You are on page 1of 15

A Semi-Empirical Model for Dynamic Stall

J. G . Leishman* T. S. Bcddoes
Senior. Aerodynamicist Aerorlyr~arnicsS ~ ~ e c i a l i s l
Aerodyrlarnics Department
Westland Helicopters Ltd.
Yeovil, E ~ l g l a n d

A semi-empirical model is formulated to represent thc unsteady lift, drag, and pitching moment characteristics
of an airfoil undergoing dynamic stall. The model is presented in a form which is consistent with an indicial
formulation far the unsteadv aerodvnamics under attached now conditions. The onset of vortex shedding.during
dynamicstallis represented using a criterionfor leading edgeor shockinducedseparatian based on the attainment
- " .oressure. The induced vortex lift is re~resentedempirically along.with the associated
of a critical leadine edee
pitching moment which is obtained by allowing the center of pressure to move in a time dependent manner
durine-dvnarnic
. stall. Sienificant nonlinearities in the airfoil behavior associated with trailing edge
~~ - reparation
~

are represented using a KirehhoR flaw madcl in which the separation point is related to the airfoil behavior.
These effects are represented in such a way as to allow progressive transition between the dynamic stall and
the static stall characteristics. It is shown how the above features may be implemented as an algorithm suitable
for inclusion within rotareraft airloads or aeroclasticity analyses. Validation of the model is presented with force
and moment data from two-dimensional unsteady tests on the NACA 0012, HH-02, and SC-1095 airfoils.

Nomenclature c, = Pressure coefficient


= Vortex lift increment
a = Sonic velocity, rnls c,.
= Coefficients of indicia1 response function cw = Work per cycle coefficient = - $ C M d a
At
= Exponents of indicia1 response function D,D,,D, = Deficiency functions
bi
c = Airfoil chord, m f = Trailing edge separation point
k = Reduced frequency = o,l2V
c, = Chord force coefficient
K = Noncirculatory time constant multiplier
CD = Pressure drag coefficient
= Zero lift (viscous) drag coefficient K,,,K,,K2 = Coefficients representing pitchlng moment
Cuo curve fit
c~ = 114-chord pitching moment coefficient
M = Mach number
GI = Vonex induced pitching moment coefficient
11 = Current time sample
c~ = Normal force coefficient S = Nondimensional distance travcled by airfoil in
CN, = Critical normal force coefficient delimiting at-
semi-chords = 2Vrlc
tached flow
st, = Coefficients of separation point curve fit
c$ = Circulatory normal force coefficicnt
Sr
s
2
= Strouhal number
C!N = Noncirculatory (impulsive) normal force coef-
t = Time, s
ficient
TI = Noncirculatory time constant = cla
= Normal force under potential flow conditions
c!
CN = Vortex induced normal force coefficient
T,,T,. T,,,Ts, = Time constants (semi-chords)
T,,r = Vortex passage time constant
CN- = Normal force (lift) curve slope, rad V = Free stream velocity, r r ~ l s
c.,, = Maximum normal force coefficient x = Nondimensional chord
CP,. = Voltex induced center of pressure ID< = Aerodynamic center
X,Y = Circulatory deficiency functions
a = Angle of attack, deg
'Currently Assistant Professor. Department of Aerospace Engineering,
University of Maryland. a~ = Effective angle of attack, deg
Revised version of a paper presented at the 42nd Annual Forum of thc l3 = Prandtl-Glauert compressibility factor =
American Helicopter Society. Washington, D.C., Jun 1986. VTTP
4 J.G. LEISHMAN JOURNAL OF THE AMERICAN HELICOPTER SOCIETY

7. = Nondimensional vortex time operating regimes, unsteady aerodynamic effects are of low
$ = Indicia1 response function magnitude and can be justifiably neglected in any analysis.
T = Chordwisc force recovery factor However, if the angle of attack of the blade sections bccomes
UJ = Circular frequency, radls large enough, dynamic stall may occur. Typically, this occurs
on the retreating hlade under conditions of high blade loading
and in high speed forward flight. The rotor operational limi-
Introduction tations (i.e., vibration, aeroelastic stability, maximum control
loads and fatigue limits) are all determined by the onset of
T he successful design of advanced rotorcraft requires the
. . to confidently predict thc large unsteady and vi-
abillty
bratory loads generated and transmitted by the rotor system.
transient flow separation such as dynamic stall.
Many experimental tests (Refs. 1-5) have shown that the
The capability to accomplish this has improved significantly in distinguishing feature of dynamic stall compared with static
recent years as a result of advances in the analytical modeling stall is the shedding of significant concentratcd vorticity from
of hlade structural dynamics, the rotor wake geometry, and the airfoil leading-edge region. This vortex disturbance is suh-
unsteady aerodynamics. While structural dynamic modeling sequently swept over the airfoil chord and induces a strong
has now reached a good level of maturity, the development of moving pressure wave on the airfoil surface. These pressure
accurate andcomputationally efficient aerodynamic models that changes result in significant increases in airfoil lift and in large
represent the unsteady behavior of the hlade sections still poses nose-down pitching moments well inexcess of the static values.
a major challenge to the rotor analyst. Generally, the analyst A typical example is illustrated in Fig. 1 which also identif~es
needs to determine an appropriate compromise between the the main features of the flowfield during dynamic stall. Further
accuracy of a given aerodynamic model and the need to keep qualitative descriptions of the dynamic stall phenomenon can
computational requirements within practical limits. In many be found in Refs. 6 and 7.
circumstances, these requirements are conflicting and the rotor For repeated excursions into stall, considerable hysteresis in
analyst is often forced to resort to a relatively simple quasi- the force and moment behavior can arise and may lcad to
steady representation for the aerodynamics. Unfortunately, this reduced or negative pitch damping. If stall becomes sufficiently
cdn restrict the rangeof flight conditions over which the analysis severe on a rotor, this may excite the bladc torsion mode at its
can be applied, and thereby severely limit its generality as a natural frequency and can lead to a dynamic instability known
practical design tool. as stall flutter. Thus, to define the rotor operating envelope, it
Within a helicopter rotor flow field, the blades encounter is necessary to he able to predict this dynamic stall phenomenon
complex time varying changes in aerodynamic angle of attack and model its consequences on the performance and dynamic
due to imposed control inputs, the dynamic motion of the blades, response of the rotor.
and large local variations in inflow velocity that arise from the The unsteady aerodynamic response of an airfoil to a specific
l complex three-dimensional vortex wake system. In many rotor time history of forcing can now he determined with consid-

-A
-.-I NACA 0012, M=0.3, k=0.1 I
FLOW STRUCTURE

Unsteady FLOW REVERSALS WITHIN BOUNDARY U Y E R .


......... Static FORMATION OF VORTEX
CN " -.5 - ... ..' '..... * VORTEX DETACHES AND MOVES OVER
AIRFOIL SURFACE
VORTEX PASSES TRAILING EDGE.
FULL STALL DEVELOPS
....
.. L REATTACHMENT OF FLOW
-.5
-5 0 5 10 15 20 25

FORCES AN0 MOMENTS


EXCEEDS STATIC MAXIMUM LIFT.
EXTRAPOUTE UNEAR RANGE
PITCHING MOMENT DIVERGENCE.
2 VORTEX u n PRESENT
MAXIMUM Urn. RAPID DECAY.
-Unsteady
......... Static V MINIMUM PITCHING MOMENT, MAXIMUM DRAG

READJUST TO LINEAR RANGE


- 1
o 105 15 20 25
Angle of attack, a (deg.)
Fig. 1 Typical dynamic stall behavior of a NAC:A 0012 airfoil at M = 0.3. (Adapted from Ref. 6)
JULY 1989 DYNAMIC STALL MODEL 5

erable detail and accuracy using computational fluid dynamic the sum of two parts; one for the initial noncirculatory loading
(CFD) methods. For example, numerical solutions to the un- (which comes from piston theory, e.g. Ref. 19) and another
steady Navier-Stokes equations are now becoming increasingly for the circulatory loading which builds up quickly to the steady
feasible (Refs. 8 and 9) and have shown some recent succcss state value. The scaling normally adopted to nondimensionalize
in modcling dynamic stall. Unfortunately, these aerodynamic time produces the parameter S = 2Vtlc which corresponds to
solutions are extremcly complex and the required computational the relative distance traveled by the airfoil in terms of semi-
resources are, for mist routine rotor ahlyses, prohibitive. chords. For a step change in angle of attack, the normal force
Nevertheless, CFD methods are providing considerable insight and pitching moment coefficients can be written in theS domain
to the aerodynamic problems encountered by rotorcraft (Rcf. as
10) and ultimately will offer the most complete model of the
flowfield. It appears, however, that for the foreseeable future
more approximatc aerodynamic solution methods must con-
tinue to be used in most rotor performance and aeroelasticity
analyses. Clearly, this poses somewhat of a dilemma for the
analyst as the consequences of complex viscous effects must
be modeled within the practical constraints imposed by the
comoutational enormitv of overall rotor andlvsis.
'1'0 prov~dox1111croprosollt;!tlon of thc unstc:tdy d c r o d ) r ~ a n ~ ~ c
behavior of the bl;ale sections. :I numbcr of C~irlvso~h~\lic:ttcd
semi-empirically based modelshave been developed (e.g., Refs.
11-15). A good review of some of the capabilities of some
other recent dynamic stall prediction methods is given in Ref.
16. Most of the semi-empirical models in use for helicopter
rotor analyses have incompressible flow approximations for the
unsteady aerodynamics under attached flow conditions and, for
The circulatory part of the indicial response, +C(S, M), for a
the dynamic stall regime, they often rely heavily on the syn-
step change in angle of attack has been shown by Heaslet et
thesis of wind tunnel data from unsteady airfoil tests. In thc
a / . (Ref. 18) to be similar to a Kiissner type gust function.
interests of computational simplicity, some models sacrifice
physical realism and so may have limited generality in appli- This indicial response can be approximated empirically in terms
of the exponential function
cation. Despite these limitations, developed methods have met
with good success and have been shown to give significant
improvements in performance prediction capability for heli-
copters. However, with the increasing operational demands that where the constants are given by Beddoes (Ref. 21) as A , =
arc placed on helicopters and the increasing use of advanced
0.3, A, = 0.7, b, = 0.14, and b, = 0.53. It should be noted
blade technology and modern airfoil sections, there is still a
that the exponents are generalized to scale with p2 to account
fundamental requirement for improved aerodynamic models
for the effects of compressibility on the build-up of the cir-
that can be applied more generally and used with greater con- culatory loads. The noncirculatory indicial functions +h and
fidence in rotor design procedures.
It is the purpose of this paper to present the results of some $LA, are also approximately by exponential functions and can
be generalized in terms of Mach number. A recent papcr by
research directed towards improving the representation of un-
steady aerodynamics within a comprehensive rotor analysis. Leishman (Ref. 22) discusses this approach in detail in which
To this end, the effects of compressibility and dynamic stall indicial moment responses and contributions due to pitch rate
motion are also defined.
are included. Furthermore, the method is presented in the time
The above step response solutions can subsequently be ma-
domain which is a necessary prerequisite to fully account for
the flowfield encountered by a helicopter rotor. The main ob- nipulated by superposition using a finite-difference approxi-
jective of the work is not simply to synthesize the unstcady mation to Duhamel's integral to construct the cumulative effect
airloads but to attempt to tackle the problem at a more physical to an arbitrary time history of angle of attack. Development of
the numerical algorithms to accomplish this were initially pre-
level of approximation. As a consequence, it is necessary to
model, in some approximate but physically reprcscntative way. sented in Refs. 21 and 23. The basics of these procedures are
the key features of the processes involved in the dynamic stall recapitulated here in the light of recent enhancements to the
representation of the indicial response functions over those
of an airfoil. It is shown in this paper that this approach can
significantly reduce the very large number of parameters that given in Ref. 23.
are often inherent within somc semi-empirical dynamic stall The circulatory normal force due to an accumulating series
of step inputs in angle of attack has been obtained using
models. Numerical procedures are presented with the primary
objective of providing a practical engineering method for in-
corporating the dynamic stall model in routine rotorcraft aero-
dynamics and aeroelastic response analyses.
where rr is the current sample. The deficiency functions are
Methodology given by

Attached Flaw Behavior


X,, = X,,_,
exp(- 6,P'AS)
A prerequisite to any unsteady aerodynamic model is the
ability to model accurately the attached flow behavior. It is
possible to formulate this problem in terms of a superposition
of indicial aerodynamic responses. The derivation of the in- Y,, = Y , j _ , exp(-bzp2AS)
dicial lift and moment functions for compressible flows has
been presented in numerous sources (e.g., Refs. 17-20). In
the present analysis, the indicial response functions given by
Beddoes (Ref. 21) and later refined by Leishman (Ref. 22) arc where AS is the incremental distance travelcd by the airfoil (in
used. These indicial responses arc assumed to be composed of semi-chords) over the sample interval, At = I,, - I , , - , , and
6 J.G. LEISHMAN JOURNAL OF THE AMERICAN HELICOPTER SOCIETY

Aa is the corresponding change in a from one time step to the gradient which may be used to denote theonset of static leading-
next. The deficiency functions represent the "deficiency" in edge stall. This criterion was subsequently evaluated by Bed-
angle of attack due to unsteady aerodynamic effects and contain does (Ref. 26) within the context of rotor airfoil performance,
all the information about the time history of the shed wake under both steady and unsteady conditions. For practical pur-
effects on the airloads. poses, Beddoes determined that, although under time-depen-
In a similar way, the noncirculatory normal force can he dent forcing conditions the pressure gradient on the airfoil at
ohtained from a given angle of attack was significantly modified, it was pos-
sible to predict the onset of leading edge separation (and hence,
dynamic stall) using a criterion in which the attainment of a
critical local leading-edge velocity (pressure) was the primary
factor. The analysis was subsequently extended by Beddoes
(Ref. 27) to encompass higher Mach number flows, where the
where in this case the deficiency function is given by attainment of a critical leading edge pressure corresponding to
shock reversal was used to denote the onset of shock induced
D,, = D,,_I exp
(2)
- stall.
In application, the leading edge pressures are related to the
normal force, C,. SO it is possible to obviate the need to com-
pute airfoil pressures by transforming the calculation to the C,
domain. From an analysis of airfoil static test data, a critical
value of CN(static) = CN,may be ohtained which corresponds
to the critical pressure for separation onset at the appropriate
The deficiency function in Eq. 8 essentially accounts for time- Mach number. Thus. a Mach number dependent, separation
history effects on the airloads due to the accumulation of wave- onset (stall) buunrlary may bc deilncd. A ~ y p ~ chotlndary
al for
like pressure disturhances. The noncirculatory time constant the NACA 0012 airfoil is shown in Fig. 2. Bound;~rieslor uthcr
governing the decay of the loads due to the propagation of airfoils may he derived accordingly if the static chordwise
pressure disturbances is given by T, = cia, which is a result pressure distributions are known. In practice, however, CN,can
given previously in Ref. 23. The factor K,, which is associated be ohtained from the value of CN(static) that corresponds
with the noncirculatory time constant, is a function of Mach to either the break in the pitching moment or the chord force
number at stall.
For unsteady conditions, there is a lag in CN(t)with respect
to changes in angle of attack: however, there is also a lag in
the leading edge pressure response with respect to CN(t).Thus,
for an increasing angle of attack, the lag in the leading edge
and is derived using the exact linear thcory of Lomax (Ref. pressure response results in the critical pressure being achieved
17). ForM+ 0, K,+ 0.75, which is the valueusedpreviously at a higher value of CN, and hence, at a higher angle of attack
in Ref. 23. The evaluation of all the noncirculatory time con- than for the quasi-ste"ady case. Thus, this mechanism signifi-
stants is discussed in detail in Ref. 22. cantly contributes to the overall delay in the onset of dynam~c
The total normal force coefficient under attached flow con- stall. To implement the critical pressure criterion under un-
ditions, CE;, is given by the sum of circulatory and noncircu- steady conditions, a first-order lag may he applied to CN(i) to
latory components produce asubstitutevalue C&(t)withthe presumption that whatever
properties apply to the pressure must also apply to Ch(t). For
a discretely sampled system, this compensation to CN(t) may
he written in numerical form as
The numerical procedures for the evaluation of pitch rate and
pitching moment terms can be evaluated in an identical way
using the indicia1 response functions of Ref. 22.
The unsteady chord (in-plane) force coefficient, C,, may be
ohtained using the effective angle of attack a,

c , ,= c," tan a," = CNa(M)ain (11)

Subsequently, the unsteady pressure drag is obtained by re-


solving the total normal force and chord force coefficients through
the pitch angle to obtain

CDn= CNnsin a,,- Ccmcos a,, (12)

It should he noted that, under unsteady (potential flow) con-


ditions, the instantaneous pressure drag may actually become
negative. Further details on modeling the unsteady chord force -0- Shook ravarrol crlferlan
Leading adga crll.rlon
...1..-
and pressure drag are given in Refs. 23 and 24.
C

Leading Edge Separation


-C1
0
The most critical aspectpf modeling dynamic stall is to define 0
I
the conditions under whlch leading edge separation occurs. U 0 .I .2 .3 .4 .5 .6 .7 .8 .g
Because of its importance to the dynamic stall problem, this Mach number. M
will be discussed first.
Evans and Mort (Ref. 25) present a useful criterion equiv- Fig. 2 Critical normal force separation anset boundary for the NACA
alent to a critical leading edge pressure and associated pressure 0012 airfoil.
I JULY 1989

where the deficiency function is given by Trailing Edge


DYNAMIC STALL MODEL

Separation
7

Before discussing the subscqucnt airloads induced after the


initiation of dynamic stall, it is necessary to discuss another
mcchanism that is also involved in most types of airfoil stall.
This is progressive trailing edge separation. Thc associated loss
The results of this compensation procedure are shown in Fig. of circulation due to trailing edge separation introduces a non-
3 where the leading edge pressure coefficient at I percent chord linear force and moment behavior, especially with the cambercd
for a harmonic pitch oscillation is plotted versus the coms- airfoils more typically used on modern helicopters. Wilby (Ref.
ponding CN(f) and C;(r). It can be seen that, for different 29) suggests that trailing edge separation may play a significant
reduced frequencies, when pressure is plotted versus C;(I) rolc in the onset of dynamic stall. However, as also discussed
using a suitable value of T,, the curves collapse to a single by Wilhy, . experimental
. tests have indicated that the occurrence
line which in fact represents the static relation. Thus, in terms 11; tr;l~lingcJgc \cparation I \ supprcsscd by incrcas~~~g p~tcl~
of the value of C;(t), the onset of leading edgclshock induced rhtc Thcdyn:~micstall prnscs, may thcn bc initi;~tc(lby lcad~na-
separation under dynamic conditions will be initiated when edge sepa;ation or shbck induce2 separation if supercritical
C;(t) exceeds the critical CN,(M)boundary. This means that flow is allowed to develop. Evcn so, when the primary source
there will be a delay in the onset of leading cdge separation to of separation is either at the leading edge or at the foot of a
higher angles of attack for increasing reduced frequency (or shock wave, it appears that this is generally sufficient to pro-
increasing pitch rate). Furthermore, if the value of C;(t) is mote some separation at the trailing edge and hence initlate
monitored throughout the calculation into stall, then it may be some nonlinear behavior in thc force and moment response.
used as an indicator for the conditions which permit flow re- One theory which models separated flow regions on 2-D
I
I attachment (i.e., if C;(t) < C,,). bodies is attributed to Kirchhoff and is reviewed in Refs. 30
It should be noted that thc time constant T, is a function of and 31. A specific casc of Kirchhoff flow is a simple model
Mach number and must be determined empirically from un- for the trailing edge separation phenomenon in which the airfoil
steady airfoil data. (Appropriate values are given for the NACA normal force coefficient, CN, may be approximated as
0012 airfoil in Ref. 23.) However, it appears that the values
of T, are largely independent of airfoil shape. Such a conclusion
may be drawn by performing a similar analysis on unsteady
data obtained by McCroskey (Ref. 28) for various airfoils.
albeit over a limited range of Mach numbers.
where 2 n is the force-curve-slope for incompressible flow, f
is the trailing edge separation point and a is thc angle of attack.
Thus, if the separation point can he determined, it is a trivial
calculation to determine the normal force. In practical cases,
this expression may be extended to encompass compressible
flows where 2.rr is qcplaced by the force-curve-slope at the
appropriate Mach number

To implement this procedure, the relationship between the ef-


fective separation point, f , and the angle of attack, a , can be
dcduced from the airfoil static CN behavior by rearrangement
of Eq. 16 to solve directly for f. The relationship between a
-2 4 . , . , , , , , , , ,
and f can be generalizcd empirically in a fairly simple manner
using the relations
-.l .I .3 .5 .7 .9 1.1
Measured CN

E 1 Test data
The coefficients S, and S, define the static stall characteristic,
while a, defines the break point corresponding to f = 0.7. It
should he noted that f ; .0.7 closely corresponds to the static

stall angle of attack for most airfoil sections. S,, S,, and a ,
are easily determined for different Mach numbers from the
static lift data.
A general expression for the pitching moment behavior can-
not be obtained from Kirchhoff theory, and an alternative em-
pirical relation must be formulated. From the airfoil static data,
the center of pressure at any angle of attack may be determined
from the ratio CM/CN(allowing for the zero lift moment CMo).
The variation can be plotted versus the corresponding value of
-2 4 , , . , . , , , . , , , the separation point and curve fitted using a suitable polynom-
-.1 .1 .3 .5 .7 .9 1.1 ial. One suitable fit is to use the form
Compensated CN via 1st order log. CN1
Fig. 3 Compensation of unsteady leading edge pressurn for the NACA
0012 at M = 0.4.
8 J.G. LEISHMAN JOURNAL OF THE AMERICAN HELICOPTER SOCIETY

where KO = (0.25 - x,) is the aerodynamic center offset inviscid flow, CDo= 0, .rl = I and f = 1; so, for steady con-
from the 114-chord. The constant K, gives the direct effect on ditions, Eq. 12 reduces to the classical result that C , = 0.
the center of pressure due to the growth of the separated flow For unsteady flow, there exists a modified separation point
region and the constant K2 helps describe the shape of the location due to the temporal effects on the airfoil pressure
moment break at stall. The values of KO,K,, K2 and rn can be distribution and the boundary layer response. An open loop
adjusted for different airfoils, as necessary, to give the best procedure has been developed to represent these effects on the
moment reconstruction. For the NACA 0012 a value of in = trailing edge separation point and thereby permit the evaluation
2 has been used; however values of m = 112 or rn = I can of nonlinear forces and moments under dynamic conditions via
be used as necessary for other airfoils to obtain the best fit with the application of the Kirchhoff theory, as above.
the test data. The procedure is performed by firstly incorporating the air-
Using the above equations, the reconstructed static lift and foil unsteady pressure response via Eqs. 13 and 14. This may
moment versus angle of attack relationships are shown for the then he used to define an effective angle of attack, a,, which
NACA 0012 airfoil at Mach numbers of 0.3, 0.5, and 0.7 in gives the same unsteady leading edge pressure as for the equiv-
Fig. 4. This procedure has also been validated for other airfoils, alent quasi-steady case
and can he applied to almost any airfoil if the static stall char-
acteristics are known a prior;.
An expression for the chord force Cc may also be deduced
from the Kirchhoff solution to the trailing edge stall problem

C& = q cNma2* (19) This value of a, may be used to determine a value for the
effective separation point f' at this a, from the static f versus
a relationship in Eq. 17. Secondly, the additional effects of
where the factor fl accounts for the influence of trailing edge the unsteady boundary layer response may be represented by
separation. Even with no separation on the airfoil, the airfoil application of a first-order lag to the value off' to produce the
does not realize 100 percent of the chord force which would final value for the unstcady trailing edge separation point f".
he attained in potential flow. Allowance for this nonrealization For a discretely sampled system, this may he represented in
is made through the recovery factor .rl which can be obtained numerical form as
empirically from static airfoil test data. Typically, q = 0.95.
The pressure drag in separated flow can be obtained through
resolution of the chord force and normal force through the
pitch angle as in Eq. 12. The total section drag is obtained
by addition of the viscous drag component C ! . Note that for where the deficiency function is

As in the case of T,, TI is a Mach number dependent time


constant (see Ref. 23), albeit a much weaker variation than
with T,. However, it is more difficult to define how this time
constant will change with airfoil shape. Without access to un-
steady airfoil data, an unsteady boundary layer code along the
lines of Ref. 32 can he practically used to determine how TI

- Reconstruction
. . also Ref. 27 for discussion).
varies with airfoil shave (see
Finally, the (nonlinear) normal force Cl, incorporating the

0
M = 0.3
M = 0.5 f" is given by the ~irchhoffreiation
" " .
effects of the modified lunsteadv) trailine edee senaration ooiut

A M = 0.7

.04 and the pitching moment by

.a2

C" O
where C$ is the circulatory normal force coefficient and CMO
-.a2 is the zero-lift moment. The contributions of the other unsteady
circulatory and noncirculatory moment terms are additive to
-.04 Eq. 24. Similarly, the chord force is given as
- Reconstruction
-.06 M = 0.3

Dynamic Stall
-.l
0 2 4 6 8 10 12 14 16
The general case of dynamic stall involves the formation of
Angle of a t t a c k . a (deg.) a vortex near the leading edge of the airfoil which subsequently
separates from the surface and is transported downstream. It
Fig. 4 Reconstruction of static normal force and pitching moment appears that, until thc vortex begins to detach, there arc no
behavior for the NACA 0012 at various Mach numbers. gross changes in the airfoil pressure distribution, and the forces
JULY 1989 DYNAMIC STALL MODEL 9

and moments can be rcpresentcd as if the developing vortex is dctermined empirically from unsteady test data, and values are
ignored. After the vortex detachcs, the stall development ap- presented for the NACA 0012 in Ref. 23. Both T,, and T,,,
pears to be govemcd by a basic common proccss; qualitatively appear rclativcly independent of Mach numbcr over most of
similar effects have been obtained for different modcs of forcing the Mach number range. However, no formal conclusion can
such as oscillatory pitch, plunge, and ramp motions (c.g., Rcfs. be made here regarding the variability of these parameters wtth
29 and 33). airfoil shape. The dynamic stall experiments performed by
.
A ohvsicallv
, accentable model for the vortex induced lift has McCroskey e t a / . (Refs. 28 and 34) indicate that, while there

I been formulated by viewing the vortex lift contribution as an


excess accumulation of circulation that is retained in the vicinitv
of the airfoil until some critical condition is reached. For a
is a significant effect of airfoil shape under light stall conditions,
all airfoils behave very similarly undcr strong dynamic stall
conditions. Thus, it can bc tentatively concludcd that the pa-
discretely sampled system, the vortex lift force coefficient rameters TI, and T,., should bc relatively insensitive to airfoil
C; is represented by assuming that, for a givcn sample period, shape.
the increment in vortex lift C,, is dctcrmincd by the difference From the above, the required total loadings can be obtained
between the instantaneous linearized value of thc unsteady cir- by superposition. For example, the total normal force coeffi-
culatory lift and the corresponding unsteady nonlinear lift as cient CN under dynamic stall conditions is given by
given by the Kirchhoff approximation, i.e.

with a similar equation for the pitching moment.

I where

KN,?= (I + flI2/4 (27)


Indications of multiple vortex shedding phenomena during
dynamic stall were apparent in many of thc ramp and oscillatory
pitch test data, especially in thc highcr incidcncc range and at
lower reduced frequencies and pitch rates. This was tnanifcstcd
by multiple peaks in the normal force, drag, and pitching mo-
At the same time, the total accumulated vortex lift CLis allowed ment rcsponse for which the secondary peak was generally of
to decay exponentially with time, but may also be updated by most significance. In the high incidence range, the frequency
a new increment. Following the approach used above, this of vortex shedding was dctermined to correspond closely to a
process may be writtcn in discrete time form as
! Strouhal number St = 0.19, which appeared relatively indc-
pendent of Mach number. A simple model for the process was
+ (C,,,, - C,,,,-,) exp developed by allowing secondary increments in vortex lift to
build up again via Eqs. 26-28 after a nondimensional time T ,
(instead of T,,,) corrcsponding to an cffcctivc flat platc Strouhal
Consequently, when the rate of change of lift is low, the vortex number based on the position of the trailing edge separation
lift is being dissipated as fast as it accumulates; in the limit as point, i.e.
the rate of changc tcnds to zero, thc airfoil charcteristics revert
smoothly back to the static (nonlinear) bchavior.
Abrupt airloading changes occur when the critical conditions
for leading edge or shock induced separation effects are mct
(i.e., Ch(t) excecds thc CN,/Mach number boundary). At this Using this simple model, the forcc and momcnt characteristics
ooint. the accumulated vortex Itft is assumed to start to convect of multiple vortex shedding were found to be reasonably well
bver the airfoil chord. The rate at which this convection process represented during both the ramp and oscillatory pitch tests (scc
occurs has been determined from exnerimental tests to be some- results below).
what less than half of the free-stream velocity, with a weak
dependence on Mach number. During the vortex convection
process, the vortex lift is assumed to continuc via Eqs. 26-28 Modifications to the Model
but the accumulation is terminated when the vortex reaches the Although the above system equations describe, in an opcn
airfoil trailing edge and is shed into the wake. To track the loop sense, the basic physical flow phenomena likely to be
position of the vortex, a nondimensional vortex time parameter encounfcrcd on most airfoils, the elements of the model are,
T, (in semi-chords) is used such that T,. = 0 at the onset of in fact, physically couplcd. For examplc, trailing cdgc sepa-
separation conditions and T,, = TI,, when thc vortex reaches the ration development will be curtailed with the onset of leading
trailing edge. edge separation. Also, the initiation of flow reattachment w ~ l l
The center of pressure on the airfoil also varies with the
chordwise position of the vortex and will obtain a maximum
value when the vortex reaches the trailtng cdgc. Based on an
shedding. -
be delayed if the airfoil is undergoing leading edge vortex
-. even thoueh thc anelc of attack mdv bc low enough
to initiate flow reattaFhment under normal con&tions. As shown
-.
\ analysis of much experimental data involving dynamic stall .
in Ref. 23. the counline " of the elements of the nonlinear model
over a wide range in Mach number, a fairly general represen- can be readily represented by a temporary modification to the
l
I
tation of the center of pressurc behavior (aft of IM-chord) was
fomiulated empirically as
appropriate time constant associated with the behavior (i.e..
by a simple halving or doubling of thc titnc constant associated
with one process during some critical phase of another process).
For example, airfoil test data indicate that leading edge or shock
induccd sc~arationmav occur very abru~tly. . . To im~lement
these phc110111cnavlathc cr~tlcalIciallng cdgc prc*surccrltcrlan.
Finally, the increment in pitching moment about the quarter- - cdcc
it is suiCictcnt to ovrr-ride thc 1;ic iisu>~.iatcJwilh tr;ailinx -
separation by halving the T, time constant. To tnaxilnize gen-
chord due to the aft-moving center of pressurc is given by erallty for other airfoil sections, these modifications were re-
stricted to the two time constants T, and T,, only.
While modifying the time constants may not bc particularly
desirable, this does avoid coupling the flow elements of the
Both the vortex decay time constant T, and the nondimcn- model using the addition of further equations and possibly the
sional time for thc vortex to traverse the chord T,,, were also use of more timc constants. It should bc noted, however, even
10 J.G. LEISHMAN JOURNAL OF THE AMERICAN HELICOPTER SOCIETY

if the modifications to the time constants TI and TI, are not a=2.1+8.2"sinwt
implemented, the basic elcmcnts of the model are still sufficient .2 7
to produce accurate predictions of the force and moment char-
acteristics of dynamic stall. Thc various stratcgics to modify
the time constant have been developed using experimental data
as a guide and by correlating the model with some 200 test
cases of dynamic stall over a range of Mach numbers from 0.3
to 0.8. These strategies are detailed in Ref. 23 and involve
minimal additional l o g ~ cin the overall algorithm. It should be
noted that modifications to the time constants are intended to
be fixed features of the model. . . . a
EXPERIMENT
Other minor modifications have bccn made to the model -.l1 ~ . ~ . ~ . ~ r . , , , , , , , . , , , , , , . ,
based on expcnence in correlating with a large amount of ex- -10 -5 0 5 10 15
perimental data for dynamic stall. These included modifications
to the chord force calculation under deep stall conditions and
the modeling of the centcr of pressure during flow reattachment
fromdynamic stall. Again, all the details of these modifications
are discussed in Ref. 23.

Results a n d Discussion
This section documents a selection of comparisons of thc
aerodynamic niodcl with unsteady test data. The bulk of the
-.l
-10
1 -5 0 5 10 15
data presented are for a NACA 0012 airfoil and wcre obtained Angle of attack. a
in A blow-down wind tunncl at the Aircraft Research Associ-
ation (A.R.A.) in the U.K. Thc tunncl has a test section which Fig. 6 Prediction of chord farce and pressure drag undcr stlsrhcd
is 45cm in hetght and 20cm wide with slotted upper and lowcr flow conditions at M = 0.4.
walls. Theairfoil model, whichspanned thc widthof the tunncl,
had a chord of IOcm. Surface pressures were nicasured using to give a Reynolds number of approximately Re = M x 10'
miniature pressure transducers distributed at 30 positions arouni which is close to full scale helicopter rotor values. Both os-
the chord at mid-span. Forces and momcnts were ohtaincd from cillatory pitch and constant pitch rate (ramp) tests were con-
the integration ofpressures. The wind-tunnel was pressurizcd ducted. For the purposes of comparing the tnodel with test
data, the timc history of the anglc of attack forcing as measured
in the experimcnt was used dircctly as an input to thc model.
Typical nomial force. and pitching moment responses are
shown in Fig. 5 for a harmonic pitch oscillation at a nominal
reduced frcquency of 0.075 and M = 0.4 under attached flow
conditions. It should be noted that, for a pure sinusoidal os-
cillation undcr attached flow conditions, nominally elliptical

Mz0.381, k = 0 . 0 7 4

-
- MODEL
EXPERIMENT

I ..... EXPERIMENT

-.l I,
-10 -5 0 5 10 15
Angle of attack, a
Fig. 5 Prediction of normal force and pitching moment under at- Fig. 7 Prediction of farccs and moments during light dynamic stall
tached flow conditions at M = 0.4. at M = 0.4.
JULY 1989 DYNAMIC STALL MODEL

1 - VORTEX DETACHES
2 - VORTEX REACHES T.E.
3 - FULL SEPARATION
4 - START O F FLOW
REATTACHMENT

a= 10.3"+8.1 "sinwt

0.0I
0 5 10 15 20
a (deg.1 MODEL
0.11 --MODEL (NO VORTEX)
EXPERIMENT (2 CYCLES) -
0.6 -
0.L-

0.2-
CD
0.0-

Fig. 8 Prediction of farces and moments during strong dynamic stall at M = 0.4.

lift and moment loops should be obtained. However, it can be In Fig. 7, the mean angle of attack is increased to 5.2" which
seen that there is some distortion in the loops, particularly for is a typical case of light dynamic stall pcnctration (i.e., the
the pitching moment. This behavior was traced to a significant maximum angle of attack is just sufficient to cause leading
third harmonic in the pitch forcing that has the effect of in- cdge separation). It can he seen that all three components of
creasing the pitch rate contributions to the airloads near the the loadine show deviations from the attached flow behavior
maxim& and minimum angles of attack. In Fig. 6, the cor- near the maximum angles of attack because of the development
responding predictions of the chord force and pressure drag are of limited flow separation. As the anele of attack is reduced,
shown. 1t:s-worth noting the behavior of the &steady prcisurc the flow quickly reattaches. ~ l t h o u &it was found that light
drag, which clearly shows that the drag becomes negative (i.e., dynamic stall was quitc difficult to model, the predictions were
a propulsive force) on the downstroke of the motion. The ex- in good agreement with the test data. The pitch damping for
cellent correlations obtained for all components of the loading this case is quite low and indicates that the stall onsct conditions
give considerable confidence in the ability to model the attached may be conducive to stall flutter.
flow behavior. In Fig. 8, the mean angle of attack is increased such that
J.G. LEISHMAN JOURNAL OF THE AMERICAN HELICOPTER SOCIETY

a=7.8"+8.4"sinwt
M=0.481, k=0.060

MODEL
- EXPERIMENT

0 20
a (deg.1
0.1 0.63
-
-

CM

-0.1 . . . . . . . . . . . . . . . . . . . . .
0 20 0 20
a (deg.1 a (deg.1
Fig. 9 Prediction of farces and moments during strong dynamic stall at M = 0.5.

2.0-
SECONDARY VORTEX

1.5-
PRIMARY VORTEX / 0.1 a=l 5.3+5.20sinwt, k=0.076

CN
1.0-

0.5-
- MODEL -0.2
--- MODEL (NO VORTEX)

0.0,
- EXPERIMENT -030 5 10
SECONDARY VORTEX
15 20 25
o 5 10 1s 20 25 a (deg.)
Fig. 10 Prediction of secondary vortex shedding during dynamic stall.
JULY 1989 DYNAMIC STALL MODEL 13

strong dynamic stall occurs. (Thc samc rcduced frequency of Of the various empirical constants that are used for the model,
0.075 is n~aintained,as in the previous ca%es.)The modeling all but four of the parameters are derived from the static airfoil
is shown with and without the contribution from thc vortcx characteristics. The necessary static parameters for the HH-02
lift. It can he seen that the vortex shedding contributes a mod- and the SC-I095 have been derived for the model using exactly
erate increment to CN?,&,,but glvcs a particularly large in- the same procedure as for the NACA 0012 discussed previ-
-
crease in nosc-down ~ l t c h i n gmoment. The maximum lift and
minimum moments are predicted very accurately along with
ously. The dynamic time constants T,,, T,, T,, and T , , were
retained from the values used for the NACA 0012 at the same
the phasing of the loads during stall. Flow reattachlncnt (in- Mach number.
dicated b y t h e return to thc nominally clliptical shapes of 'the Predictions of the lift and moment response arc shown with
loops) is delayed to a fairly low anglc of attack during the
downstrokc of the motion. It should be noted that two succes-
-
the test data in Fie. 14 for a case of strone
- dvnamic
, stall. Thc
rcsults arc in remarkably good agreement with the test data and
sive cycles of oscillatory pitch data are shown, and this serves demonstrate the abilltv of the orcsent model to oredict. at least
to illustrate thc inhcrcnt variability of the aerodynamic loads to an engineering 1e;el of approximation, th; dynamic stall
in the separated flow regime. characteristics of other airfoil sections givcn thc static stall
Another case of strong dynamic stall is shown in Fig. 9 and behavior. The results also tend to support the previously ~ n a d c
should be examined in conjunction with Fig. 8. This figure hvpothesis that the dvnamic time constants are relativelv in-
illustrates the basic similarity of the effects of dynamic stall at s;nsitivc to airfoil shape. Howcver, the validity of this fatter
a higher Mach number of 0.5, but adjusted to a lower mean statement can onlv bc confirmed whcn unstcadv airfoil data
angle of attack datum. become available ?or higher Mach numbers.
In Fig. 10, the mean angle of attack is increased further, but
the oscillation amplitude is reduced to So, still at a reduced
frequency of 0.075. Under these conditions, strong secondary
vortex shedding occurs which, in fact, is responsible for the
maximum loads on the airfoil. It can be seen that the model Conclusions
captures this secondary vortex shedding phenomenon quite wcll. 1. The objective behind the work outlined in this paper has
Some evidence of tertiary vortcx shcdding is also indicated on been to develop an improved semi-empirical model for the
the downstroke of the airfoil motion, but no attempt was made effects of dynamic stall. The approach adopted has bcen to
to model this phenomenon here. tackle the problem at a morc physical level of approximation.
Figure 11 shows the effect of ramp pitch rate on the airfoil hut still in a sufficiently simple manncr and computational form
normal forcc, moment, and drag response. It is worth noting to include it within a comprehensive rotor perfomlance or
the delav of the scparation onsct conditions with increasing aeroelasticity analysis. The ultimate objective of this line of
pitch rate, the overail increase in maximum lift, and the diskin: approach is directed towards the development of a more gclicral
periodicity of the airloads in the post stall regimc. This figure engineering model for the effects of dynamic stall that can be
also indicates the effectiveness of the leading edge separayion applied to a variety of conventional and advanccd airfoils used
criterion in predicting the delay and final onset of dynamic stall for new rotor designs.
for widely different pitch rates. For the modcling, the multiple 2 . An indicia1 Gsconse and superposition method has bcen
vortex shedding option gives a reasonable correlation with the selected for thc attached flow aerodyna~nicsalong with com-
test data in the post stall regime. pressibility corrections in order to accurately capturc both am-
Figure 12 illustrates the cffcct of Mach number on the dy- plitude and phasing of the unstcady aerodynamic response. The
namic stall development (drag not shown). In this figure, the nonlinear effects of trailing edge scparation have been implc-
Mach nu~nbcris increased from 0.488 to 0.692 but under the mented under time-dependent conditions using the Kirchhoff
same nominal forcing conditions and at approximately the same theory as a means of relating the force and moment character-
rcduced frequency. Of particular note hcre are the increasing istics to the location and progression of the trailinr edge SCD-
amounts of hysteresis in the force and moment as the Mach aration point. Features bf reading edgc or shock &duc;d
number is increased. This occurs because, at a constant reduccd separation have bcen reviewed and are implcmcntcd via a crit-
frequency, the angle of attack for the onset of leading edge ical normal force coefficicnVMach number boundary. This pro-
separation will decrease with increasing Mach numbcr. In all cedure has been extended to the dynamic regime to denote thc
threc cases, the predictions of the model are in good agreemcnt initiation of dynamic stall. Finally, the induccd lift and pitching
with the tcst data showing that the effects of Mach number on moment behavior during dynamic stall have been rcprcsented
dynamic stall are well reprcsentcd. in a physically realistic manner. All the above phenorncna are
Figure 13 shows that, by increasing the reduced frequency modeled in a manner that can be readily integrated into a sub-
for thc same nominal angle of attack forcing (c.f. M = 0.488 routine for the blade sectional aerodvnamics.
case in Fig. 12), the amount of flow separation can be sup- 3. An extensive validation of thc model has been conductcd
pressed. While nominally elliptical loops are ohtaincd at these with unsteady test data for NACA 0012 undergoing both os-
conditions, the loss of lift during the reattachment process in- cillatory pitch and ramp changes in anglc of attack. Correlation
dicates that the amount of trailing cdgc separation is still suf- with the test data was good, particularly in terms of predicting
ficient to affect the airloads. Again, the behavior predicted by the onset of dynamic stall. Thc subsequent magnitude phas~ng
the model is in good agreement with the tcst data and proves of the dynamic stall loads for variations in both Mach nunibcr
that the effects of frequency in terms of suppressing scparation and rcduced frequency was also in good agrccment with the
development are adequately represented. test data.
While the previous cascs havc rcferred specifically to mod- 4. A brief validation has also been shown for the HH-02
eling of the NACA 0012 airfoil characteristics, the ultimate and SC-1095 airfoil sections using only changes to the param-
proof of anv aerodvnamic model is the abilitv to oredict the eters of the model that can be derived directly from the static
airfoil characteristics. The predictions for these airfoils are also
in good agreement with the tcst data, and it appears the model
the scope of the present paper, it is sufficient to demonstrate may bc sufficiently general to allow its application to other
here computed predictions for two representative modern rotor airfoil sections, at least whcn cngincering levels of prediction
airfoils, na~nclya Hughes HH-02 and a Sikorsky SC-1095 capability arc required. The level of correlation obtained with
airfoil. The test data havc bccn taken from those recorded in this new unsteady aerodynamic model provides considcrablc
Ref. 28 and are for a Mach number of 0.3. confidence when applied in the design of new roton.
J.G. LEISHMAN JOURNAL OF THE AMERICAN HELICOPTER SOCIETY

2.0
M=0.501, &=8020/s " O l M=0.496. &= 1 4930,s

CN

-
-
EXPERIMENT
- - - - STATIC EXP'T

a (deg.1
0 a (deg-1 30 0 30

1 .o 1.o

cD cD

0 30 0 30
a (deg.1 a (deg.1
Fig. 11 E~Tectsof ramp pitch rate on normal farce, pitching moment, and pressure drag.
JULY 1989 DYNAMIC STALL MODEL

a=6.0+4.80sinwt a=6.0+4.6"sinwt a=5.8"+4.5"sinwt

0 20 O a (deg.) 20 0 20
a (deg.1 a (deg.)

Cw=0.277 Cw=0.328 Cw=0.381

-0.2 -0.2 -0.2


0 20 0 20 0 20
a (deg.1 a (deg.1 a (deg.1
Fig. 12 Example of the effecls of increasing Mach number on the airloads during dynamic stall.

Acknowledgments nical monitor. Some aspects of the work in this paper were
The research work described in this paper has been funded conducted by the first author with thc financial support of thc
by the Procurement Executive, U.K. Ministry of Defence. Mr. U.S. A m y Rcsearch office under contract DAAL-03-88-C-
Peter Wilby, Superintendent of the Structures Department at 002. The first author also wishes to thank colleagues at both
the Royal Aircraft Establishment, Farnborough, was the tcch- Westland Helicopters and the University of Maryland for their
helpful discussions.

References
'Manin, 1. M., Empey. R. W., McCmskey, W. I . . Caradanna. F. X.,
"An Experimental Analysis of Dynamic Stall an an Oscillating Airfoil,"
Joerrtnl of l l ~ eA,nericn,r Helico/,ler Sociely. Vol. 19, (I), Jan 1974.
'Carla. F. 0.. "AnaiysisofOscillatory PressureDatalncluding Dynamic
Stall Effects," NASA CR-2394. 1974.
-MODEL
- EXPERIMENT
'Cam, L. W., McAlister, K. W.. McCroskey, W. I.. "Analysis of the
Development of Dynamic Stall bascd on Oscillating Airfoil Experiments,"
NASA TN D-8382. 1977.
'Can, L. W., McAlister, K. W., McCroskey. W. J.. "Dynamic Stall
Experiments an the NACA 0012 Airfoil," NASA TP-1100. 1978.
'McAlister. K. W., Can. L. W., "Water Tunncl Expcrimcnts on an
Oscillating Airfoil." NASA TM 78446, 1976.
'Beddoes, T . S.. "AQualitative Discussion of Dynamic Stall," AGARD
Special Course on Unsteady Aerodynamics. AGARD Report 679, 1979.
'McCroskey, W. J . , "The Phenomenon of Dynamic Stall," NASA TM-
81264. 1981.
sJanakiram. R. ~ . , ~ a n k aL. r , N., "Emerging Role of First Principles
..
Bascd Comoutational Aerodvnamics for Rotarcraft Aoolications." ~. Pro-
ccedi!lgs of rite Zndl,~rrnrnrio,ralCo,fc~.enccon Rolorcrafl Basic Reseorck.
College Park. Maryland, Feb 1988.
'Tunccr, I . , Wu, J.. Wang, C . , "Theoretical and Numerical Studies of
Oscillating Airfoils." Paper 89-0021, presented at the 27th Aerospace
Scicnces Meeting. Reno, Nevada, Jan 9-12. 1989.
I0Srinivasan, G. R., McCroskey, W. I., "Navier-StokesCalculationsof
Haverine- Rotor Flowfields." AIAA Joar,rnl of Aircrofl, Vol. 25. (101,
Fig. 13 Suppression of flow separation by increasing the reduced 1988.
frequency. "Garment. R. E., "A Mathematical Model of Unsteady Aerodynamics
16 J.G. LEISHMAN JOURNAL O F T H E AMERICAN HELICOPTER SOCIETY

Sikorsky SC--1095 Hughes HH- 02

CD
EXPERIMENT
-MODEL
STATIC

EXPERIMENT
~ ~
\f
Y -.4 -
-5
-
----.

0
EXPERIMENT
MODEL
STATIC

5
.. a

EXPERIMENT
MODEL
STATIC
10
(deg.1
15 20 25

Fig. 14 Prediction of sirloads on HH-02 and SC-1095 airfoil sections at M = 0.3.

and Radial Flow far the Application to Helicopter Rotor," USAAMRDL "Lomax, H., Heaslet, M. A., Fullcr. F. B.. Sluddcr. L., "Two and
Technical Repon 72-67, 1973. Three Dimensional Unsteady Lift Problems in High Spced Flight," NACA
"Beddoes. T. S., "A Synthesis of Unsteady Aerodynamic Effects In- Report 1077. 1952.
cluding Stall Hysteresis." Verrico, Vol. 1, (2). 1976. "Heaslet, M. A,, Spreiter, J. R., "Reciprocity Relations in Aeradyn-
"Bielawa, R. L., "Synthesized Airfoil Data Method with Applications amics," NACA Report 1119, Fcb 1952.
to Stall Flutter Calculations," Pmceedings of Ilte 31sr A#zrzsalFonrra of "Bisplinghoff, R. L., Ashley, H., Halfman, R. L., Aernel~sricir~~,
Ad-
tlze Anreric0,t Helicopler Society, 1975. dison-Weslcv Publishine- Co., Reading,
~~~~~
. Mass.. 1955.
T r a n , C. T . , Pitot, D., "A Semi-Empirical Model for the Dynalnic 2"Lomax, H., "Indicia1 Aerodynamics:" AGARD.Manual a n Aeroelas-
stall of Airfoils in View of the Application to the Calculation of Responses ticity. Part 2. Chapter 6. 1968.
,,fa Helicopter Blade in Fanvard Flight," Verrica, Vol. 5, (I), 1981. 21Bcddoes. T. S.. "Practical Computation of Unsleady Lifl," 8th Eu-
1 5 ~ ~S. T., ~ "Synthesized
~ ~ ~ Airfoil ~ i Data, Melhod for Prediction of ropcan Rotarcraft and Powered Lift Aircraft Forum, 1982. Also Venicn,
~~~~~i~ stall and Unsteady Airloads," Proceedi8zxs of ?Ire3911, A,v,rral Vol. 7, (2). 1983.
F~~,,,, of ~ , " ~ r i c n ,Helicopler
r Sociery, 1983. Also Verrico, Voi. 8. "Leishman, 1. G., "Validation of Approximate Indicial Aerodynamic
1984. Functions for Two-Dimensianal Subsonic Flow," AlAA joa,nal @Air-
T. S. R., Kaza, K. R. V., "A Comparative Study of some crnfr, Vol. 25, (10). Oct 1988.
Dynamic Stall Models," NASA TM-88917. "Leishman, J. G.. Bcddoes, T. S., "AGeneraliredModel forunsteady
JULY 1989 DYNAMIC STALL MODEL 17

ceedittgs of
-
Airfoil Behavior and Dvnamic Stall Usine the Indicia1 Method" P.m-- ~

!he 42nd A,l,,aal Foruaz of lire Antrrirar8 Heiicoprer Sorirry.


Range af Airfoil Design Features on Dynamic Stall Onsct," Paper No. 2,
10th European Rotorcraft Forum, The Hague. The Nethcrlands. Aug 28-
Washington D.C.. Jun 1986. 31. 1984.
24Leishman, I. G.. "An Analytic Model for Unsteady Drag," AIAA 30Thwaites, B. (ed.), hrcontprrssihle Ao.odv,,a,nics. Oxford University
Joarnol qf Aircrafr, Val. 25, (7). Jul 1988. Clarendon Press. 1960.
"Evans, W. T., Mort, K. W . , "Analysis of Computcd Flow Parameters "Waods, L. C., Tire Tltro~?of Subsonic Plor~eFlotv. Cambridge Uni-
far a Sm of Sudden Stalls in Low Soeed Twa-Dimensional Flow." NACA versity Press. 1961
TND-85, 1959. "Scmggs. R. M.. Nash. J. F.. Singleton. R. E., "Analysis of Dynamic
26Beddaes, T. S.. "Onset of Leading Edge Separation Effects Under Stall using Unstcady Boundary Layer Theory." NASA CR-2426, Oct
Dynamic Conditions and Low Mach Numbcr," Pmccerlirtxs of !he 3411, .,. ..
1974
A~znenlFo,s,,t of rl,e Ajner-ice,, Helicoprer-Soriev, 1978. "Liiva, I., Davenport. F. I . , Grey, L.. Walton. I. C.. "Two-Dimcn-
"Beddoes, T. S., "Representation of Airfoil Behavior. AGARD Spe- sional Tests of Airfoils Oscillating Near Slall," USAAVLABS TR-68-13.
cialists Meeting on the Prediction of Aerodynamic Loads on Rotorcraft," 1968.
AGARD CP-334. 1982. Also Verlica. Val. 7. (2). 1983. 34McCraskey. W. I . , McAlisler. K. W.. Carr. L. W.. Pucci. S. K..
"McCroskev. W. I.. McAlister. K. W... Carr.. L. W.. Pucci..S. I . "An
~ ~ ~~~ Lambcn. 0.. Indergand. R. F., "Dynamic Stall an Advanccd Airfoil
~xpcrimental'~1udyof ~ y n a m i cStall on Advanccd ' ~ i r f o i lSections;' Sections." P,arrrdi,zjis of the 36118 A~lvlrrolFonsn of llze A,n$l.irorr He/-
Vols. 1, 2 & 3 NASA TM-84245, 1982. icoplo-Socirv. Washington DC, May 1980.
"Wilby, P. G . , "An Experimental Investigation an the lnflucnce of a

You might also like