Você está na página 1de 12

State-Space Model for Unsteady Airfoil Behavior and Dynamic Stall

J. G o r d o n Leishman *
Gilbert L. Crouse, J r . t

C e n t e r for Rotorcraft Education a n d Research, D e p a r t m e n t of Aerospace Engineering, University of Maryland, College P a r k , M D 20742

Abstract
A t i m e - d o m a i n m o d e l h a s been f o r m u l a t e d t o r e p r e sent, a t an e n g i n e e r i n g level of a p p r o x i m a t i o n , t h e u n s t e a d y lift, d r a g and p i t c h i n g m o m e n t c h a r a c t e r i s t i c ~of a t w o - d i m e n s i o n a l airfoil u n d e r g o i n g d y n a m i c stall. The m o d e l is g i v e n as a set o f f i r s t o r d e r differential s t a t e e q u a t i o n s ; (1)an e i g h t s t a t e l i n e a r a t t a c h e d flow s o l u t i o n d e r i v e d f r o m i n d i c i a l r e s p o n s e f u n c t i o n s valid f o r c o m p r e s s i b l e flow, (2) a t h r e e s t a t e s o l u t i o n f o r t h e p r o g r e s s i v e n o n l i n e a r effects o f t r a i l i n g e d g e flow s e p a r a t i o n a n d , (3) a s i n g l e s t a t e s o l u t i o n f o r t h e c a t a s t r o p h i c l e a d i n g e d g e flow s e p a r a t i o n w h i c h i s c h a r a c t e r i s t i c o f d y n a m i c stall. T h e d y n a m i c s o f e a c h p a r t o f t h e m o d e l are c o u p l e d i n s u c h a w a y to a l l o w p r o g r e s s i v e t r a n s i t i o n b e t w e e n t h e airfoil s t a t i c s t a l l a n d t h e d y n a m i c s t a l l c h a r a c t e r i s t i c s . A n i m p o r t a n t f e a t u r e o f t h e m o d e l is t h a t t h e effects o f flow c o m p r e s s i b i l i t y a r e i n c l u d e d a n d a s s u c h t h e m e t h o d is p a r t i c u l a r l y u s e f u l i n t h e p e r f o r m a n c e , a e r o e l a s t i c response, a n d real-time simulation analysis of helicopter r o t o r s . To v a l i d a t e t h e m o d e l , c o r r e l a t i o n s a r e p r e s e n t e d w i t h u n s t e a d y force a n d m o m e n t d a t a f r o m oscillatory p i t c h t e s t s o n NACA 0012, H H - 0 2 a n d S C - 1 0 9 5 airfoils.

Time constant for leading edge pressure response Time constant for vortex lift Time constant for vortex traverse over chord Free-stream velocity State variables Angle of attack Compressibility factor = 4 Indicia1 response function Subscripts (.hc Refers Refers (.)A4 (.)e Refers Refers (.)a Superscripts (dC Refers Refers (.)I Refers (.)p Refers (.)"

to to to to

aerodynamic center pitching moment about the 114-chord pitch rate angle of attack

to to to to

circulatory loading noncirculatory (impulsive) loading potential (attached flow) loading loads with trailing edge separation

Refers t o vortex loading

Nomenclature
Sonic velocity Coefficients of indicial functions Elements of the system state matrix Exponents of indicial functions Elements of the system output matrix Airfoil chord Chordwise force coefficient Pressure drag coefficient Pitching moment about the 1/4chord Zero lift pitching moment Normal force coefficient Maximum normal force coefficient Critical normal force coefficient Normal force curve slope Vortex lift center of pressure Separation point location Mach number Nondimensional pitch rate = c k / V Distance travelled in semi-chords = 2 V t l c Time Noncirculatory time constant = c/a Time constant for separation point movement

Introduction
T h e successful design of advanced rotorcraft requires the ability to confidently predict the large unsteady and vibratory loads generated and transmitted by the rotor system. T h e capability t o accomplish this has improved significantly in recent years as a result of advances in the analytical modeling of blade structural dynamics, the rotor wake geometry and unsteady aerodynamics. While rotor structural dynamic modeling has now reached a good level of maturity, the development of accurate and computationally efficient aerodynamic models which represent the unsteady behavior of the blade sections, still poses a major challenge t o the rotor analyst. Generally the analyst needs t o determine an appropriate compromise between the accuracy of a given aerodynamic model and the need t o keep computational requirements within practical limits. In most circumstances, these requirements are conflicting and the rotor analyst is forced to resort t o a relatively simple representation for the aerodynamics. Unfortunately, this can restrict the range of flight conditions over which the analysis can be applied, and thereby severely limit its generality as a practical design tool. Within a helicopter rotor flowfield, the blades encounter complex time varying changes in aerodynamic angle of attack due t o imposed control inputs, the dynamic motion of the blades and local variations in inflow velocity due the complex three-dimensional vortex wake system created by each blade. Understanding and modeling the effects of this complex time varying flowfield on the unsteady aerodynamic behavior of the blade sections is one of the major challenges still facing rotor analysts. In many rotor operating regimes, unsteady aerodynamic effects are of low magnitude and can be justifiably neglected in any arialysis. However, if the

'Assistant Professor, Member AHS, AIAA. tGraduate Fellow, Member AHS, Student Member AIAA. Presented a s paper 89-1319 at the A I A A / A S M E / A S C E / A H S / A S C 80th Structures, Structural Dynamics and Malerials Conference, Mobile, Alabama, April 3-5, 1989. Copyright 01989 by J.G. Leishman and G.L. Crouse. Published by the American Institute of Aeronautics and Astronautics, Inc. with permission.

angle of attack of the blade sections becomes large enough, dyform has a number of advantages, particularly when applied to namic stall may occur. Typically, this occurs on the retreating aeroelasticity analyses: (1) T h e form of the aerodynamic represenblade under conditions of high blade loading and in high speed fortation is particularly elegant. (2) T h e representation is compatiward flight. In general, the rotor operational limitations, i.e. vible with the structural dynamic equations13 and can be integrated bration, aeroelastic stability, maximum control loads and fatigue simultaneously in time. (3) T h e model is given as a continuous limits are all determined by the onset of transient flow separation time representation and no particular form of discretization or numerical solution technique is imposed. Any standard ODE solver such as dynamic stall. Many experimental t e ~ t s l -have shown that the distinguish~ can be used t o integrate the state equations. (4) Under attached flow regime the airloads are linear functions of the forcing, thus ing feature of dynamic stall compared with static stall is the shedding of significant concentrated vorticity from the airfoil leadingthe state space representation lends itself to eigenanalyses. This edge region. This vortex disturbance is subsequently swept over however, cannot be done under stalled conditions unless some the airfoil chord and induces a strong moving pressure wave on form of linearization is performed. the airfoil surface. These pressure changes result in increases in T h e approach of using differential equations t o describe the airfoil lift and large nose-down pitching moments well in excess of unsteady aerodynamic behavior of a 2-D airfoil was first adopted the static values. For repeated excursions into stall, considerable by Tran ei aL9 who used experimental airfoil d a t a from small hysteresis in the force and moment behavior can also arise. These amplitude pitch oscillations t o estimate the equation parameters. conditions may lead t o reduced or negative pitch damping, and if T h e objective here was t o obtain a linearized model for the unstall is sufficiently severe this may excite the blade torsion mode steady aerodynamics through stall, and thereby provide a model a t the natural frequency, which can lead to a dynamic instabilwhereby blade aeroelastic stability (with stall) could be computed ity known as stall flutter. Thus, to define the rotor operating using Floquet theory. Reasonable success has been demonstrated envelope, i t is necessary t o be able t o predict this dynamic stall with this method, although the quantitative prediction capability phenomenon and model its consequences on the dynamic response for the airloads could probably be improved. Furthermore, this of the rotor. method requires a large d a t a base of unsteady data for a particular T h e unsteady aerodynamic response of an airfoil to a speairfoil section over a wide range of Mach numbers. Experimencific time history of forcing can now be determined with contally, these d a t a are not easy to obtain. For general application siderable detail and accuracy using computational fluid dynamic t o a variety of airfoil sections, it makes more sense to establish a (CFD) methods. For example, numerical solutions t o the unmodel which (a) is based on well proven classical unsteady aerodysteady Navier-Stokes equations are now becoming increasingly namic methods for attached flows and (b) can be used to e s i i m a i e feasible5*' and have shown some recent success in modeling dythe dynamic stall characteristics from the static stall characterisnamic stall. However, these aerodynamic solutions are complex tics. However this requires, a t a minimum, the modeling of key and the required computational resources are, for most rotor analfactors which affect the dynamic stall mechanism and must inyses, prohibitive. Nevertheless, C F D methods are providing conclude both leading edge separation and trailing edge separation siderable insight t o the aerodynamic problems encountered by effects together with leading edge vortex shedding where approrotorcraft7. It appears, however, t h a t for the foreseeable future priate. These effects must be modeled within the practical commore approximate aerodynamic solution methods must continue putational constraints imposed by the aeroelastic or performance t o be used in most rotor performance and aeroelasticity analyses. analysis of the rotor system. Clearly, this poses somewhat of a dilemma for the analyst as the In this paper, it is demonstrated how a relatively elegant and computationally efficient method can be created and used t o credconsequences of complex viscous effects must be modeled within the practical constraints imposed by the computational enormity ibly reproduce the dynamic stall characteristics of an airfoil secof overall rotor performance and/or aeroelasticity analysis. tion given, primarily, the static (nonlinear) airfoil lift and pitching To provide some representatibn of the unsteady ae;odynamic moment characteristics. T h e method draws on the work of both behavior of the blade sections. a number of fairlv so~histicated bed doe^'^^^^^^^ and Leishman and Beddoes", t o establish a syssemi-empirically based models have been developed (e.g. Refs. t e m of first order ODE's which represent specific aspects of the 8-11). Many of these semi-empirical models have reasonable apunsteady airfoil behavior. T h e method is useful in the perforproximations for the unsteady aerodynamics under attached flow mance and aeroelastic response analysis of helicopter rotors and conditions, however for the dynamic stall regime they often rely propellers. To support the development of the method, illustraheavily on the synthesization of wind tunnel data from unsteady tive comparisons with experimental d a t a are presented for the unairfoil tests. In the interests of computational simplicity, many steady lift, pitching moment and drag on three typical helicopter models sacrifice physical realism and so may have limited generrotor airfoils undergoing oscillatory pitch forcing into dynamic ality in application. Despite these limitations, developed methstall. ods have met with good success and have been shown t o give Methodology significant improvements in performance prediction capability for helicopters. However, with the increasing overational demands - . T h e objective is t o derive a concise but comprehensive description t h a t are placed on helicopters and the increasing use of advanced of the unsteady aerodynamic behavior of a 2-D airfoil undergoing blade technology and modern airfoil sections, there is still a fundynamic stall as a finite number of first order differential state damental requirement for improved aerodynamic models that can equations. T h e background t o this form of representation can be be used with greater confidence in rotor design procedures. found in many excellent texts on modern control system theory, I t is t h e purpose of this paper t o document t h e results of some such as Refs. 17 or 18. In general, an nth order differential system recent research directed towards improving the representation of with m inputs and p outputs may be represented by n first order unsteady aerodynamics within a comprehensive rotor analyses, differential equations including the important effects of flow compressibility and dynamic stall. T h e method is presented in the time domain which k=Ax+Bu is a necessary prerequisite t o fully account for the flowfield enwith the output equations countered by a helicopter rotor. The approach used is to express the unsteady airloads as a series of elementary systems which are described by first order ordinary differential state equations (ODE's), i.e. state space form. Preliminary results with this forwhere z = d + / d t ; T h e vector x = z;,i = 1 , 2 , ...,n is a n x 1 mulation were first reported in Refs. 12 and 13. T h e state space column vector called the state-vector; u is a m x 1 input column

. .

vector; y is an output column vector; A , B, C and D are coefficient matrices of appropriate dimensions. The inputs u t o the system are the angle of attack and pitch rate, and the outputs y are the required lift, drag and pitching moment. In general, the states describe the internal behavior of the system and are simply the information required a t a given instant in time t o allow the determination of the future outputs from the system given future inputs. In other words, the state of the system determines its present condition. It should be noted that it is desirable t o obtain a system with a minimum number of states (since each state imposes extra computational overhead) while a t the same time maximizing the performance of the system in terms of aerodynamic prediction capability.

10-

l
(

--.-M

M=O (Wagner) M = 0.2 = 0.8

/../.-.

Attached Flow Behavior


A prerequisite in any unsteady aerodynamic theory is the ability t o accurately represent the unsteady aerodynamic response under attached flow conditions. Classical theories in this category for incompressible flows have been formulated by c he odor sen'^, Greenbergzo, sears2', and von Karman and searsz2. In addition, Wagnerz3 has obtained a solution for the indicial response. T h e indicial response is particularly useful since the state equations representing t h e unsteady airfoil behavior can be obtained by the direct application of Laplace transforms to the indicial response t o get the aerodynamic transfer function. While the well known Wagner indicial function is used in some form in many aeroelastic analyses, strictly speaking it is restricted to incompressible flows. However, most practical aerodynamic problems involve compressibility to some degree and there is no exact equivalent of Wagner's function for a compressible flow. Nevertheless, as shown by various researchers, including M a z e l ~ k i ~ '~ e~ ~ o e s " , ~ o w e 1 1 ~ ~ Leishmanm, a practical ~ dd , and representation of the indicial responses can be made using an exponential series approximation. By suitably generalizing the indicial response functions in terms of Mach number, as shown by Leishman30, the corresponding state equations may be obtained in a relatively general form. In Ref. 30, the indicial functions are assumed t o be idealized into two parts. One part of the response is for the initial noncirculatory loading which comes from piston theoryz4. This result is valid for any Mach number for time equal t o zero. For subsequent time, pressure waves from the airfoil propagate at the local speed of sound and the loading decays rapidly with time from its initial value. In fact, these time dependent noncirculatory loads may be considered the compressible analog of the apparent mass terms used in many incompressible analyses. T h e second part of the indicial response is due t o the circulatory loading which builds up quickly t o the steady state value. The behavior of the indicial lift response is shown in Fig. 1 for a step change in angle of attack a t Mach numbers of 0.2 and 0.8 in comparison with the classical Wagner function. By convention, the functions are plotted versus nondimensional time S = 2 V t l c which corresponds to the relative distance travelled by the airfoil in terms of semi-chords. In general, the indicial normal force and quarter-chord pitching moment response t o a step change in angle of attack a and a step change in pitch rate q can be written as

o
0

.
2

,
4

,
6

,
8

,
10

,
12

,
14

.
S

,
16

Distance travelled In semi-chords,

Figure 1: Indicia1 lift response to a step change in angle of attack where the indicial response functions , , ,: : : ,d , 4 4fM 4 5 q : : 4 , and , : are expressed in terms of both aerodynamic time S and 4 the Mach number M. These functions are fully defined in Ref. 30. T h e superscripts C and I refer t o the components of circulatory and noncirculatory (impulsive) loading respectively and the subscript M refers t o the pitching moment contribution. T h e second term in Eq. 4 represents the contribution t o the pitching moment due t o a Mach number dependent offset of the aerodynamic center, xac, from the airfoil quarter-chord axis and must be obtained from either experiments or C F D codes. At subsonic speeds, the aerodynamic center lies close t o t h e 114-chord although for transonic speeds the effective aerodynamic center moves quickly t o the vicinity of the mid-chord as the free-stream Mach number approaches unity. To illustrate the form of the state equations, consider the normal force response t o changes in angle of attack. T h e circulatory indicial response function is written as

and the noncirculatory function as

R o m ~ e d d o e s ' ~h e constants of the circulatory lift function are t, A1 = 0.3, A2 = 0.7, bl = 0.14 and b2 = 0.53 and the function is generalized t o different Mach numbers by scaling the exponents by P2. From Leishmanm, the noncirculatory time constant, T, = Z(,TI is given based on an approximation to the exact linear theory results of ~ o r n a x ~ ' where TI = c l a and

T h e circulatory normal force response t o a variation in angle of attack can be written in terms of the differential state equations

where x = d x l d t . For the circulatory component, the pitch term q can be coupled into the above equations by using the angle of attack a t the 3/4-chord, i.e.

T h e output equation for the normal force coefficient is given by

where 2 x / P is t h e force-curve-slope for linearized compressible flow. Similarly, the noncirculatory normal force due t o angle of attack can b e written a s t h e differential s t a t e equation
13

= -13 I(, Tz

+ a ( t ) = a3323 + a ( t )

(11)

with the o u t p u t equation for t h e normal force coefficient given by

T h e remaining (five) s t a t e equations for t h e pitching moment and pitch r a t e terms can be derived in a similar way, and a r e given in Ref. 25 and 26. T o obtain t h e total airloads under attached flow conditions, t h e individual components of loading are linearly combined t o obtain t h e overall aerodynamic response. For example, t h e total normal force coefficient CL is given by

Figure 2: Force resolution on a thin airfoil in unsteady flow

and a similar equation holds for t h e pitching moment. T h e net unsteady aerodynamic response in attached flow can b e described in terms of a two i n p u t , two o u t p u t system where the inputs are t h e airfoil angle of attack and pitch rate and t h e outputs are t h e unsteady normal force (lift) and pitching moment. I t can be readily shown t h a t by rearranging all (eight) s t a t e equations for attached flow conditions, they can b e represented as a diagonal canonical set of s t a t e equations

T h e corresponding chord force in potential flow Cg is given in terms of a~ a s

which involves t h e appropriate combination of t h e states 2.1 and 12. T h u s , as a by-product of the above system representation for t h e unsteady lift, t h e necessary information may be extracted from t h e system t o obtain t h e unsteady chord force component. Finally, t h e instantaneous pressure drag can be obtained by resolving t h e components of the normal force and chordwise forces through the geometric angle of attack a using C L ( t ) = C&(t) sin a ( t )

where t h e matrices are of t h e form

- Cg(t) cosa(t)

(I61

Further details and validation of the unsteady chord force and pressure drag representation are given by ~ e i s l l m a n ~ ~ .

Nonlinear Aerodynamics and Dynamic Stall


Having established t h e differential s t a t e equations which govern t h e attached flow (linearized) behavior of t h e airfoil, it is required t o extend t h e analysis to encompass the nonlinear airfoil behavior and dynamic stall. T o d o this, it is necessary to identify and model the key features of t h e stall process, rernembering t h a t this must b e done using a minimum number of state equations. It is also required t o attribute a physical significance t o each of t h e s t a t e equations t h a t are defined.

T h e total aerodynamic lift a n d pitching moment response t o a n arbitrary time history of n and q can be obtained from t h e above s t a t e equations by integrating using a n ODE solver (see Ref. 26). Referring t o Fig. 2, t h e unsteady chord force (in-plane force) and pressure drag o n t h e airfoil may also be obtained in terms of t h e s t a t e variables already defined. For a fixed-wing problem t h e wing has a high in-plane stiffness and s o t h e chord force component rarely participates in t h e aeroelastic problem. However, for a helicopter rotor t h e relatively low lead/lag stiffness of t h e blades makes t h e chord force component much more significant for t h e aeroelasticity problem. F t o m t h e o u t p u t equation (Eq. 10) t h e effective angle of attack of t h e airfoil, a ~due t o t h e shed wake (circulatory) terms , can be written in t e r m s of t h e states rl and 1 2 as

Stall Onset
T h e most crucial aspect of t h e modeling of dynamic stall is t h e identification a n d representation of t h e conditions for the onset of leading edge separation. T h e implementation of any criterion must also be sufficiently general t o allow for the prior history of t h e airfoil motion, including angle of attack and Mach number variations. Evans and Mort33 present a useful criterion equivalent to a critical leading edge pressure and associated pressure gradient which m a y be used t o denote the onset of static leading-edge stall. T h i s criterion was subsequently evaluated by 13eddoes14 within t h e context of rotor airfoil performance, under both steady and unsteady conditions. For practical purposes, Beddoes determined t h a t although under time-dependent forcing conditions the

Trailing Edge Separation

NACA 0012, woods*


C
0

.4

...8.NACA 0012. ~ c ~ r o s k r y "

s
]L

.1

.2

.3

.4

.5

.6

. 7

.8

.9

Mach number, M
Figure 3: Stall boundary for the NACA 0012 airfoil section pressure gradient on the airfoil a t a given angle of attack was significantly modified, it was possible 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 ~ e d d o e s 'to encompass higher Mach number flows, ~ where the attainment of a critical leading edge pressure was again used to denote the onset of shock induced stall. In application the leading edge pressures P are related to the normal force CN, so it is possible t o obviate the need t o compute airfoil pressures by transforming the calculation to the C N domain. From an analysis of airfoil static test data, a critical value of CN(static)= CN, may be obtained which corresponds t o the critical pressure for separation onset a t the appropriate Mach number. In practice CNl is close to the maximum static C N . Thus a Mach number dependent separation onset (stall) boundary may be defined. A typical boundary for the NACA 0012 airfoil is shown in Fig. 3. For unsteady conditions, there is a lag in CN(t) with respect t o the forcing (above), however there is also a lag in the leading edge pressure response P ( t ) with respect t o C N ( t ) Thus for an increasing angle of attack, the lag in the leading edge pressure response results in the critical pressure being achieved a t a value of CN and hence a t a higher angle of attack than the quasi-steady case. Thus, this mechanism significantly contributes to the overall delay in the onset of dynamic stall. To implement the critical pressure criterion under unsteady conditions, a first order lag may be applied to CN(t) t o produce a substitute value C&(t) with the presumption t h a t whatever properties apply t o P ( t ) must also apply t o CEy(t). This representation may be written as the differential state equations

A phenomenon t h a t is also involved in most types of airfoil stall is progressive trailing edge separation. The associated loss of circulation due to trailing edge separation introduces a nonlinear force and moment behavior, especially with the cambered airfoils more typically used on modern helicopters. ~ i l suggests t h a t ~ b ~ ~ trailing edge separation may play a significant role in the onset of dynamic stall. However, as also discussed by Wilby, experimental tests have indicated t h a t the occurrence of trailing edge separation is suppressed by increasing pitch rate. The dynamic stall process may then be initiated by leading-edge separation or shock induced separation if supercritical flow is allowed t o develop. Even so, when the primary source of separation is at the leading edge or the foot of a shock wave it appears t h a t this is generally sufficient t o promote some separation a t the trailing edge and hence initiate some nonlinear behavior in the force and moment response. One theory which models separated flow regions on 2-D bodies is attributed t o Kirchhoff and is reviewed in Refs. 35 and 36. A specific case of Kirchhoff flow is a simple model for the trailing edge separation phenomenon (Fig. 4) in which the airfoil normal force coefficient, C N , may be approximated as

where 2 a is the force-curve-slope for incompressible flow, f is the trailing edge separation point and a is the angle of attack. Thus, if the separation point can be determined it is a trivial calculation t o determine the normal force. In practical cases, this expression may be extended t o encompass compressible flows where 27r is replaced by the force-curve-slope at the appropriate Mach number

As first shown by ~ e d d o e s ' ~ , order t o implement this proin cedure the relationship between the effective separation point, f , and the angle of attack, a , can be deduced from the airfoil static lift behavior by rearranging Eq. 20 t o solve directly for f . T h e relationship between a and f can be generalized empirically in a fairly simple manner using the relations

1 - 0.3exp { ( a - al)/S1) 0.04 - 0.66exp {(a1 - a)/S2)

if a if ff

< ffl > a1

(21)

T h e coefficients Sl and Sz define the stall characteristic, while a 1 defines the break ~ o i n t corresponding to f = 0.7. I t should be t noted t h a t f x 0.7 closely co~responds o the static stall angle for most airfoil sections. S l , S2 and a1 are determined empirically for each airfoil and vary with Mach number. Using the above equations, the reconstructed lift versus angle of attack relationships are shown for the NACA 0012, HH-02 and SC-1095 airfoils a t a Mach number of 0.3 in Fig. 5. This procedure has also been thoroughly validated for higher Mach numbers and other

where the input to the above equation is the total unsteady lift under attached flow conditions, C;. The time constant T p is a function of Mach number and can be determined empirically from unsteady airfoil data. T h e value of T p is largely independent of airfoil shape but is more dependent on Mach number1'. Thus by monitoring the value of C h ( t ) the onset of leading edge/shock induced separation under dynamic conditions will be initiated when C h ( t ) exceeds the critical CNl ( M ) boundary. Furthermore, if the value of C h ( t ) is monitored throughout the calculation into stall, then it may be used as an indicator for the conditions which permit flow reattachment, i.e. if C h ( t ) < Cjv1

Figure 4: Kirchhoff model for separated flow past a flat plate

A
O 0

-Reconstruction
I I I

NACA 001 2 HH-02 SC-1095

A
1

-10

10

20

30 1

-Reconstruction
I

O 0

NACA 001 2 HH-02 SC-1095

Angle of attack, a (deg.)


1 1

Figure 5: Reconstruction of the static lift behavior using t h e Kirchhoff model airfoils, and may be applied t o almost any airfoil if t h e static stall characteristic is known apriori. A general expression for the pitching moment behavior cannot b e obtained from Kirchhoff theory and an alternative empirical relation must be formulated. From t h e airfoil static d a t a , t h e center of pressure a t any angle of attack may b e determined from (allowing for t h e zero lift moment CM,,). T h e t h e ratio C M / C ~ variation can be plotted versus t h e corresponding value of t h e separation point and fitted in a least squares sense t o the form

10

20

30

Angle of attack, a (deg.)


Figure 6: Reconstruction of t h e static ri~omentbehavior tribution and t h e boundary layer response. A relatively simple open loop procedure was first developed by ~ e d d o e s " and later by Leishman and Beddoes" t o represent t h e variation of the timedependent location of t h e trailing edge separation point. This procedure can be used a s a means of extending t h e evaluation of nonlinear forces and moments into t h e dynamic regime via t h e application of t h e Kirchhoff theory, as above. T h e procedure is performed by firstly incorporating the airfoil unsteady pressure response via Eqs. 17 and 18 which may then b e used t o define an effective angle of attack af which gives t h e s a m e unsteady leading edge pressure a s for t h e quasi-steady case

where ICo = (0.25-X,,) is t h e aerodynamic center offset from t h e 114-chord. T h e constant K1 gives the direct effect on the center of pressure due t o t h e growth of the separated flow region and the constant helps describe t h e shape of the moment break a t stall. T h e values of 1 6 , K l , and m can be adjusted for different airfoils, a s necessary, t o give t h e best moment reconstruction. Typically, m = 2. Static moment reconstructions for the NACA 0012, HH-02 and SC-1095 airfoils are shown in Fig. 6. An expression for t h e chord force Cc may also be deduced from t h e Kirchhoff solution for the trailing edge stall problem

T h i s value of af may be used t o determine a value for t h e effective separation point f a t this af from t h e static f versus a ' relationship in Eq. 21. Secondly, t h e additional effects of t h e unsteady boundary layer response may be represented by applying a first order lag to t h e value of f t o produce t h e final value for t h e unsteady trailing ' edge separation point f N . In s t a t e variable form this additional lag in t h e response may be represented as

Due t o viscous effects on t h e pressure distribution, t h e airfoil does not realize 100% of t h e chord force which would be attained in potential flow. Allowance for this nonrealization is made through t h e factor q which can obtained empirically from static airfoil test d a t a . Typically q = 0.95. Note t h a t for inviscid flow, q = 1 a n d f = 1 s o for steady conditions CL = 0. However for unsteady conditions CL # 0 as t h e angle of attack in Eqs. 20 and 23 must be replaced by t h e effective angle of attack a~ (i.e. Eq. 14). It should be noted t h a t under unsteady conditions t h e instantaneous pressure drag can even become negative. Further details on modeling t h e unsteady chord force and pressure drag are given by Leishman in Refs. 11 and 30. For unsteady flow there will exist a modified separation point location d u e t o t h e temporal effects on t h e airfoil pressure dis-

As in t h e case of T p , t h e time constant Tf is Mach number dependent. Substantiation for t h e modeling of the above two components which contribute t o the overall unsteady trailing edge separation response have been given previously in Refs. I 1 and with t h e the modi15. Finally, t h e (nonlinear) normal force fied (unsteady) trailing edge separation point f" is given by

CL

and the pitching moment by

CL = [KO+ K l ( l - f") + Kz ~ i n ( ? r ( f " ) ~ )C$ + C M ~ (28) ]


where C$ is t h e circulatory normal force coefficient and CMois t h e zer+lift moment. T h e contributions of the other unsteady circulatory and noncirculatory moment terms are additive t o E q . 28. Similarly, t h e chord force is written as

Hence, t h e corresponding pitching moment C b produced by the vortex lift component will b e given by

Both t h e vortex decay time constant T,, and t h e nondimensional time Turhave been determined statistically from a variety of dynamic stall test data" and appear t o b e relatively independent of both Mach number and airfoil shape.

Modifications to the Model Modeling of Dynamic Stall


T h e general case of dynamic stall involves t h e formation of a vortex near t h e leading edge of t h e airfoil which subsequently separates from t h e surface and is transported downstream. After the vortex detaches, t h e induced airloads appear t o be qualitatively similar for different airfoils, for different Mach numbers and for different modes of forcing such a s oscillatory pitch and ramp motions34. An approximate, but physically acceptable, model for t h e dynamic stall process has been formulated by viewing t h e vortex lift contribution C& as an excess circulation which is not shed into t h e airfoil wake until some critical condition is reached. T h e critical condition used here is of course when C h ( t ) exceeds CN,. At this point, catastrophic flow separation occurs and t h e accumulated circulation passes over the airfoil and into t h e wake. T h e vortex lift process can be represented by t h e differential s t a t e equations Although t h e above system equations describe, in a n open loop sense, t h e basic physical flow phenomena likely t o be encountered, t h e elements of t h e model are physically coupled. For example, trailing edge separation development will interact t o some extent with t h e onset of leading edge separation and the subsequent vortex lift generation. In addition, under general forcing conditions, separation effects m a y b e compounded by a t h e airfoil kinematics. As shown in Ref. 1 1 , in most cases interactional effects can be readily represented by modifying t h e appropriate time constant associated with t h e behavior, i.e. by reducing or increasing the time constant associated with t h e process. T h e various strategies a r e documented in Ref. 11 and are not repeated here. All the modifications a r e incorporated into t h e algorithm using simple logic a n d t h e values of t h e time constants are updated during each pass through t h e algorithm. O t h e r modifications have been made t o t h e model based on experience with t h e discrete time version presented previously11. These included modifications t o t h e chord force calculation under deep stall conditions and the modeling of secondary vortex shedding phenomena. In addition, t h e mean center of pressure during flow reattachment from t h e deep stall regime was represented by using Eq. 28 with a different effective separation point t h a n for t h e lift. T h i s parameter, f M , is computed by using t h e quasi-steady separation point location f,, (for the same angle of attack, a j ) as an input t o t h e system

where

C$ [ c u = { Ol - ( 1

+fl)2/4]

for for

TU

T ,

5 > 2Tui 2Tu/

(32)

as derived from t h e Kirchhoff approximation. T h e vortex lift is essentially generated by the time rate of change of circulation d I ' / d t K C, which also dissipates with a characteristic time cons t a n t T,. Consequently, when the rate of change of lift is low t h e vortex lift is being dissipated a s fast as it accumulates and in t h e limit a s t h e flowfield becomes steady, t h e airfoil characteristic will revert smoothly back t o the static nonlinear behavior. Abrupt airloading changes occur when the critical conditions for leading edge o r shock induced separation effects are m e t , i.e. C h ( t ) exceeds CN,. At this point there is a catastrophic loss of leading edge suction and t h e accumulated vortex lift is assumed t o s t a r t t o convect over t h e airfoil chord. When this occurs, t h e nondimensional time r, = 0 is defined t o track t h e vortex passage. T h e rate a t which this vortex convection process occurs has been shown from a variety of experimental tests t o be somewhat less t h a n half of t h e free-stream velocity, t h e actual rate which is also somewhat dependent on Mach number. During t h e vortex convection process, the vortex lift is assumed to continue t o accumulate via Eqs. 30-32 but the accumulation is terminated when t h e vortex reaches t h e airfoil trailing edge, i.e. after a suitable nondimensional time interval Tur. Assuming t h e airfoil maintains a high angle of attack, t h e airloads quickly approach their quasi-steady values, even though the angle of attack may still be changing. T h e center of pressure on t h e airfoil also varies with the chordwise position of t h e vortex and will obtain a maximum value when the vortex reaches the trailing edge (r, = TUl). A fairly general representation of t h e center of pressure behavior (aft of 114-chord) can be approximated a s

T h i s requires t h e addition of a final s t a t e , namely xlz, however t h e overall improvement in t h e correlation obtained with test d a t a justifies t h e inclusion of this e x t r a s t a t e .

Total Aerodynamic Response


T h u s , four additional states z9,zlo,x11,x12 are required t o represent t h e nonlinear aerodynamics. Essentially, this part of t h e model is a series of simple first order systems t h a t are connected by nonlinear gains. (This can be more easily seen if the t h e systems a r e written o u t a s a simulation block diagram). T h e input t o t h e first part of t h e nonlinear model is t h e unsteady potential lift Ch, with t h e other system inputs being derived from t h e o u t p u t of t h e previous system. By suitable manipulation of t h e o u t p u t s from t h e various subsystems, the required total loadings can be obtained. For example, the total normal force coefficient C N under dynamic stall conditions is given by

with similar equations for t h e itching moment and drag force components.

Results and Discussion


T o evaluate t h e model, a computer program was developed using F O R T R A N and implemented on a MicroVAX I1 computer. T h e

CP, = 0.25 1 - cos

(8

NACA 0 0 1 2

SIKORSKY SC-1095 HUGHES H H - 0 2

Figure 7: Geometry of the airfoils considered in this study program was used t o study a variety of examples of the unsteady airloads on airfoils subject t o prescribed forcing below stall, and the results were correlated with experimental d a t a . For these particular calculations, the integration of the state equations was performed using the O D E solver D E / S T E P given in Ref. 37, which is a general purpose Adams-Bashforth O D E solver with variable step size and variable order. Further discussion of the performance of this O D E solver is given in Ref. 38. There are many good examples of unsteady airfoil behavior available in the published literature which can be used t o illustrate the performance of the theory. However in the interests of brevity, representative examples of oscillatory pitch airfoil motion under light and deep dynamic stall conditions will be considered. Validation of the modeling under attached conditions has been previously considered in Refs. 25 and 26. To avoid difficulties in comparing test data from different wind tunnels, all the experimental d a t a are taken from the tests perThe selected set of data are for formed by McCroskey e t harmonic pitch oscillations a t various mean angles of attack with a constant oscillation amplitude of 10' a t a (nominal) reduced

frequency of 0.1. This reduced frequency is representative for the once-per-rev cyclic pitch change on a helicopter rotor. The Mach number in all the tests was 0.3. All the experimental data presented are the ensemble average of some 50 pitch cycles. The input was supplied t o the model was obtained by performing an F F T on the the time history of the angle of attack forcing as measured in the experiment. T h e first three harmonics were then used t o reconstruct the forcing. Three airfoils were selected from Ref. 41 for this study; the NACA 0012 (as a baseline section), the HH-02 and the SC-1095. T h e latter two airfoils are typical modern helicopter rotor sections. T h e NACA 0012 is a 12% thick symmetric airfoil, whereas the HH-02 and SC-1095 are cambered airfoils with approximately 9.5% thickness t o chord ratios. In addition, the HH-02 is considerably more cambered than the SC-1095 although has the distinction of a large trailing edge tab which cancels most of the pitching moment associated with the camber. The geometries of the three airfoils are shown in Fig. 7. It should be mentioned, t h a t for the purposes of the comparisons with test d a t a the lift curve slope 2 x l P was replaced by the appropriate quasi-static value obtained from the experimental data, as appropriate. Similarly, the aerodynamic center z,, was obtained from the quasi-static test d a t a and was implemented via Eq. 28. T h e numerical values of the parameters used in the model for each airfoil section are given in Table 1.

Stall Onset, a ( t )= 5'

+ 10' sin wt

Fig.8 showns typical normal force, pitching moment and drag predictions compared with test d a t a for stall onset conditions, where the angle of attack is just enough to initiate leading edge separation. Under fully attached flow conditions, nominally elliptical lift

NACA 001 2

SC- 1095

-5

10

15 -5

10

15

Angle of attack, a

(deg.)

Figure 8: Comparison of theory with test data for the unsteady normal force, pitching moment and pressure drag a t stall onset

1379

;MaNACA 001 2 SC- 1095


L
Model

****

** .**

- - - Stat~cdata

NACA 00 1 2

SC- 1095

-Model

'.

Angle of attack, a

(deg.)

Figure 9: Comparison of theory with test data for the unsteady normal force, pitching moment and pressure drag during moderately strong dynamic stall and moment curves are obtained and this is consistent with predictions from linear theory. With the initiation of some Row separation however, the loops become distorted near the maximum angle of attack. In all cases, all three airfoils exhibit an increase in C N ~ , ,over the static values due to the lag in the development of flow separation under unsteady conditions. I t is clear however, t h a t the HH-02 and SC-1095 maintain attached flow to higher angles of attack with higher values of CN and thereby exhibit a superior performance t o the the NACA 0012. O u t of all the likely cases of dynamic stall, the stall onset condition was found t o be the most difficult t o model. For all three airfoils, the critical CN, value was just exceeded and so these stall onset examples were a good test for the algorithm. It can be seen from Fig. 8 t h a t the model does quite well in predicting the onset of stall for all three airfoils. The point of stall onset is perhaps seen most clearly in the drag characteristics. T h e NACA 0012 exhibits a somewhat greater maximum drag, indicating t h a t it has lost more leading edge suction and penetrated slightly deeper into stall. Although stall onset was quite well predicted, the subsequent behavior during the downstroke of the motion was considerably more difficult t o predict, especially for the pitching moment. Previous studies" have also shown some difficulties with stall onset conditions so difficulties of this nature were not entirely unexpected. It should also be noted t h a t considerable cycle-tecycle variability of the experimental airloads are obtained under these conditions. Overall, the correlation with the test d a t a was quite acceptable bearing in mind that the stall onset condition is where most aerodynamic models are likely to have difficulties.

Moderate Dynamic Stall, a ( t ) = 10'

+ 10sinwt

Fig. 9 shows force and moment predictions in comparison with test d a t a for a case of moderately strong dynamic stall. Under these conditions leading edge vortex shedding is initiated and the characteristic lift overshoot and strong nose down pitching mcment behavior are exhibited. Considerable hysteresis in the force and moment behavior is resent. All three airfoils exhibit a aualitatively similar type of behavior, although there are certainly some quantitative differences. The NACA 0012 exhibits moment stall (points M) a t a lower angle of attack to either the HH-02 or the SC-1095, although the lift stall (points L) occurs a t approximately the same angle of attack for all three airfoils. T h e moment stall occurs just after onset of leading edge separation.

2.5

1.5 CN

NACA

001 2

HH-02

:
\

/'-

Model Experlmenl

---

Static data

NACA

001 2

HH-02

SC-

1095

Model Experlrnent

Angle of

attack, a

(deg.)

Figure 10: Comparison of theory with test d a t a for the unsteady normal force, pitching moment and pressure drag during deep dynamic stall Both the NACA 0012 and the SC-1095 airfoils exhibit a well rounded moment break at the onset of dynamic stall, in comparison with,the HH-02 which has a very abrupt moment break. This suggests t h a t some trailing edge separation is present on the NACA 0012 and SC-1095 prior t o the onset of leading edge separation. This is consistent with the static airfoil chacteristics, since both the SC-1095 and NACA 0012 exhibit fairly gradual static stall characteristics which is symptomatic of trailing edge separation. This suggests t h a t t o some extent the static stall behavior is carried over into the dynamic stall regime. It was interesting t h a t both the HH-02 and SC-1095 exhibited than the NACA 0012. Again, the static a slightly greater CNMA, CNMA, gains are carried over somewhat into the dynamic regime. for T h e gains in CNMA, the HH-02 and SC-1095 are also reflected in the drag curves where higher peak values of CD are obtained as a consequence. On the other hand, the NACA 0012 clearly exhibits a lower peak value of CM than the other two airfoils. This suggests t h a t the strength of the shed leading edge vortex may be somewhat less for the NACA 0012. Overall, the model performed particularly well in predicting the stall onset and the subsequent magnitude and phasing of the induced airloads. T h e peak values of the lift and minimum pitching moment coefficients were well predicted for all three airfoil sections. I t was clear that the slight differences in the onset of dynamic stall between the NACA 0012 and the other two airfoils could be predicted with the model.

Deep Dynamic Stall, cr(t) = 15"

+ 10' sinwt

Fig. 10 shows force and moment predictions in comparison with test d a t a for a cases of deep dynamic stall. Under these conditions the airfoil reaches a high maximum angle of attack. Strong leading edge vortex shedding occurs giving significant increments in normal force, pitching moment and drag. As in the previous case, all three airfoils exhibit a qualitatively similar type of behavior. Both the HH-02 and SC-1095 exhibit increased values of CNMAX

NACA 0012 0.113

Table 1: Parameters used for each airfoil over the NACA 0012. It is interesting, however, that while under static conditions the SC-1095 exhibits a gain in C N M A X about of 0.1 over the HH-02, under dynamic conditions there is almost no between these two airfoils. Similarly, the difference in CNMAx minimum values of C..- are almost identical. This indicates t h a t M while CNMAx be a useful measure of the airfoils performance may under static conditions, this is not necessarily an indication of the performance under unsteady conditions. The aerodynamic model did reasonably well in predicting the magnitude and phasing of the airloads for this deep stall condition. I t should be noted t h a t for all three airfoils there is evidence of secondary vortex shedding near the maximum angles of attack. This manifests itself as smaller secondary peaks in the normal force, pitching moment and drag behavior. This aspect of the behavior is also captured reasonably well with the model, being based on an effective Strouhal number for bluff body vortex shedding". As in the previous cases, most of the differences between the model and the test d a t a are apparent during the reattachment phase. In this regime, there are generally significant variations in the airloads from cycle t o cycle due t o the inherent randomness of the flow. In addition, it appears t h a t there may be some influence of the tunnel and/or test configuration on the airloads in this flow regime. T h e unsteady lift response of the NACA 0012 has also been compared with results from three different test f a ~ i l i t i e s ~ ~ ,appears~ . It ~ ~ , ~ during the onset of separation and during dynamic stall the airloads are comparable, however there appears significant variations between the measured airloads during reattachment. It should also be noted here that recent calculations of dynamic stall using the Navier-Stokes equations6 also show significant deviations from the test d a t a in the flow reattachment regime.

porary rotor designs. Based on the novel work of Beddoesl4>l5, individual flow features are represented in a sufficiently simple manner that permits inclusion within the overall sectional aer+ dynamics calculation. T h e nonlinear effects on the airloads due t o trailing edge separation have been implemented using the Kirchhoff flow theory as a means of relating the force and moment characteristics to the location and progression of the trailing edge separation point. Features of leading edge or shock induced separation have been reviewed and are implemented in terms of a representation of the unsteady leading edge pressure response in which the attainment of a critical value is used to denote the initiation of dynamic stall. Finally, the induced force and pitching moment behavior during dynamic stall have been represented in a physically realistic manner. All the above phenomena are modeled as a series of differential equations t h a t can be implemented as a subroutine for the blade sectional aerodynamics. Validation of the model has been conducted with 2-D test d a t a for three rotor airfoils, namely the NACA 0012, HH-02 and SC-1095. Correlation with the test d a t a was good, particularly in terms of predicting the onset of dynamic stall. T h e phasing of the dynamic stall induced airloads was also modeled quite well. It is considered t h a t the model is sufficiently general to allow its application to other airfoil sections, a t least when engineering levels of prediction capability are required. The level of correlation obtained with the aerodynamic model provides considerable confidence when applied in the design of new rotors.

Acknowledgements
This work was supported by the U.S. Army Research Office under the Center of Excellence for Rotary Wing Technology Program a t the University of Maryland. Drs. Robert Singleton and Tom Doligalski were the technical monitors. The authors wish t o thank Prof. Roberto Celi and Rotorcraft Fellow Khanh Nguyen of the University of Maryland for their stimulating discussions on the state-space formulation. T h e authors also wish t o thank Prof. Inderjit Chopra of the University of Maryland for his continued encouragement during the course of this research.

References
[I] Martin, J.M., Empey, R.W., McCroskey, W . J . , Caradonna, F.X. "An Experimental Analysis of Dynamic Stall on an Oscillating Airfoil," Journal of the American Helicopter Soc., Vol. 19., No. 1, 1974, pp. 26-32. [2] McCroskey, W.J., McAlister, K.W., Carr, L.W. "Dynamic Stall E x p e r i m e ~ t son Oscillating Airfoils," AIAA Journal, Vol. 14, 1976, pp. 57-63.

[3] Carr, L.W., McAlister, K.W., McCroskey, W . J . , "Analysis


of the Development of Dynamic Stall Based on Oscillating Airfoil Experiments," NASA T N D-8382, 1977. [4] McAlister,K.W ., Carr, L.W ., "Water Tunnel Visualization of Dynamic Stall," Nonsteady Fluid Dynamics, Proceedings of the Annual Winter Meeting, ASME, San Francisco, CA, 1978, pp. 103-110. [5] Janakiram, R.D., Sankar, L.N., Principles based Computational craft Applications," Proceedings Conference on Rotorcraft Basic Maryland, February 1988. "Emerging Role of FirstAerodynamics for Rotorof the 2nd. International Research, College Park,

Summary and Conclusions


Rotor aeroelasticity and performance analyses require versatile and relatively simple methods for evaluating the unsteady aerodynamic behavior of the blade sections. T h e main objective behind the work outlined in this paper has been to provide an improved model for the unsteady force and moment characteristics of an airfoil undergoing dynamic stall, but in a sufficiently simple manner and in a computational form t h a t can be included within a comprehensive analysis of the rotor system. To this end, a state space formulation has been selected as particularly appropriate. T h e main emphasis has been on the development of a fairly general model for the effects of dynamic stall t h a t can be applied t o a variety of conventional and advanced airfoils used for contem-

[6] Tuncer, I , Wu, J . , Wang, C., "Theoretical and Numerical


Studies of Oscillating Airfoils," Paper 89-0021, Presented a t t h e 27th. Aerospace Sciences Meeting, Reno, Nevada, J a n 9-12. 1989.

Srinivasan, G.R., McCroskey, W.J., "Navier-Stokes Calculations of Hovering Rotor Flowfields," AIAA Journal of Aircraft, Vol. 25, No. 10, 1988, pp. 865-874. Beddoes, T.S., "A Synthesis of Unsteady Aerodynamic Effects including Stall IIysteresis," Vertica, Vol. 1, pp. 113-123, 1976. T r a n , C.T., Petot, D., "Semi-empirical Model for the Dynamic Stall of Airfoils in view of the Application t o the Calculation of t h e Responses of a Helicopter Blade in Forward Flight," Vertica, Vol. 5, No. 1 , 1981, pp. 35-53. Gangwani, S.T., "Synthesized Airfoil D a t a Method for Prediction of Dynamic Stall and Unsteady Airloads," Verlica, Vol. 8, 1984, p p . 93-118. Leishman, J.G., Beddoes, T.S., "A Generalized Model for Airfoil Unsteady Aerodynamic Behavior and Dynamic Stall using t h e Indicial Method," Proceedings of the 42nd. Annual Forum of the American Helicopter Society, Washington D C , J u n e 1986. Elliott, A.S., Leishman, J.G., Chopra, I., "Rotor Aeromechanical Analysis using a Nonlinear Unsteady Aerodynamic Model," Proceedings of the 44th. Annual Forum of the A m e r ican Helicopter Society, Washington D C , 1988. Elliott, A S . "Calculation of t h e Steadily Periodic and G u s t Responses of a Hingeless Rotor Helicopter using 2-D T i m e Domain Unsteady Aerodynamics," P h . D . Thesis, Departm e n t of Aerospace Engineering, University of Maryland, September 1987. Beddoes, T . S . , "Onset of Leading Edge Separation Effects under Dynamic Conditions and Low Mach Number," Proceedings of the 34th Annual Forum of the American Helicopter Society, Washington D.C., May 1978. Beddoes, T . S . , "Representation of Airfoil Behavior," Veriica, Vol. 7, No. 2, p p . 183-197, 1983. Beddoes, T.S., "Practical Computation of Unsteady Lift," Vertica, Vol. 8 , No. 1, 1984, p p . 55-71. Ziemer, R.E., T r a n t e r , W.H., Fannin,R.D., Signals and Syst e m s - Coniinuous and Discrete, MacMillan Publishing Co., New York, 1983. Kailath,T. Linear Systems, Prentice-Hall, Englewood Cliffs, N J , 1980. Theodorsen, T . , L'GeneralTheory of Aerodynamic Instability and t h e Mechanism of Flutter," NACA Report 496, 1935. Greenberg, J . M . , "Airfoil in Sinusoidal Motion in a Pulsating Stream," NACA T N 1326, 1977. Sears, W . R , "Some Aspects of Non-Stationary Airfoil Theory and its Practical Application," Journal of the Aeronautical Sciences, Vol. 8, No. 3, J a n . 1941, p p . 1 0 4 1 0 8 . [22] von K a r m a n , T h . , Sears, W . R . , "Airfoil Theory for NonUniform Motions," Journal o f t h e Aeronautical Sciences, Vol. 5 , No. 10, Aug. 1938, p p 379-390. [23] Wagner, H., "Uber die Entstehung des Dynamischen Auftriesbes von Tragflugeln," Zeitschrifl fur Angewandte Malhematik und Mechanik, Vol. 5, No. 1, Feb. 1925. [24] Bisplinghoff, R.L., Ashley, H., Halfman, R.L., Aeroelasticity, Addison-Wesley Publishers, 1955.

[25] Leishman, J . G . , Nguyen, K.Q. "State-Space Model for Unsteady Airfoil Behavior," T o appear in t h e A I A A Journal. Available o n request from the authors. [26] Leishman, J . G . , Crouse, G.L., "A State-Space Model of Unsteady Aerodynamics for Flutter Analysis in a Compressible Flow," Paper 89-0022, Presented a t t h e 27th. Aerospace Sciences Meeting, Reno, Nevada, J a n . 9-12, 1989. [27] Mazelski, B., "Numerical Determination of Indicial Lift of a TwwDimensional Sinking Airfoil a t Subsonic Mach Numbers f r o m Oscillatory Lift Coefficients with Calculations for a Mach Number of 0.7," NACA T N 2562, 1951. [28] Mazelski, B., Drischler, J . A . , "Numerical Determination of Indicial Lift and Moment Functions of a Two-Dimensional Sinking and Pitching Airfoil a t Mach Numbers of 0.5 and 0.6," NACA T N 2739, 1952. [29] Dowell, E.II., "A Simple Method for Converting Frequency Domain Aerodynamics into the T i m e Domain," NASA Technical Memorandum 81841, 1980. [30] Leishman, J . G . , "Validation of Approximate Indicial Aercdynamic Functions for TwwDimensional Subsonic Flow," A I A A Journal of Aircraft, Vol. 25, No. 10, O c t . 1988, pp. 914-922. [31] Lomax, H . , Heaslet, M . A . , Fuller, F . B . , Sluder, L., "Two and T h r e e Dimensional Unsteady Lift Problems in High Speed Flight," NACA Report 1077, 1952. [32] Leishman, J . G . , "An Analytic Model for Unsteady Drag," A I A A Journal of Aircrafi, Vol. 25, No. 7 , July 1988, p p . 665-666. [33] Evans, W . T . , Mort, K . W . , "Analysis of Computed Flow Parameters for a Set of Sudden Stalls in Low-Speed T w w Dimensional Flow," NASA TND-85, 1959. [34] Wilby, P.G., "An Experimental Investigation on the Influence of a Range of Airfoil Design Features on Dynamic Stall Onset," Paper No. 2, 10th European Rotorcraft Forum, T h e Hague, T h e Netherlands, August 28-31, 1984. [35] Thwaites, B, (ed.) Incompressible Aerodynamics, Oxford University Clarendon Press, 1960. [36] Woods, L.C., T h e Theory of Subsonic Plane Flow, Cambridge University Press, 1961. [37] Shampine, L.F., Gordon, M . K . , Computer Solution of Ordinary Differeniial Equatzons - The Initial Value Problem, W . H . Freeman and Co., San Francisco, 1975. [38] Shampine, L.F., Watts, II.A., Davenport, S.M., "Solving Nonstiff Ordinary Differential Equations - T h e S t a t e of the Art," SIAM Review, Vol. 18, No. 3 , July 1976, pp. 376-411. [39] McCroskey, W . J . , AZcAlister, K . W . , C a r r , L . W . , Pucci, S.L. "An Experimental S t u d y of Dynamic Stall on Advanced Airfoil Sections," Vols. 1,2 and 3. NASA TM-84245, 1982. [40] Woods, Mrs. M.E. "Results from Oscillatory Pitch and R a m p Tests on t h e NACA 0012 Blade Section," Memo 220, Aircraft Research Association, Redford, U.K., 1979. (411 S t . Hilaire, A . O . , C a r t a , F.O., "Analysis of Unswept and Swept Wing Chordwise Pressure. D a t a from an Oscillating Airfoil Experiment", Vol. 2 - D a t a Report, NASA C R 165927, 1983.

Você também pode gostar