributi on on the CONCLUDING REMARKS
pressure surface between the two geometries. The effect of thi s An aeroelastic analysis program based on the avier-
motion is to reduce the overall stabilizing effect fro m the Stokes equation has been used to calculate the flutter
pressure 's urface, reduci ng the aerodynamic damping for the characteristics and to investi gate the behavior of ae rodyna mi c
deformed geometry. damping vari ation for a forward swept transonic fa n geometry.
Calculated vari ations of aerodynami c dam ping with back
pressure and IBPA is presented for a condition where flu tter
occurred during testing. Although a negative aerodynami c
damping was not ca lc ul ated, the analys is yielded some good
results and trends. The analysis correctly predicted the mode
and !BPA of flutter. Also, the trend for ca lculated aerodynamic
damping clearl y indicated a negative aerodynami c damping
wo uld res ul t with further increasing the back pressure.
Unfortunately, the analys is showed that a stall ed flow field
wo uld emerge for any further increase of the back press ure .
Although flow separation was captured along with vo rtex
shedding, deep bl ade stall po ed a numerical pro bl em. Res ul t
also sho wed that a majority of the bl ade surface does not
contribute towards the stabili ty characteristics. But the
stabilizing and destabilizing effects were concentrated near the
shock location and were restri cted to outboard stations of the
blade.
Detail ed examination of the fl ow features indicated that
nom1al shock on suction surface has a destabili zing effect and
results in a large area of positi ve work centered around the
nom1al shock. The area of the destabilizing effects depended
(a) Ori gi nal geo metry on the location of the shock. Higher back pressure moved the
normal shock forward increasing the area of positive work.
This coupl ed with the mode shape res ulted in an increased
variation of the blade surface pressure over one oscillation
cycl e causing the destabilizing influence to increase.
A study was also performed to understand the effect of
various parameters on the ca lcul ation of ae rodynamic damping.
Th e stud y showed that the aerodynamic damping was
insensiti ve to the variations in vibration frequency. However, it
should be noted that onl y the frequency was altered without
altering the mode shape. It was also fo und that the blade shape
under operating conditions is important. A small change of 0.5
deg in twist at the blade ti p led to a better correlation of
damping with measurements, where the original shape had
predi cted marginal pos itive damping. Changes in bl ade shape,
although moved the suction surface shock further upstream in
the outboard sections, do not measurably change the level of
positive work. On the other hand, the negative work on
pressure surface was reduced in the fro nt section of the blade
reducing the overall blade stability, resul ting in better
correlati on with measurements. Thus it is important to use the
(b) Deformed geometry correct shape in the analys is for more accurate calcul ation of
damping, especially near th e flutter boundary.
Figure 12. Instantaneous pressure conto urs over one vibrati on Changes in time step fo r unsteady analysis were fo und to
cycl e with time period T, near press ure surface leading edge at have a moderate impact. Although the trends di d not change,
95 percent span for back pressure of 16.4 psi the small er time step reduced the aerodynam ic damping.
However, th ese di fferences wo uld be of sign ifica nce fo r
Approximately 95 Mega Words of computer memory and dampin g calcul ati ons onl y in the vicinity of neutral stability.
one hour of computational time per 100 steps was required on
the Cray C-90 computer located at the AS fac ility of ASA.

ASAJTM-2002-2118 18 7
 From these studies the following conclusions can be drawn [8] Giles, M ., Haimes, R. , "Validation of a umerical
that could be useful in similar analyses : Method for Unsteady Flow Calculations," ASME Paper
91 - GT- 27 I, 1991.
(a) Shock location and its strength have a strong influence
on blade stability [9] Siden, L.D.G ., " umerical Simulation of Unsteady
(b) In-board sections of the blade do not contribute Viscous Compressible Flows Applied to a Blade Flutter
towards establishing the stability characteristics, thus a Analysis," ASME Paper o. 91 - GT- 203 , 1991.
coarser grid in this region can be used to reduce
computational cost. [10] He, L. and Denton, J.D., "Three Dimensional Time
(c) Accurate blade operating shape is important for Marching Inviscid and Viscous Solutions for Unsteady
accurate damping prediction as it establishes the shock Flows Around Vibrating Blades," ASME Journal of
structure on the blade, which in turn affects the TurbomachinelY, Vol. 116, pp. 469-476, 1994.
stability. Variations in vibration frequency were found
to be less significant. [II] Williams, M.H ., Cho, J. , Dalton, W .N ., " Unsteady
(d) Larger time steps can be used away from the neutral Aerodynamic Analysis of Ducted Fans," Journal of
stability region for determining the trends, however, Propulsion and Power, Vol. 7, o . 5, pp. 800-804,
close to neutral stability; smaller time steps need to be 1991 .
used for accurate stability predictions.
[12] Gerolymos, G.A, "Coupled 3- D Aeroelastic Stability
REFERENCES Analysis of Bladed Disks," ASME Paper no . 92- GT-
[1] Lane, F. and Friedman, M ., "Theoretical Investigation of 171,1992 .
Subsonic Oscillating Blade-row Aerodynamics," NACA
T 4136, 1958. [13] Srivastava, R., and Reddy, T.S .R., " Comparative Study
of Coupled-Mode Flutter-Analysis Methods for Fan
[2] Smith, S. ., "Discrete Frequency Sound Generation in Configurations," Journal of Propulsion and Power,
A)lial Flow Turbomachines," R&M 3709, British Vol. 15, No.3 , pp. 447-453 , 1999.
Aeronautical Research Council, London, England, UK,
1972 . [14] Srivastava, R ., Bakhle, M .A., Keith, T.G. Jr., Stetko
G.L. "Aeroelastic Stability Computations for
[3] Hall, K.C. , Clark, W .S., "Linearized Euler Predictions Turbomachinery," Proceedings of the 2001 International
of Unsteady Aerodynamic Loads in Cascades," AIAA Forwn on Aeroelasticity and Structural Dynamics,
Journal, Vol. 31 , pp.540-550, 1993 . Madrid, Spain, Vol. III pp. 77- 86, June 200t.

[4] Verdon, J.M ., "Unsteady Aerodynamic Methods for [15] Chen, J.P., " Unsteady Three-Dimensional Thin-Layer
Turbomachinery Aeroelastic and Aeroacoustic Navier-Stokes Solutions for Turbomachinery in
Applications," AIAA Journal, Vo!' 31 , o. 2, pp . 235- Transonic Flow," Ph. D . Dissertation, Mississippi State
250, 1993. University, 1991.

[5] He, L. , "An Euler Solution for Unsteady Flows Around [16] Janus, J.M ., " Advanced 3- D CFD Algorithm for
Oscillating Blades," ASME Journal of Turbomachinery, Turbomachinery," Ph. D. Dissertation, Mississippi State
Vol. 112, 0.4, pp. 714-722, 1989 . University, Mississippi , 1989.

[6] Gerolymos, G.A. , and Vallet, 1., "Validation of 3- D [17] Carta, F.O ., " Coupled Blade-Disk-Shroud Flutter
Euler Methods for Vibrating Cascade Aerodynamics," Instabilities in Turbojet Engine Rotors," Journal of
ASME Paper 94-GT- 294, 1994. Engineering f or Power, Jul y 1967, pp . 419-426.

[7] Bakhle, M. A , Srivastava, R ., Keith, T. G . Jr., Stefko, [18] Fite, B.E., Private Communications, ASA Glenn
G. L. , " A 3D EulerlNavier-Stokes Aeroelastic Code for Research Center, January, 2001.
Propulsion Applications," AIM Paper 97- 2749, 1997.

NASA/TM-2002-211818 8

I
I
 Form Approved
REPORT DOCUMENTATION PAGE
I OMB No. 0704-0188
Public reporting burden for this co/lection of information is estimated to average 1 hour per response, including the time for reviewing instructions, searching existing data sources,
gathering and maintaining the data needed. and completing and reviewing the coliection of information. Send comments regarding this burden estimate or any other aspect of this
collection of information, including suggestions for reducing this burden, to Washington Headquart ers Services, Directorate for Information Operations and Reports, 1215 Jefferson
Davis Highway. Suite 1204 , Arlington. VA 22202-4302. and to the Office of Management and Budget. Paperwork Reduction Project (0704·0ISS). Washington. DC 20503 .
1. AGENCY USE ONLY (Leave blank)

4. TITLE AND SUBTITLE

r' REPORT DATE
September 2002
r' REPORT TYPE AND DATES COVERED
TechmcaJ Memorandum
5. FUNDING NUMBERS

Flutter Analysis of a Transonic Fan

VVU-781-30-11-00
6 . AUTHOR(S)

R. Srivastava, M.A. Bakhle, T.G. Keith, Jr., and G.L. Stefko

7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES) 8 . PERFORMING ORGANIZATION

REPORT NUMBER
National Aeronautics and Space Administration
John H. Glenn Research Center at Lewis Field E-13511
Cleveland, Ohio 44135- 3191

9. SPONSORING/MONITORING AGENCY NAME(S) AND ADDRESS(ES) 10. SPONSORING/M ON ITORING

AGENCY REPORT NUMBER

National Aeronautics and Space Administration

Washington , DC 20546-0001 NASA TM-2002-21181 8

11 . SUPPLEMENTARY NOTES
Prepared for the Turbo Expo 2002 cosponsored by the American Society of Mechanical Engineers and the International
Gas Turbine Institute, Amsterdam, The Netherlands, June 3-6, 2002. Portions of this material were presented at the
meeting. R. Srivastava, M.A. Bakhle, and T.G. Keith, Jr., University of Toledo , Toledo, Ohio; and G.L. Stefko, NASA
Glenn Research Center. R. Srivastava and M.A. Bakhle, NASA Resident Research Associates at Glenn Research Center.
Responsible person, R. Slivastava, organization code 5930, 216-433-6045.
12a. DISTRIBUTION/AVAILABILITY STATEMENT 12b. DISTRIBUTION CODE

Unclassified - Unlimited
Subject Category: 07 Distribution: Nonstandard
Available electronicall y at http://gltrs.grc.nasa.gov
This publication is available from the NASA Center fo r AeroSpace InfOlmation, 301--621-0390.
13. ABSTRACT (Maximum 200 words)

This paper desclibes the calculation of flutter stability characteristics for a tran sonic forward swept fan configuration
using a viscous aeroelastic analysis program. Unsteady Navier-Stokes equations are solved on a dynanlically deforming,
body fitted, grid to obtain the aeroelastic characteristics using the energy exchange method. The non-zero inter-blade
phase angle is modeled using phase-lagged boundary conditions. Results obtained show good con'elation with measure-
ments. It is found that the location of shock and variation of shock strength strongly influenced stability. Also, outboard
stations primarily contributed to stability characteristics. Results demonstrate that changes in blade shape impact the
calculated aerodynamic damping, indicating importance of using accurate blade operating shape under centrifugal and
steady aerodynamic loading for flutter prediction. It was found that the calculated aerodynamic damping was relatively
insensitive to vruiation in natural frequency.

14. SUBJECT TERMS 15. NUMBER OF PAGES

14
Aeroelasticity; Turbomachine; Fan; Viscous; Numerical 16. PRICE CODE

17. SECURITY CLASSIFICATION 18. SECURITY CLASSIFICATION 19. SECURITY CLASSIFICATION 20. LIMITATION OF ABSTRACT
OF REPORT OF THIS PAGE OF ABSTRACT

Unclassified U nclassifi ed Unclassified

NSN 7540-01-280-5500 Standard Form 298 (Rev. 2-89)
Prescribed by ANSI Std. 239·18
298-102

______ _ _ ____________ ~ ___________ ~~~ __ ~ _______ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ __ _ _ _ _ _ _ _ _ _ ~ _ ___.l