Professional Documents
Culture Documents
Unsteady Aerodynamics
Contents
1 Kinematics interface
1.1 Structural kinematic quantities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.2 Airstation kinematic quantities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1
2
2
3
et al.
. . . .
. . . .
. . . .
. . . .
. . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
4
4
5
6
6
7
rigid airfoil
Corrections for wind tunnel measurements . . . . . . . . . .
Unsteady aerodynamic loads . . . . . . . . . . . . . . . . .
Corrections to for wind tunnel drag measurements . . . . .
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Relationship of Peters formulation to Theodorsens theory
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
8
8
9
9
10
10
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
11
11
12
13
14
14
14
15
15
17
17
19
20
21
22
24
24
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
25
Kinematics interface
An airstation location involves two distinct elements: its geometric position characterized by its position vector,
xa , and its curvilinear coordinate, si , along the composite curve that defines the lifting line. An orthonormal basis
A = (
a1 , a
2 , a
3 ) defines the orientation of the airfoil. At each airstation, an airfoil is defined; it lies in a plane passing
1.1
The first step towards the determination of the kinematic quantities of the airfoil is the evaluation of the kinematic
quantities at point S, the point located on the reference line of the beam at curvilinear coordinate si . Let uk , k =
1, 2, . . . K be the displacements of the nodes of the beam element, see fig. 1, and K the number of nodes of the element.
The shape functions of the element are defined in terms of a local variable, r [1, +1], within the element [1]. Let
r = ri determine the position of point S. The displacement, us , of point S is now interpolated using standard finite
element techniques to find
K
X
hk (ri ) uk (t),
us (t) = u(r = ri , t) =
(1)
k=1
where h (r) are the shape functions of the beam element. The time history of the displacement of point S can be
determined since the above equation can be used at any time t.
Let r k , k = 1, 2, . . . K be the rotation parameter vector of the nodes of the beam element; using a similar
interpolation process, the rotation parameter vector, r s (t), at point S is determined easily. The rotation tensor,
Rs (t), that determines the rotation of the beams cross-section is computed from these rotation parameter vectors.
Within the framework of the finite element discretization of the structure, both nodal displacements, velocities, and
accelerations are evaluated. From the nodal velocities, the linear velocities, v s (t), and angular velocities, s (t), of point
S are obtained from a similar interpolation procedure. Finally, the linear accelerations, v s , and angular accelerations,
s (t), of point S are obtained.
In summary, the structural kinematic quantities determined in this first step are: (1) the inertial displacement,
velocity, and acceleration vectors of point S, denoted us (t), v s (t), and v s (t), respectively, (2) the inertial rotation
tensor, Rs (t), of the beams cross-section at point S, and (3) the inertial angular velocity and accelerations vectors
of the same cross-section, denoted s (t) and s (t), respectively. All these vectors and tensors are resolved in inertial
basis I = (1 , 2 , 3 ).
1.2
Once the structural kinematic quantities of point S and associated beam cross-section have been determined, the
kinematic quantities of the airfoil are derived.
Figure 2 shows the system in the reference and present configurations. In the reference configuration, the inertial
position vector of point S is given as us0 , and the inertial orientation of the cross-section is given by rotation tensor
Rs0 . Similarly, the inertial position vector of the airstation is given as ua0 , and the inertial orientation of the airfoil
basis is given by the rotation tensor Ra0 .
_
a03 Reference
_
configuration
a02
_
Airstation location
a01
Airstation
_ s coordinates
A
a3
u
_s, Rs _
d_0 S
= a2
_
a1
u
_s0, Rs0 _
A
=
S
_
d
u
_a0, Ra0
i3
=
u
_ a, R a
=
_
Present
_
i2
configuration
i1
(3)
where the last equality follows from the rigid body motion assumption: d = Rs d0 . Similarly, because the airfoil basis, A, undergoes rotation Rs ,
Ra = Rs Ra0 .
(2)
(4)
(5a)
(5b)
and the angular velocity and acceleration vectors of bases E and A are, of course, identical
= a = s
= a = s
(6a)
(6b)
The formulation of two-dimensional, unsteady aerodynamic theories typically requires the velocity of the flow with
respect to the airfoil, which simply writes
va = V + v a ,
(7)
where V is the far field flow velocity, assumed to be of constant magnitude and orientation and the average inflow
velocity over the airfoil. When using the internal aerodynamic model, airstations must be located at the airfoils
quarter-chord point, to comply with the usual aerodynamic convention; hence, eq. (7) gives the relative velocity of
the flow with respect to the airfoil quarter-chord point.
As shown in fig. 3, the components of the relative velocity
_
vector resolved in basis A are denoted
a3
_
a2
U1
V
va = RTa va = U2 ,
(8)
U3
U3
U2
V
where notation () indicates components of vectors and tensors resolved in basis A. Note the minus sign for the U2 component: to comply with the sign convention used in aerodynamic Figure 3: Relative velocity of the flow with respect
theories, the velocity component parallel to the chord direction to the quarter-chord point.
is taken positive towards the trailing edge. Velocity components U1 , U2 , and U3 are sometimes denoted UR , UT , and
UP , respectively, indicating the radial, tangent and perpendicular components of the relative flow.
T va + RTa v a =
T va RTa v a . Equation (7)
Next, the relative acceleration of the flow is evaluated as v a = RTa
was used to obtain the last equality and the following approximation was made: v a v a because the far field
flow, including the inflow, is assumed to be of constant magnitude and orientation. The components of the relative
acceleration vector resolved in basis A are denoted
A1
a
a = A2 .
(9)
A3
3
It follows that A3 = 2 U1 + 1 U2 v a3
, and hence
A3 1 U2 v a3
,
(10)
U3 = V sin .
(12)
_
a2
_
a3
_vqc
_vmc
_vtc
quarter
three-quarter
The relative velocities at the airfoil quarter-, mid- and
chord
_
qc
mc
tc
chord
three-quarter-chord points, denoted va , va and va , respecmid
tively, are sometimes used in unsteady aerodynamics theories,
chord
as illustrated in fig. 4. For instance, the unsteady aerodynamic
developed by Peters, see section 3, is based on the relative ve- Figure 4: Velocity vectors at the airfoil quarter-, midlocity of the flow with respect to the airfoil mid-chord point. and three-quarter-chord points.
Since the airfoil is assumed to be rigid, the following relationships must hold: vqc
mc
a and vtc
mc
a , where is the relative position vector of the quarter-chord
a = v
a +
a = v
a
point with respect to the mid-chord point.
3
3.1
This section summarizes the state-space unsteady aerodynamic theory developed by Peters and coworkers for flexible
airfoils. Full details of the derivation of the the theory can be found in Peters and Johnson [2], Peters et al. [3], and
Ahaus and Peters [4].
Figure 5 depicts the general coordinate system used for the development of the theory. The airfoil is undergoing an arbitrary unsteady
v0
y
motion through a mass of still air. At a given instant, the velocity
v1
of the air as viewed by an airfoil attached observer is denoted u0
Leading
u0
along the positive x direction and v0 along the positive y direction.
edge
The velocity gradient, denoted v1 , is counted positive for a nose-up
rotation of the airfoil. The origin of the coordinate system is located
-b
+b x
at the airfoils mid-chord location, and hence, spans b x +b.
h(x, t)
The deformation of the airfoil, denoted h(x, t) is assumed to remain
small, i.e., h/b 1, h/x 1, and h/t u0 .
All the variables of the problem are expanded in terms of the
Figure 5: General airfoil coordinate system.
Glauert angle, , defined as
x = b cos , b x +b, 0 .
(13)
For direct flow, the trailing and leading edges correspond to = 0 and , respectively.
The bound circulation per unit length on the airfoil is expanded in terms of the Glauert angle as
"
#
s
0 cos X
b = 2
+
n sin n ,
sin
sin
n=1
and a similar expansion holds for the pressure drop across the airfoil
#
"
0 cos X
s
+
n sin n ,
P = 2
sin
sin
n=1
(14)
(15)
where is the air density. The first two terms in these expansions are the singular potential functions that allow
suction peaks at either end of the airfoil, when appropriate.
The total bound circulation, denoted
= 2b(s + 1 /2).
(16)
4
n=0
n cos n,
(17a)
hn cos n,
(17b)
n=0
X
X
nhn
n sin n
h/x =
hn
=
+
b
sin
b
n=0
n=1,3,5
n=1
k=n+1,n+3,
2khk
cos n,
b
(17c)
Of course, inverse relationships can be found by using the orthogonality properties of trigonometric function. For
instance,
h0 =
hi =
h d,
(18a)
h cos i d,
i = 1, 2, . . . , .
(18b)
(19a)
(19b)
2
) + u0 1 = 1 ,
2
( n1 n+1 ) + u0 n = n
,
2n
n
(19c)
b( 0
n 2.
(19d)
The inflow states are related to the time derivative of the total bound circulation,
2
) + u0 1 = ,
2
(n1 n+1 ) + u0 n =
,
2n
n
b( 0
(20a)
n 2.
(20b)
(21a)
(21b)
n > 2.
(21c)
P cos n b sin d.
(22)
+b
P cos n dx =
Introducing the expansion for the pressure drop across the airfoil, eq. (15), then yields
1
L0
= (f 0 + ),
2b
2
1
2
L1
= (0 ),
2b
2
2
Ln
1
= (n1 n+1 ),
2b
4
3.1.1
(23a)
(23b)
n 2.
(23c)
(24)
where f is the reverse flow parameter, which is designed to enforce the vanishing pressure drop at the trailing edge
of the airfoil. Because the trailing edge can change position depending on the sign of u0 , different values of this
parameter can be selected
1
u
/|u
|
fully reversed flow regime,
0
f=
(25)
q0
If f = 1, the vanishing of the pressure drop across the airfoil is enforced at x = b; this is equivalent to ignoring
the reversed flow regime. If f changes sign with u0 , the vanishing of the pressure drop across the airfoil switches to
x = b when u0 changes sign; this corresponds to a fully reversed flow regime. The last option enforces a smooth
transition between the direct and reversed flow regimes with symmetrical suction peaks at x = b when u0 = 0; the
reverse flow parameter can be interpreted as f = cos , where is the angle of attack at mid-chord.
3.1.2
Loading components
Introducing the total velocity components into eqs. (21) yields the components of the pressure drop across the airfoil
and eqs. (23) finally give the generalized loads acting on the airfoil as
L0
2b
L1
2b
L2
2b
Ln
2b
u0
b
w 2
w1 (w 0
),
2
2
2
u0
u0
b
=
(w0 0 ) w2 (w 1 w 3 ),
2
4
16
u0
b
w 2
b
=
(w1 w3 ) + (w 0
) (w 2 w 4 ),
4
4
2
24
u0
b
b
=
(wn1 wn+1 ) +
(w n2 w n )
(w n w n+2 ),
4
8(n 1)
8(n + 1)
= f u0 (w0 0 )
(26a)
(26b)
(26c)
n 3.
(26d)
1
1
= f (w0 0 ) + w1 1 .
(27)
2b
2
2
The components of the total velocity are expressed in term of the frame motion, u0 , v0 , and v1 and the components
of airfoil deformation, hn , as
w0 = v0 + h 0 + u0
X
khk
,
b
k=1,3
w1 = v1 + h 1 + 2u0
wn =
h n + 2u0
khk
,
b
k=2,4
k=n+1,n+3
3.1.3
(28a)
(28b)
khk
,
b
n 2.
(28c)
Introducing the total velocity components given by eq. (28) into eqs. (26) gives the generalized loads acting on the
airfoil in terms of the frame motions and airfoil deformation parameters
L0
1
0 h2 ) u0 u0 (v1 + h 1 ) 1 ( u 0 h1 u0 v1 ),
= (v 0 + h
2
2b
2
2
b
b
2 b
b
2
1
L1
3 ) + 1 u0 (v0 + h 0 0 ) 1 u0 h 2 + 1 u0 h1 1 u 0 h2 ,
= (v 1 +
h1 h
2
2
2b
16
2 b
2 b
2b
4 b
2
1
L2
0 ) h2 + h4 + 1 u0 ( v1 + h 1 h 3 ) + u0 h2 + 1 u 0 (h1 h3 ),
= (v 0 + h
2b2
4
6
24 2 b 2
b2
4 b
2
hn2
hn+2
nhn
1 u0
Ln
n+1 ) + 1 u0 nhn + 1 u 0 (hn1 hn+1 ).
=
+
+
(
h
h
n1
2b2
8(n 1) 4(n2 1) 8(n + 1) 2 b
2 b2
4 b
(29a)
(29b)
(29c)
(29d)
X
h0
k1 h
k+1
1
u
v
u
1
u
D0
h
0
0
1
0
0
h1 + v 1 h2 ,
= (v0 + h 0 0 ) +
+ k h k +
khk hk +
+
+
2b
4
b
2 b
2
4
b 2
4
k=1
(30)
1
= + (v1 + h 1 1 ).
2b
2
In these equations, the following notation was defined
X
X
u
0
f
= f (v0 + h 0 0 ) +
khk +
khk .
b
k=1,3,5
3.1.4
(31)
(32)
k=2,4,6
Peters theory was developed in the general airfoil coordinate system depicted in fig. 5 and eqs. (22) imply the positive
sign conventions for loading components L0 and L1 shown in the upper portion of fig. 6. Note that the loading
components are evaluated at the airfoils mid-chord location.
In airfoil theory, it is customary to compute the lift, denoted L, aerodynamic moment, denoted M , and lift induced
MidPeters theory
drag, denoted D, with respect to the airfoils quarter-chord.
chord sign convention
Typically, the lift is counted positive up, the aerodynamic mo-b
D0 +b
ment is positive when causing a nose-up rotation, and the drag
L1
x
positive towards the leading edge, as depicted in the lower porL0
L
tion of fig. 6. The following relationships follow
b/2
b/2
b
L = L0 ,
(33a)
D
L0
b
M = b(L1 +
) = bL1 L,
(33b)
-b
+b x
M
2
2
D = D0 ,
(33c)
QuarterAirfoil theory
sign convention
chord
and will be used to transform the loading components obtained
from Peters theory to the lift, moment, and drag components
Figure 6: Sign conventions for Peters and airfoil thecommonly used in airfoil theory.
In section 2, the relative velocity of the air with respect to ories.
the airfoil was evaluated, and the component of this relative
velocity vector resolved in basis A were denoted U2 and U3 along unit vectors 2 and 3 , respectively, as shown in
fig. 3. The following notational transformations follow,
u 0 = U2 ,
v0 0 = U3 ,
(34a)
(34b)
v1 = b1 .
v 0 = A3 = 1 U2 v a3 ,
(34c)
(34d)
where the last equality results from eq. (10). In the following sections, all vector and tensor components are resolved
in basis A; consequently, superscript () is dropped to simplify the writing.
With these new definitions, the loading components on the airfoil given by eqs. (29) become
# "
#
"
U h
0
2
U2
1
h
h
U2
U2
L0
2
1
f U3 +
f h0 + h 1
+
,
(35a)
=
+ U2 1
v a3
+
2b2
b
b
b
b 2
2
2
4
# "
#
"
1
3
L1
b
U2 h 0 h 2
h
h
U22 h1
U 2 h2
1 U2
+
+ 1
,
(35b)
=
U3 + 2
+
2b2
2 b
b 2
b
2
b 4
16
16 16
# "
#
"
2
0
2
4
L2
1
1
h
h
h
U2 h 1 h 3
U 2 h1 h3
U2 2h2
+
+ v a3 +
,
(35c)
=
U2 1 +
+
+
2b2
b2 2
2
b
2
b
4
4
4
6
24
# "
#
"
2
n2
n+2
n
U2 h n1 h n+1
h
h
U 2 hn1 hn+1
nh
U2 nhn
Ln
+
+
.
(35d)
=
+
+
2b2
b2 2
b
2
b
4
8(n 1) 4(n2 1) 8(n + 1)
The drag force, given by eq. (30), becomes
"
X
U2
D0
U2
= U3 f U3 +
+ 1 U2 h1 2f U3 +
+ f h0 h0 +
2b
b
b
k=1
!
#
"
X
b
hk1 hk+1
h0
1
+
v a3 h1 + 1 h2 +
hk ,
4
2
4
4
k=1
U2
U 2 khk
k hk +
b
b 2
hk
(36)
h 1
U2
1
b
.
= f U3 +
+
1 + f h0 +
2b
b
2
2
2
(37)
In the evaluation of the total bound circulation, the inflow gradient, 1 , is often neglected because it is not easily
evaluated. In these equations, scalar was defined as
=f
khk +
k=1,3,5
khk .
(38)
k=2,4,6
b2
v a3 ,
2
b4
b
b2
U2 U3 a0 1 L.
2
16
2
(39a)
(39b)
In the expression for the lift, the first term represents the steady lift component, the second the aerodynamic damping
in pitch, and the last the inertial effect. The drag force is
D = a0 f bU32 ,
(40)
b
b2
1 a0 1 .
2
2
The theoretical value of the slope of the lift curve, 2, was replaced by a0 in the above expressions.
= a0 f bU3 + a0
4.1
(41)
The steady components of the lift and drag forces are obtained from eqs. (39a) and (39b), respectively, as
Ls = a0 f b U2 U3 ,
s
D = a0 f b
M s = a0
U32 ,
(42a)
(42b)
b
b
U2 U3 L s .
2
2
(42c)
These expressions were derived for a thin airfoil in a two-dimensional inviscid, incompressible flow, and hence, are
approximate in nature.
Steady wind tunnel measurements have been acquired for numerous airfoils and the corresponding lift, drag, and moment coeffi a_
3
cients are available in tabulated form as functions of angle of attack
wt
L
_
and Mach numbers. It is also possible to use computational fluid
a2
Zero-lift-line
Ls
dynamics codes to compute the same coefficients based on theories
s
D
that remove many of the assumptions made in Peters theory. It then
seems appropriate to correct eqs. (42) to make sure they yield the
V
wind tunnel measurements for steady flows.
Figure 7 depicts the typical experimental setup used in wind
tunnel experiments. The airfoil is set at an angle of attack with Figure 7: Test set-up for an airfoil in a wind
respect to the steady, far-field flow velocity, V . The angle of attack tunnel.
is defined as the angle between the far-field flow velocity direction
and the zero-lift-line of the airfoil. The force in the direction normal to the flow is defined as
Lwt = c (, M) bV 2 ,
(43)
where c are the experimentally measured lift coefficients. Typically, these coefficients are tabulated as functions of
the angle of attack, , and Mach number, M.
8
The lift and drag forces acting on the airfoil in a steady flow are given by eqs. (42a) and (42b), respectively and
can be manipulated in the following manner
U3 U2
= bV 2 (a0 f sin ) cos ,
V V
U3 U3
= bV 2 (a0 f sin ) sin ,
a0 f
V V
Ls = b a0 f U3 U2 = bV 2 a0 f
(44a)
Ds = b a0 f U3 U3 = bV 2
(44b)
respectively, where is the quarter-chord angle of attack defined by eq. (12) and V the far field flow velocity in the
plane of the airfoil given by eq. (11). A cursory look at fig. 7 reveals that the steady lift and drag forces defined
by eqs. (42a) and (42b), respectively, are the projections of the lift measured in the wind tunnel along axes a
3 and
a
2 , respectively, i.e., Ls = Lwt cos and Ds = Lwt sin . Clearly, wind tunnel test measurements and theoretical
predictions become identical if the following choice is made
c (, M) = a0 f sin .
(45)
Next, correction of the steady quarter-chord moment is also derived. The expression for the steady moment is
given by eq. (42c) and can be manipulated in the following manner
s
Mqc
= a0
b2
(1 f )U2 U3 .
2
(46)
For direct flow, f = 1, and the quarter-chord moment vanishes, as expected. In the wind tunnel test, a non-vanishing
quarter-chord steady moment is measured and the corresponding moment coefficient, denoted cm , is defined as
wt
Mqc
= cm (, M) 2b2 V 2 .
(47)
Here again, the moment coefficient, cm , is now a function of the angle of attack, , and Mach number, M.
In summary, after corrections to account for wind tunnel measurements, the steady lift, drag, and moment expressions become
s = c (, M) bV U2 ,
L
s = c (, M) bV U3 ,
D
s = cm (, M) 2b2 V 2 .
M
qc
4.2
(48a)
(48b)
(48c)
The unsteady components of lift, drag, and moment are obtained from eqs. (39a) and (39b), respectively, as
Lus = a0 b2 U2 1 a0
b2
v a3 ,
2
Dus = 0,
us
Mqc
= a0
(49a)
(49b)
b
b
1 Lus ,
16
2
(49c)
where 1 the pitch rate of the airfoil, counted positive about axis a
1 , i.e., it is positive for a nose-up rotation.
4.3
Because Peters theory is based on a two-dimensional inviscid, incompressible flow around a thin airfoil, it does not
predict the viscous drag. Here again, wind tunnel test measurements can be used to supplement the theory.
Figure 8 shows the drag force, Dwt , measured in the wind
_
_
tunnel and expressed as
a3
a2
Dwt = cd (, M) bV 2 ,
(50)
Dwt
S3
V
S2
where cd is the drag coefficient. The components of the drag force
along axes
a2 and a
3 are denoted S2 and S3 , respectively, and
can be expressed as
Figure 8: Measurement of the drag force in the
S2 = cd bV 2 cos = cd bV U2 ,
(51a) wind tunnel.
S3 = cd bV 2 sin = cd bV U3 .
(51b)
In the presence of flow along the blade, the flow velocity component along axis a
1 is denoted U1 . Defining the
resultant flow velocity VR2 = U12 + U22 , the associated skin friction drag is Dsf = cd0 bVR2 , where cd0 is the skin friction
drag coefficient, estimated to be equal to the drag coefficient at zero lift force. The components of this force along
axes a
1 and
a2 are S1 = cd0 bVR U1 and S2 = cd0 bVR U2 , respectively.
The drag forces acting on the airfoil are obtained by summing up the pressure and skin friction drag components.
The net force along axis
a2 will be S2 = cd bV U2 + cd0 bVR U2 . Clearly, this expression is incorrect because it
double counts the skin friction drag, and hence, it will be corrected as S2 = cd bV U2 cd0 bU22 + cd0 bVR U2 . This
last expression recovers wind tunnel measurements in the absence of radial flow. In summary, the pressure and skin
friction drag components are
S1 = cd0 (, M) bVR U1 ,
S2 = cd (, M) bV U2 + cd0 (, M) b(VR U2 )U2 ,
(52a)
(52b)
S3 = cd (, M) bV U3 .
4.4
(52c)
Summary
Peters unsteady aerodynamic theory, corrected by wind tunnel test measurements, predicts aerodynamic forces along
axes a
1 , a
2 , and a
3 , denoted F1 , F2 , and F3 , respectively and moments about axes a
1 , a
2 , and a
3 , denoted M1 , M2 ,
and M3 , respectively. The expressions for these forces and moments are found by summing up their corrected steady
counterparts, eq. (48), unsteady components, eq. (49), and the drag and skin friction components, eq. (52), to find
F1 =
S1 ,
(53a)
s
S2 ,
F2 = ftl D
s
us
+ S3 ,
F3 = ftl L + ftl L
s
us
+ ftl M ,
M1 = ftl M
(53b)
M2 = 0,
M3 = 0.
(53e)
(53f)
qc
(53c)
(53d)
qc
The tip loss correction factor, denoted ftl , was introduced in these expressions.
4.5
The unsteady aerodynamic formulation described in the previous section is closely related to Theodorsens theory, as
described by Bisplinghoff et al. [5].
h
i
1
2
ba
+ 2V bC(k) h + V
L = b h + V
+ b( a)
,
(54a)
2
1
1
V b( 1 a)
+ 2V b2 (a + 1 )C(k) h + V
mc = b2 bah
M
b2 ( + a2 )
,
(54b)
+ b( a)
2
8
2
2
Z
U2
U3
b
h0
V
^
X
U2
b
^
b
ba h
U3
Figure 10: Theodorsens theory. V is the flow velocity.
h(t) is the downward displacement of the axis (x = ba)
of rotation,
(t).
Figure 9: Peters formulation. U2 is the uniform x velocity, and U3 is the uniform z velocity. h0 (t) is the motion
in the z direction.
In Peters formulation, the effect of the trailed vortices is taken into account through the inflow velocity that is
added to the far filed flow velocity, whereas in Theodorsens theory, the same vorticity enters the formulation through
the lift deficiency function C(k). Hence, identical expressions should be recovered from the two approaches by setting
= 0 and C(k) = 1 in Peters and Theodorsens formulations, respectively. Since Peters formulation is based on the
relative velocity of the flow with respect to the airfoil mid-chord point, we set a = 0 in Theodorsens theory. For a
10
and h h0 . Also, 1 =
is the airfoil pitch rate. With these approximations, the lift and mid-chord pitching moment
then become
2
= a0 U2 b h 0 + U3 + a0 b2 h
0 + U 3 + a0 U2 b 1 ,
L
(56a)
2
2
2
2
mc = a0 b2 U2 b 1 b 1 + a0 U2 b (h 0 + U3 ) + b 1 = a0 b4 1 + a0 b2 U2 h 0 + U3 ,
(56b)
M
2
2
8
2
2
16
2
0 into U 3 , since h 0 and U3 are both the velocities of the flow
Note that we can incorporate h 0 into U3 , and h
with respect to the airfoil in Peters formulation. The first term in the lift expression is the steady lift, identical to
eq. (42a), whereas the last two terms represent the unsteady lift, identical to eq. (49a). This shows the equivalence of
the lift expressions in the two theories. To show the equivalence of the pitching moment expressions, the mid-chord
mc bL/2
to yield
moment given by Theodorsen is computed at the quarter-chord as Mqc = M
mc b L
= a0 b4 1 + a0 b2 U2 h 0 + U3 b L.
Mqc = M
2
16
2
2
Introducing the expression for the lift, eq. (56a), then leads to
b2
b2
b
b4
a0 (h
+
U
)
+
a
U
Mqc = a0 1
0
3
0
2
1 .
16
2
2
2
(57)
(58)
Since the bracketed term of this equation is identical to the unsteady lift given by eq. (49a), Theodorsens quarter-chord
moment becomes equivalent to its counterpart in Peters formulation, eq. (49c).
The results presented in the previous section summarize the application of Peters theory to a rigid airfoil. Peters
theory, however, also applies to deformable airfoils, as presented in section 3. Airfoils with trailing edge flaps can be
viewed as deformable airfoils, as shown in fig. 11. The main wing moves with the rigid body coordinate system and
the trailing flap has a mid-span defection denoted and an angular deflection denoted ; positive sign conventions
are depicted in the figure.
5.1
(59)
where is the Glauert angle defining the position of the trailing edge
flap and db the distance this hinge is located aft the airfoils mid
chord. The configuration depicted in fig. 11 shows a discontinuity
between the main wing and the flap. For typical constructions, the
Flap
bd
Main wing
flap is connected to the wing by means of brackets and revolute
b(1 + d)/2
joints. At these attachment points, the main wing and flap connect
b
without discontinuity; on the other hand, small discontinuities are
likely to occur at intermediate points. When the main wing and flap
connect smoothly, the flap mid-point deflection is not independent Figure 11: Configuration of the trailing edge
of its angular deflection
flap.
=
b
(1 d).
2
(60)
11
The components of the Glauert expansion of the camber line of the airfoil are found by introducing the airfoil
displacement field (59) into the generic expression (18) to find
1+d
b
h0 = +
sin
,
(61a)
2
sin 2
b
2 sin
(1 + d) sin +
,
(61b)
+
h1 =
2
2 sin n
b sin(n 1)
sin n sin(n + 1)
hn =
, n 2.
(61c)
+
(1 + d)
+
n
n1
n
n+1
It is convenient to introduce the following matrix
sin cos
sin
cos
2
sin
sin
3
cos sin 2 +
b
3
2
1
,
T =
..
(62)
and using eq. (60), the airfoil displacement components given by eq. (61) now become
h = T .
(63)
(64)
where L is the array of airfoil loading components, whose entries are defined by eqs. (35).
5.2
The loading components defined in eqs. (35) for a general deformable airfoil apply to the present airfoil with a trailing
edge flap. Because the flap kinematics is characterized by the single flap angular deflection, constant is given by
eq. (38) and the loading components become
# "
#
"
U h
0
2
U2
1
h
h
U2
U2
L0
2
1
f U3 +
f h0 + h 1
+
,
(65a)
=
+ U2 1
v a3
+
2b2
b
b
b
b 2
2
2
4
# "
#
"
1
3
L1
b
U2 h 0 h 2
h
h
U22 h1
U 2 h2
1 U2
+
+ 1
,
(65b)
=
U3 + 2
+
2b2
2 b
b 2
b
2
b 4
16
16 16
# "
#
"
2
0
2
4
1
1
U 2 h1 h3
h
h
h
U2 h 1 h 3
U2 2h2
L2
+
+ v a3 +
,
(65c)
=
U2 1 +
+
+
2b2
b2 2
2
b
2
b
4
4
4
6
24
# "
#
"
2
n2
n+2
n
Ln
U2 h n1 h n+1
h
h
U 2 hn1 hn+1
nh
U2 nhn
+
+
.
(65d)
=
+
+
2b2
b2 2
b
2
b
4
8(n 1) 4(n2 1) 8(n + 1)
The drag force given by eq. (36) becomes
"
X
U2
D0
U2
= U3 f U3 +
+ 1 U2 h1 2f U3 +
+ f h 0 h 0 +
2b
b
b
k=1
!
#
"
Xh
k1 h
k+1
0
b
1
h
v a3 h1 + 1 h2 +
hk ,
+
4
2
4
4
U 2 khk
U2
k hk +
b
b 2
hk
(66)
k=1
h
U2
b
1
.
= f U3 +
+
1 + f h 0 +
2b
b
2
2
2
12
(67)
5.3
The steady components of airfoil loading are obtained from eqs. (65) as
Ls0
2b2
Ls1
2b2
Ls2
2b2
Lsn
2b2
and the drag force becomes
U2
U2
(f U3 +
),
b
b
1 U2
U 2 h1
=
U3 + 22
,
2 b
b 2
U 2 2h2
= 22
,
b 2
U 2 nhn
= 22
,
b 2
=
D0s
U2
= U3 (f U3 +
),
2b
b
(68a)
(68b)
(68c)
(68d)
(69)
Using a procedure similar to that outlined in eqs. (44a) and (44b), the steady components of lift and drag become
U3
U2 U2
+ a0
= bV 2 a0 f sin + a0 cos cos ,
(70a)
Ls = b a0 f U3 + a0 U2 U2 = bV 2 a0 f
b
V
b V
V
b
U3
U2 U3
Ds = b a0 f U3 + a0 U2 U3 = bV 2 a0 f
+ a0
= bV 2 a0 f sin + a0 cos sin ,
(70b)
b
V
b V
V
b
respectively, where is the angle of attack defined by eq. (12) and V the far field flow velocity in the plane of the
airfoil given by eq. (11). A cursory look at fig. 7 reveals that the steady lift and drag forces defined by eqs. (42a)
and (42b), respectively, are the projections of the lift measured in the wind tunnel along axes a
3 and a
2 , respectively,
i.e., Ls = Lwt cos and Ds = Lwt sin . Clearly, wind tunnel test measurements and theoretical predictions become
identical if the following choice is made
c (, M, ) = a0 f sin + a0
cos .
b
(71)
Note that the lift coefficient, c , is now a function of the angle of attack, , flap angular deflection, , and Mach
number, M.
Next, correction of the steady quarter-chord moment is also derived. The expression for the steady moment is
given by eq. (42c) and can be manipulated in the following manner
s
Mqc
= a0
b2
(1 f )U2 U3 + a0 b(h1 /2)U22 .
2
(72)
For direct flow, f = 1, and the quarter-chord moment is a function of the flap deflection only, as expected. In the
wind tunnel test, a non-vanishing quarter-chord steady moment is measured and the corresponding moment coefficient,
denoted cm , is defined as
wt
Mqc
= cm 2b2 V 2 .
(73)
Here again, the moment coefficient, cm , is now a function of the angle of attack, , flap angular deflection, , and
Mach number, M. Finally, the static hinge moment is obtained from eq. (64) as H s = T T Ls . In the wind tunnel
test, the flap hinge steady moment is measured and the corresponding moment coefficient, denoted ch , is defined as
H wt = ch (, M, ) 2b2 V 2 .
(74)
In summary, after corrections to account for wind tunnel measurements, the steady lift, drag, moment, and hinge
moment expressions become
s = c (, M, ) bV U2 ,
L
s = c (, M, ) bV U3 ,
D
s
qc
M
= cm (, M, ) 2b2 V 2 .
s = ch (, M, ) 2b2 V 2 .
H
13
(75a)
(75b)
(75c)
(75d)
5.4
The unsteady airfoil loading components are obtained from eqs. (65) as
# "
#
"
U h
0
2
1
Lus
U2
h
h
2 1
0
+
,
= U2 1
v a3
+
f h0 + h1
2b2
b
b 2
2
2
4
# "
#
"
1
3
b
h
h
U 2 h2
U2 h 0 h 2
Lus
1
+ 1
,
=
+
2b2
b
2
b 4
16
16 16
# "
#
"
0
2
4
Lus
1
h
h
h
1
U2 h 1 h 3
U 2 h1 h3
2
+ v a3 +
,
=
U2 1 +
+
+
2b2
2
b
2
b
4
4
4
6
24
# "
#
"
n2
hn+2
n
Lus
h
U 2 hn1 hn+1
nh
U2 h n1 h n+1
n
+
.
=
+
+
2b2
b
2
b
4
8(n 1) 4(n2 1) 8(n + 1)
The unsteady flap hinge moment is then
H us = T T Lus .
X
D0us
U2
U2
U 2 khk
hk
= 1 U2 h1 2f U3 +
+ f h0 h0 +
k hk +
2b
b
b
b 2
k=1
"
!
#
X
0
k+1
h
hk1 h
b
1
+
v a3 h1 + 1 h2 +
hk ,
4
2
4
4
(76a)
(76b)
(76c)
(76d)
(77)
(78)
k=1
and finally, the total bound circulation follows from eq. (67) as
#
"
1
h
b
us
1
.
=
1 + f h 0 +
2b
2
2
2
More specifically, the unsteady lift and aerodynamic moment at the quarter-chord are
"
"
#
#
U h
h0
2
U2
1
h
2
1
2
us
2
f h0 + h 1 +
+ a0 b v a3 +
,
L = a0 b U2 1 +
b
b 2
2
2
4
"
#
#
"
h1
3
b
b
h
U 2 h2
3
us
3 U 2 h0 h2
+ a0 b 1
Lus .
+
Mqc = a0 b
b
2
b 4
16
16 16
2
(79)
(80a)
(80b)
us
! #
X
U2
U 2 khk
U2
hk
+ f h0 h0
k hk +
=a0 b 1 U2 h1 + 2f U3 +
b
b
b 2
k=1
"
!
#
0
X
k+1
h
b
hk1 h
1
a0 b
v a3 h1 + 1 h2 +
hk ,
4
2
4
4
"
(81)
k=1
5.5
Because Peters theory is based on a two-dimensional inviscid, incompressible flow around a thin airfoil, it does not
predict the viscous drag. Here again, wind tunnel test measurements can be used to supplement the theory. The
pressure and skin friction drag components developed in section 4.3 can be added to the loading components for an
airfoil with a trailing edge flap as well.
5.6
Summary
Peters unsteady aerodynamic theory, corrected by wind tunnel test measurements, predicts aerodynamic forces along
axes a
1 , a
2 , and a
3 , denoted F1 , F2 , and F3 , respectively and moments about axes a
1 , a
2 , and a
3 , denoted M1 , M2 ,
and M3 , respectively. The expressions for these forces and moments are found by summing up their corrected steady
14
counterparts, eq. (75), unsteady components, eq. (80), and the drag and skin friction components, eq. (52), to find
F1 =
S1 ,
s
us
S2 ,
F2 = ftl D + ftl D
s
us
+ ftl L
+ S3 ,
F3 = ftl L
s
us
+ ftl M ,
M1 = ftl M
qc
qc
(82a)
(82b)
(82c)
(82d)
M2 = 0,
(82e)
M3 = 0.
(82f)
The tip loss correction factor, denoted ftl , was introduced in these expressions. In addition, the flap hinge moment is
the sum of its steady and unsteady components give by eqs. (75d) and (77), respectively,
s + ftl H us .
H = ftl H
(83)
Petot [6] extended the original ONERA model for application to the operating conditions encountered by helicopter
airfoils. This model is based on delayed dynamic stall and second-order linear differential equations to compute the
dynamic stall loads. The second-order differential equations that govern this model are in the following form
dC
d2
d
2
2 2
2
+ V
+ V = V V C + eV
.
(84)
dt2
dt
dt
Three identical equations govern the lift, drag, and quarter-chord aerodynamic moment acting on the airfoil. In
eq. (84), is the circulation per unit span of the blade, C is the difference in the aerodynamic coefficient between
its linear static value extrapolated in the stalled region and its actual value at the angle of attack considered. The
far field flow velocity is denoted V and V = V /b, where b is the semi-chord length. Parameters and e are nonlinear
functions of CL defined as
= 0 + 1 CL2 ,
(85a)
0 + 1 CL2 ,
e0 + e1 CL2 ,
(85b)
=
e=
(85c)
parameters 0 , 1 , 0 , 1 , e0 , and e1 characterize a given airfoil shape and are function of Mach number. Typical
values of these parameters [6, 7] are listed in table 1. Note that coefficients , , and e for the lift, drag, and
quarter-chord moment acting on the airfoil are considered to be functions of CL only.
For the lift, drag, and quarter-chord aerodynamic moment problems, C is defined by the following equations
CL = c (c0 + c sin ) ,
CD = cd cd0 + cd sin2 ,
CM = cm (cm0 + cm sin ) ,
(86a)
(86b)
(86c)
respectively, as the difference in the aerodynamic coefficient between its linear static value extrapolated in the stalled
region and its actual value at the angle of attack considered.
Let the solutions of eq. (84) for the lift, drag, and quarter-chord moment circulations be denoted L , D , and M ,
respectively. The corresponding airloads due to dynamic stall then follow as
LDS = V bL ,
D
DS
(87a)
= V bD ,
M DS = 2V b2 M + bLDS
(87b)
f 1
,
2
(87c)
where the second term in the moment equation accounts for reverse flow effects through the reverse flow parameter
defined by eq. (25).
6.1
Solution process
,
+ 2
= 2 C + eC
+
15
(88)
Equation
Lift
Drag
Moment
Parameters
0
1
0
1
e0
e1
d
0
1
e0
e1
d
0
1
e0
e1
d
Range
[0.1, 0.4]
[0.0, 0.5]
[0.1, 0.4]
[0.0, 0.6]
[0.0]
[0.2, 0.0]
[8.0]
[0.0, 0.5]
[0.0, 0.6]
[0.0]
[0.0, 0.05]
[0.0]
[0.0, 0.4]
[0.0, 0.6]
[0.0]
[0.0, 0.06]
[2.0]
Default Value
0.2
0.2
0.3
0.2
0.0
-0.05
8.0
0.25
0.0
0.0
-0.015
0.0
0.25
0.1
0.0
0.01
2.0
2
2
2
where subscripts ()i and ()f denote value of variables at the beginning end end of the time step, respectively and
subscript ()m = [()f + ()i ]/2 a mid-point quantity. Using the same discretization for function yields (i + f )/2 =
(f i )/ , and hence, f = 2(f i )/ i . Introducing this result into eq. (89) and solving for f yields
cf f = ci i +
f =
2
2i df Cf di Ci ,
2
(f i ) i ,
2
2
di = m
(1
em ).
(90a)
(90b)
(91a)
(91b)
(91c)
(91d)
(92)
where is the non-dimensional time, as computed by eq. (164), after the angle of attack exceeds the static stall value
1 , and di is the non-dimensional time delay listed in table 1.
Once we solve for the circulations, we can compute the corresponding dynamic stall aerodynamic coefficients as
,
V
d
,
=
V
m
1
=
+ CDS (f 1).
V
4
CDS =
(93a)
CdDS
(93b)
DS
Cm
16
(93c)
The two-dimensional unsteady aerodynamic behavior of airfoils described in this section is based on the work of
Leishman and Beddoes [8, 9, 10, 11, 12]. The formulation can be separated into two components: the unsteady
behavior of the attached flow and the separated flow behavior. These two aspects of the problem will be treated
separately in sections 7.1 and 7.2, respectively.
7.1
The formulation of the unsteady behavior of the attached flow leads to the following set of first order ordinary
differential equations
x = A x + B u,
(94a)
y = C x + D u,
(94b)
where u is the input array, y the output array, and xT = x1 , x2 , ..., x8 the state array.
The relative flow velocity at quarter-chord is given by eq. (7) and its components along unit vectors a
2 and a
3 are
U2 an U3 , respectively, as given by eq. (8). The input array is defined as
(t)
u21 =
,
(95)
q(t)
where = arctan(U3 /U2 ) is the angle of attack measured in radians from the zero-lift line, as shown in fig. 3, and
q = c/V
the normalized pitch rate, counted positive for nose-up rotation. The output array is defined as
p
CN (t)
,
(96)
y21 =
CM (t)
P
where CN
is the normal force coefficient under potential flow conditions, and CM the quarter-chord pitching moment
coefficient.
Once a solution has been obtained for theses two coefficients, the effective angle of attack of the airfoil due to the
Cc
shed wake (circulatory) terms can be written in terms of the
aE
states x1 and x2 as
CL
p
CN
E (t) = V 2 (A1 b1 x1 + A2 b2 x2 ),
(97)
aE
a
where b is the semi-chord length and M the Machnumber.
The Prandtl-Glauert compressibility factor is = 1 M 2
CD
V
and V = V /b.
The chord force and pressure drag coefficients, as shown in
fig. 12, are then found as
Cc (t) = CN 2E (t),
CD (t) =
p
CN
sin Cc cos ,
(98a)
CN V 2 A1 b1
1 0.5
CN V 2 A2 b2
1 0.5
4/(M TN )
1 0
1/(M TNq )
0 1
T
B 82 =
, and C 28 =
1 0
1 0
0
0 1
0
0
0 1
17
0
,
A3 /(M b3 TM )
A4 /(M b4 TM )
A5 b5 V /8
7/(12M TMq )
(100)
1
4
1
.
=
M (A3 + A4 ) 7/12
(101)
2. CN (M ) is the slope of the lift curve, xac (M ) the aerodynamic center, and recovery factor for viscosity effect.
Typically 0.95, for inviscid flow = 1.
3. The coefficients of circulatory indicial response functions are denoted Ai , i = 1, 2, . . . , 5. In the absence of
unsteady test data, the values of these coefficients can be taken from refs. [11, 12], as shown in Table 2 which
lists different sets of coefficients from various sources. Note that A1 + A2 = 1.
Data Source
Boeing Data
ARA Data
NASA Data
Consolidated Data
A1
0.636
0.625
0.482
0.918
A2
0.364
0.375
0.518
0.082
A3
1.5
1.5
1.5
1.5
A4
-0.5
-0.5
-0.5
-0.5
A5
1.0
1.0
1.0
1.0
b1
0.339
0.310
0.684
0.366
b2
0.249
0.312
0.235
0.102
b3
0.25
0.25
0.25
0.25
b4
0.1
0.1
0.1
0.1
b5
0.5
0.5
0.5
0.5
KN =
kN
(1 M ) + M 2
P2
18
2b
cs
i=1
Ai bi
(102)
TN = KN TI
(103)
KNq =
kNq
0.5(1 M ) + 2M 2
KM =
KMq =
P2
i=1
Ai bi
kM (A3 b4 + A4 b3 )
,
b3 b4 (1 M )
TNq = KNq TI
TM = KM TI
7kMq
,
15(1 M ) + 3M 2 A5 b5
TMq = KMq TI
(104)
(105)
(106)
7.2
To solve for the separated flow behavior, we need to add the following inputs, where in the absence of available test
data, there values can be taken from Table 4. This theory for separated flow and dynamic stall model is presented in
refs. [8, 9, 15].
1. 1 , S1 , S2 ;
From the airfoil static CN data, we can compute the corresponding trailing edge separation point f (as
fraction of the airfoil chord) using the following relation:
" s
f= 2
CN
1
CN (M )
#2
(107)
Then we can curve fit f static relation to two expressions, before and after the static stall, as following:
f=
(108)
where S1 , S2 and 1 are three empirical parameters which may be fitted to the test data. 1 is the break point
corresponding to f 0.7 (closely correspond to the static stall angle of attack for most airfoil sections). There
values in Table 4 are in degrees (and are input to the model also in degrees).
However, this relation must be modified to account for negative angle of attack, as shown in ref. [7]
(
1 0.3 exp((|| 1 )/S1 )
|| 1 ,
f=
0.04 + 0.66 exp((1 ||)/S2 ) || > 1 .
(109)
2. k0 , k1 , k2 , m;
From the airfoil static test data we can compute CM /CN ratio. For the same angles of attack we have f
(obtained as mentioned previously), then we can curve fit their relation to:
CM /CN = k0 + k1 (1 f ) + k2 sin(f m )
(110)
where: k0 = (0.25 xac ) is aerodynamic center offset from the quarter chord (xac is one of the inputs to the
attached flow module (7.1)), k1 represents the direct effect on the center of pressure due to the growth of the
separated flow region, and k2 helps describe the shape of the moment break at stall.
The values of k0 , k1 , k2 and m can be adjust to give the best moment reconstruction. m will be taken equal to
2.0 in the implementation of the model.
3. Tp (M ) and Tf (M ) which are Mach number dependent time constants used in the separated flow behavior.
As in the case of Tp , Tf is albeit weaker mach number dependent. It appears that the values of Tp are largely
independent of airfoil shape. While it is more difficult to determine how Tf changes with airfoil shape. Without
access to unsteady airfoil data, an unsteady boundary layer code can be practically used to determine how Tf
varies with airfoil shape.
19
M
1
1
S1
S2
k0
k1
k2
Tp
Tf
0.30
15.25
2.1
3.0
2.3
0.0025
-0.135
0.04
1.7
3.0
0.40
12.5
2.0
3.25
1.6
0.006
-0.135
0.05
1.8
2.5
0.50
10.5
1.45
3.5
1.2
0.02
-0.125
0.04
2.0
2.2
0.60
8.5
1.0
4.0
0.7
0.038
-0.12
0.04
2.5
2.0
0.70
5.6
0.8
4.5
0.5
0.03
-0.09
0.15
3.0
2.0
0.75
3.5
0.2
3.5
0.8
0.001
-0.13
-0.02
3.3
2.0
0.80
0.7
0.1
0.7
0.18
-0.01
0.02
-0.01
4.3
2.0
Units
deg.
deg.
deg.
deg.
non-dim.
non-dim.
non-dim.
non-dim.
non-dim.
Table 4: Airfoil coefficients for unsteady separated flow modeling, as functions of M , for NACA 0012
Now, to go through the following sections, we need to split down the components of the output vector y, eq.( 96),
into their circulatory and noncirculatory parts , and define each as function of the states or their time derivatives
p
C
I
I
CN
(t) + CN
(t) + CN
(t)
CN (t)
q
=
.
y=
(111)
C
I
C
I
CM (t)
CM (t) + CM (t) + CMq (t) + CMq (t)
where
C
CN
(t) = CN E (t)
4
x 3 .
M
1
I
(t) =
CN
x 4 .
q
M
I
(t) =
CN
C
C
(t).
(t) = k0 CN
CM
1
I
[A3 x 5 + A4 x 6 ] .
(t) =
CM
M
A5 b5 V
C
x7 .
CMq (t) =
8b
7
I
(t)
=
CM
x 8 .
q
12M
(112)
(113)
(114)
(115)
(116)
(117)
(118)
The time derivatives of the states are taken from eq.( 94). Note that we can calculate only the circulatory or the
p
noncirculatory part of CN
and CM as previously shown, then use eq.( 111) to get the other part.
C
Define CN (t) as the circulatory part of the normal force coefficient due to angle of attack only, i.e. it is calculated
similarly to eqs.( 97, 112) as
V
2
C
CN (t) = CN
(A1 b1 x
b1 + A2 b2 x
b2 ).
(119)
b
where these two new states are get from
(120)
where A(i, i) is the ith diagonal term of matrix A (eq.( 99)). Also they can be computed numerically from the corresponding first two states of the attached flow solution, by extracting the effect of the pitch rate q(t) input.
7.2.1
Under unsteady conditions, normal force coefficient CN (t) lag (t), and the leading edge pressure response lag CN (t);
therefore the critical pressure for separation onset (stall) at the appropriate mach number is achieved at a higher
angle of attack then the quasi-static case, and there is an overall delay in the onset of dynamic stall.
p
To implement the critical pressure criterion under unsteady conditions, a 1rst order lag is applied to CN
(t) to
20
S
Tp
(122)
then
= CN
CN
Dpn .
n
n
(123)
where n is the current time sample, S = V t/b is the non-dimensional distance traveled by airfoil in semi-chords,
p
S = Sn Sn1 , and CN
is the total unsteady normal force coefficient for attached flow. The equation for CN
can
n
n
be written in the form of a state equation as ref. [15]
p
CN (t)
x9
x 9 =
+
.
(124)
Tp
Tp
CN
(t) = x9 .
7.2.2
(125)
There exist a modified trailing edge separation point location due to temporal effects on the airfoil pressure distribution
and the boundary layer response. An open loop procedure is developed to represent the time dependent effects on
the trailing edge separation point f , eq. (108), and thereby permits the evaluation of nonlinear forces and moments
under dynamic conditions via the application of Kirchhoff theory. The procedure is performed by
1. First, define an effective angle of attack, f (t), which gives the same unsteady leading edge pressure as for the
equivalent quasi-steady case
C (t)
f (t) = N
.
(126)
CN (M )
2. Using this effective angle of attack, determine a value for the trailing edge separation point, f , from the static
f relationship (108). However, the static hysteresis around stall when the angle of attack is decreasing, is
modeled using an empirically derived dynamic offset, 1n , formulated as ref. [8]
(1 fn1
)0.25 1
S < 0
1n =
(127)
0
S >= 0
where
1 is function of the airfoil and Mach number, as given in degrees in Table 4.
S = (n n1 ) is used to detect the angle of attack is increasing or decreasing with time (pitch up or down).
Note that the unsteady T.E. separation point f used here is from the previous time step (eq. (132)). Then 1
used to calculate f , is calculated as
(128)
1n = 1 1n .
then
f =
|f | 1n
1 0.3 exp [(|f | 1n )/S1 ]
0.04 + 0.66 exp [(1n |f |)/ S2 ] |f | > 1n
(129)
3. Additional effects of the unsteady boundary layer response may be represented by applying a first-order lag to
f to produce final value for the unsteady trailing edge separation point f . The deficiency function is given by
)Ef0.5 .
Dfn = Dfn1 Ef + (fn fn1
(130)
(131)
where
Ef = exp
1 S
Tf
where i , i = 1, 2, 3 are three time constants parameters, and their values will be specified at each time step,
in order to solve the interaction between the different flow features.
Then, we can calculate the dynamic T.E. separation point as
fn = fn Dfn .
(132)
Alternatively, this additional lag in the T.E. separation point can be represented in state space equation as
ref. [15]
f (t)
x10
+
.
(133)
x 10 =
Tf
Tf
f (t) = x10 .
21
(134)
f
4. Finally, the nonlinear normal force coefficient CN
, incorporating the effect of the unsteady trailing edge separation point f , can be given by
fC
f
I
.
(135)
+ CN
= CN
CN
n
n
n
fC
being the circulatory part of the normal force coefficient, including the separated flow effect, comwhere CN
n
puted as
fC
C
.
(136)
CN
= KNn CN
n
n
p 2
(1 + fn )
KNn =
.
(137)
4
and the noncirculatory normal force coefficient computed from eqs. (113) and (114) as
I
I
I
= CN
+ CN
CN
.
n
qn
n
(138)
(139)
fC
being the circulatory part of the normal force coefficient
where CM0 the zero lift pitching moment, and CN
n
due to angle of attack only, including the separated flow effect, computed from eq. (120), similarly to eq. (136),
as
fC
C
CN
.
(140)
= KNn CN
n
n
and the noncirculatory moment coefficient computed from eqs.( 116, 118) as
I
I
I
= CM
+ CM
CM
.
n
qn
n
(141)
Finally, the chord force and pressure drag coefficients, for the separated flow, are given respectively as
p
Ccfn = CN (M )2En fn .
f
f
sin Ccfn cos .
= CN
CD
n
n
(142)
(143)
If the vortex effect is to be added, then we will go to next section. However, if wanted only the effect of the T.E.
separation point, then the total section drag is obtained by adding the profile drag forces, as described in section 4.3.
7.2.3
This model take its input from the previous two sections for unsteady aerodynamic attached and separated flow. We
need to add the following inputs, where in the absence of available unsteady test data, there values can be taken from
Table 5.
1. CN1 ;
From an analysis of airfoil static test data, the critical normal force coefficient CN1 , may be obtained which
corresponds to the critical pressure for separation onset at the appropriate Mach number. In practice, it can
be obtained from the value of CN (static) that corresponds to either the break in pitching moment or the chord
force at stall ref. [9].
2. Tv (M ) and Tvl (M );
The vortex decay constant, Tv (M ), and the center of pressure travel time constant, Tvl (M ), appear in Table 5
relatively independent of Mach number over most of the mach number range. The dynamic stall experiments
show that while there is a significant effect of airfoil shape under light stall conditions, all airfoils behave similarly
under strong dynamic stall conditions. Thus, it can be concluded that the parameters Tv and Tvl should be
relatively insensitive to airfoil shape.
M
CN1
Tv
Tvl
Df
0.30
1.45
6.0
7.0
8.0
0.40
1.2
6.0
9.0
7.75
0.50
1.05
6.0
9.0
6.2
0.60
0.92
6.0
9.0
6.0
0.70
0.68
6.0
9.0
5.9
0.75
0.5
6.0
9.0
5.5
0.80
0.18
4.0
9.0
4.0
Units
non-dim.
non-dim.
non-dim.
non-dim.
Table 5: Airfoil coefficients for dynamic stall modeling, as functions of M , for NACA 0012
22
3. Df ;
This constant models empirically the increased rate at which the chord force is lost (and hence the onset of drag
divergence) after the onset of the gross separation of flow in dynamic stall.
The general case of dynamic stall involves the formation of a vortex near the airfoil leading edge, which at certain
point, separates and convects downstream. The onset of leading edge separation for subcritical flow, or shock induced
separation for higher Mach number flow, under dynamic conditions will be initiated when CN
(t) eq. (123) exceeds
the critical CN1 (M ) boundary.
At this point there is a catastrophic loss of the leading edge suction, and the accumulated vortex lift is assumed to
start to convect over the airfoil chord ref. [15]. Then v , a non-dimensional vortex time parameter (in semi-chords), is
used to track the position of the convected vortex such that v = 0 at the onset of separation conditions, and v = Tvl
when the vortex reaches the trailing edge. It is computed from eq. (164). Experimental tests ref. [9] show that the
rate of vortex convection is less then half the free-stream velocity, with weak dependence on the Mach number. Vortex
speed can be calculated as ref. [7] Vvortex = 2V /Tvl .
v
For discretely sampled system, the vortex lift force coefficient CN
is represented by assuming that, for a given
sample period, the increment in vortex lift Cv is determined by
C
(1 KNn ).
Cvn = Ds CN
n
(144)
C
from eq. (112). Then the total accumulated vortex lift can be calculated
where KNn is from eq. (137), and CN
n
from
v
v
(145)
Ev + Ds (Cvn Cvn1 )Ev0.5
= CN
CN
n1
n
where
Ev = exp
2 S
.
Tv
(146)
Ds is a switch that takes at the beginning its value from the previous time step, and after that its value is zero, unless
any of the following conditions is met, where it equals unity, refs. [15, 7].
1. 0 < v Tvl ; or
2. |CN
| < CN1 and f < 0; or
f = fn fn1
is computed at the end of the time step to be used for the next one, in order to detect the flow
is separating, f < 0, or reattaching, i.e. f > 0 .
Alternatively, the accumulated vortex lift can be calculated from the following state space form ref. [15]
!
Cv
x11
+ Ds
.
(147)
x 11 =
Tv
Tv
v
CN
(t) = x11 .
(148)
Based on analysis of experimental data involving dynamic stall over wide range of mach number, a representation of
the center of pressure behavior (aft quarter-chord) was formulated empirically, if (0 < v <= 2Tvl ), as
v
) .
(149)
CPv = 0.25 1 cos(
Tvl
otherwise its default value is 0.
Finally, the increment in pitching moment about the quarter-chord due to the aft-moving center of pressure is given
by
v
v
CM
= CPv CN
.
(150)
From the above, the total loading can be get by superposition. The total normal force and pitching moment
coefficients under dynamic stall conditions are given by
f
v
CN (t) = CN
(t) + CN
(t).
(151)
f
v
CM (t) = CM
(t) + CM
(t).
(152)
23
After onset of gross separation, Kirchhoff modification for the chord force eq. (142) becomes invalid, and another
procedure, that ensure the continuity of its behavior, is used
p
Ccn = CN (M )2En fn
(153)
=
min(Df (|CN
|CN1 ),1)
fn
|CN
| CN1
|CN
| > CN1
(154)
where the constant Df values for different Mach numbers can be taken from Table 5.
The pressure drag coefficient, as shown in fig. 12, is then found similarly to eq.(143)
CD (t) = CN (t) sin Cc (t) cos ,
(155)
As mentioned before, the total section drag is obtained by adding the profile drag forces, as described in section 4.3.
7.2.4
Flow reattachment:
Flow is allowed to reattach from deep stall after a reasonable time delay v > 2Tvl , during the presence of the vortex.
|| 1n
1 0.3 exp [(|| 1n )/S1 ]
.
0.04 + 0.66 exp [(1n ||)/ S2 ] || > 1n
(157)
then
0.5
r
r
)Em
fM
Dfrn = Dfrn1 Em + (fM
n1
n
where
Em = exp
3 S
Tf
(158)
(159)
then
r
Dfrn .
fnr = fM
n
(160)
where
K N rn =
7.2.5
(1 +
p
fnr )2
.
4
(161)
(162)
As mentioned before, in order to solve the interaction between the different parts of the theory, the Time constants
Tf and Tv will be modified at each time step, through the values of their parameters 1 , 2 and 3 (with two different
parameters for Tf for both the force and the moment coefficients, for allowing different behavior during reattachment).
Their default values are: 1 = 1, 2 = 1 and 3 = 5. Note that at the beginning of each time step, their values
from previous time step will be used, then updated at its end.
Their values will be determined based on the flow is separating or reattaching, i.e. the sign of f , the angle of
attack is increasing or decreasing, i.e. the sign of S , and on the phase of the flow itself, as shown next refs. [8, 16, 7].
if (|CN
| < CN1 )
FlagState = 0
if (f 0)
1 = 3 = 1.
else
24
1 = 0.5, 3 = 5.
end
elseif (|CN
| CN1 )
FlagState = 1
if (f 0)
1 = 3 = 1.75.
else
1 = 1.0, 3 = 5.
if (0 < v Tvl )
1 = 0.25.
if (S > 0)
1 = 0.75.
end
end
end
end
Then the following test is performed
if (|CN
| > CN1 & f 0)
if (S < 0 or f 0.7 or f r 0.7 )
1 = 3 = 2.0.
end
end
where FlagState is used to determine the initiation of dynamic stall phase (|CN
| CN1 ), in order to compute the
non-dimensional vortex travel time v , which has default value equals to 0, then at the first instant of the onset of
the dynamic stall, its initial value is computed as (ref.( [7]))
!
|CN
| CN1
V
n
.
(163)
v = t
| |C
b
|CN
Nn1 |
n
where V is the relative flow velocity in the plane of the airfoil as given by eq. (11). After that, its accumulation during
vortex travel, as long as F lagState = 1, is computed simply from
V
.
b
Now, for the dynamic stall model, the determination of the value of 2 will be as follows
vn = vn1 + t
(164)
if (S < 0)
2 = 4.0.
elseif (Tvl < v 2Tvl )
2 = 3.0.
elseif (f > 0)
2 = 4.0 .
elseif (0 < v Tvl & S < 0)
2 = 2.0.
elseif (S < 0 & 2 = 1.0)
2 = 4.0.
elseif (S < 0 & f > 0)
2 = 1.0.
end
For the unsteady attached flow aerodynamics theories we have: Peters unsteady aerodynamics and Leishman-Beddoes.
For the Dynamic stall we have Leishman-Beddoes theory and ONERA EDLIN model. For their combination we have
the following options:
1. Attached Flow solution
25
References
[1] K.J. Bathe. Finite Element Procedures. Prentice Hall, Inc., Englewood Cliffs, New Jersey, 1996.
[2] D.A. Peters and M.J. Johnson. Finite state airloads for deformable airfoils on fixed and rotating wings. In
Proceedings of the Symposium on Aeroelasticity and Fluid/Structure Interaction. American Society of Mechanical
Engineers Winter Annual Meeting, Nov. 1994.
[3] D.A. Peters, M.A. Hsieh, and A. Torrero. A state-space airloads theory for flexible airfoils. Journal of the
American Helicopter Society, 52(4):329342, October 2007.
[4] L.A. Ahaus and D.A. Peters. Unified airloads model for morphing aifoils in dynamic stall. In Proceedings of the
AHS Specialists Conference on Aeromechanics, San Francisco, 2010.
[5] R.L. Bisplinghoff, H. Ashley, and R.L. Halfman. Aeroelasticity. Addison-Wesley Publishing Company, Reading,
Massachusetts, second edition, 1955.
[6] D. Petot. Differential equation modeling of dynamic stall. La Recherche Aerospatiale, 5:6072, 1989.
[7] W. Johnson. CAMRAD II components theory, volume ii. Technical Report Release 3.2, Johnson Aeronautics,
Palo Alto, California, June 1999.
[8] J.G. Leishman and T.S. Beddoes. A generalized model for airfoil unsteady behavior and dynamic stall using the
indicial method. In American Helicopter Society 42nd Annual Forum Proceedings, Washington, DC, June 1986.
[9] J.G. Leishman and T.S. Beddoes. A semi-empirical model for dynamic stall. Journal of the American Helicopter
Society, 34(3):317, July 1989.
[10] J.G. Leishman. Principles of Helicopter Aerodynamics. Cambridge University Press, Cambridge, second edition,
2006.
[11] J.G. Leishman. Indicial lift approximations for two-dimensional subsonic flow as obtained from oscillatory measurements. Journal of Aircraft, 30(3):340351, May-June 1993.
[12] J.G. Leishman and K.Q. Nguyen. State-space representation of unsteady airfoil behavior. AIAA Journal,
28(5):836844, May 1990.
[13] S.S. Davis and G.N. Malcolm. Experimental unsteady aerodynamics of conventional and supercritical airfoils.
Technical Report TM 81221, NASA, 1980.
[14] T.S. Beddoes. Practical computation of unsteady lift. Vertica, 8(1):5571, 1984.
[15] J.G. Leishman and G.L. Crouse. State-space model for unsteady airfoil behavior and dynamic stall. In Proceedings
of the 30th AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics and Materials Conference, Mobile,
Alabama, April 3-5, 1989, pages 13191330, 1989.
[16] Anon. RCAS theory manual, version 2.0. Technical Report USAAMCOM/AFDD TR 02-A-005, U.S. Army
Aviation and Missile Command, Moffett Field, California, June 2003.
26
M, V, a, xac, b, h
@ quarter-chord
Leishman-Beddoes Unsteady
Attached Flow
M
I
C N, C M,
C
C
C
C
C Na , C Nq, C Ma , C Mq,
Cd
Leishman-Beddoes Unsteady
Separated Flow
CN1
M, V, a,
f
f
f
C N, C M , C d
CNa, h, b
f, Df, DSa, CN
Leishman-Beddoes Unsteady
Dynamic Stall Vortex
v
N
v
M
C ,C ,Cd
Update
s1, s3
Figure 13: I/O for the complete Leishman-Beddoes Dynamic Stall Model.
27
CN