Perturbation Methods in Atmospheric Flight Mechanics: Air Force Armament Laboratory, Eglin Air Force Base, Fla
Perturbation Methods in Atmospheric Flight Mechanics: Air Force Armament Laboratory, Eglin Air Force Base, Fla
Perturbation Methods in Atmospheric Flight Mechanics: Air Force Armament Laboratory, Eglin Air Force Base, Fla
9, SEPTEMBER 1973
AIAA JOURNAL
1247
Downloaded by CARLETON UNIVERSITY LIBRARY on March 14, 2015 | http://arc.aiaa.org | DOI: 10.2514/3.50574
The classical scalar perturbations of flight mechanics are compared with total vector perturbations. A component
perturbation method is introduced which is the generalization of the classical technique, and is valid for all
coordinate systems. This method, together with a new tensor formalism, is used to derive the general perturbation
equations of atmospheric flight mechanics. These equations hold for any unsteady flight regime and are expressed
in a tensor form, invariant under time-dependent coordinate transformations. The perturbation equations of all
flight vehicles such as aircraft, missiles, shells, and Magnus rotors, are universally represented by these techniques.
Introduction
H
and for a second-order tensor (tensor)
A
(2)
P. H. ZIPFEL
Downloaded by CARLETON UNIVERSITY LIBRARY on March 14, 2015 | http://arc.aiaa.org | DOI: 10.2514/3.50574
1248
[Q"]=t^K^][K**]
AIAA JOURNAL
(8)
Perturbation Techniques
The classical perturbation technique of flight dynamics, as for
instance outlined by Etkin,2 proceeds as follows. First, an axis
system is defined in relationship to physical quantities, such as
the principal body axes or the relative wind velocity. The components of the state variables parallel to these axes are then
identified. A particular steady flight regime is selected with
certain values for the components, e.g., x rl , xr2, xr3, and a
perturbed flight with x pl , xp2, xp3. The differences Axf =
xpi - xri are the perturbation variables. Because the perturbations
are generated by a scalar subtraction of the respective components this technique is also called the scalar perturbation
method (see Hopkin3).
The disadvantage of the classical perturbation technique lies
in the fact that all the formulations are tied to one particular
coordinate system. A change to other coordinate systems is very
difficult to accomplish.
In theoretical work, vectors are preferred over components
and perturbations are defined as the vectorial differences between
the reference and perturbed vectors. No allusion is made to a
particular coordinate system. Because this technique considers
the total state variable rather than its components it is called the
total perturbation method. For [xp] and [xr], the state vectors
during the perturbed and reference flight, respectively, the total
perturbation is defined as
[<5x] = [xp]-[xr]
(12)
The total perturbations have the advantage over the scalar
perturbations in that they hold for any coordinate system. However, in applications, numerical calculations require that vectors
are expressed by their components, referred to a particular
coordinate system. For instance, vehicle moment-of-inertia is
given in body axes, vehicle acceleration and angular velocity
are measured by accelerometers and rate gyroscopes, mounted
parallel to the body axes; wind-tunnel measurements are
recorded in component form, and the whole framework of aerodynamics is based on force and moment components rather than
on total values.
To express the total perturbations in components, a transformation matrix must be introduced. In the notation of the
previous section, the components of the [dx] perturbation,
d5)
where the last equality follows from Eq. (11). The transformation matrix of Eq. (13) is absent. Because this technique
emphasizes the component form of a vector, it is referred to as
the component perturbation method or alternately as the perturbations.
In working with the component perturbation method, the
choice of the \_RDpDr~\ tensor, and thus the selection of the
frame D, is most important. As a general guideline, D is chosen
so that the s-perturbations remain small throughout the flight.
Particularly, in atmospheric flight mechanics the selection of D
is determined by the requirement of representing the aerodynamic forces as a function of small perturbations. A Taylor
series expansion then is possible, and the difficult task of expressing the aerodynamic forces in simple analytical form can be
achieved. The designation "dynamic frame" will be used for D,
because the dynamic equations of flight mechanics are solved in
a coordinate system associated with D.
The dynamic frame of airplane dynamics is either the body
frame B or the stability frame S. For the definitions of the
frames used in this paper, see Etkin.2 In both cases, for small
disturbances, the rotation tensors are close to the unit tensor,
expressing the fact that the frame Dp has been rotated by
small angles from Dr. For instance, in the simple case of small
perturbations about a horizontal steady reference flight, Dr is an
inertial frame itself and Dp deviates by the small angles \jj, 9, $
from D/y with ar included if D is the stability frame. As will
be outlined in more detail in the last section, the motions of
the dynamic frame relative to the wind frame are directly
related to the state variables used in the aerodynamic force
expansion.
In missile dynamics, the situation is similar except that the
aeroballistic frame replaces the stability frame as dynamic frame.
However, for a spinning missile, the body frame cannot serve
as a dynamic frame because the perturbations of the aerodynamic
roll angle can be large. To keep the perturbations small between
the wind and dynamic frame, the nonrolling body frame is chosen
as dynamic frame. The motions between the body frame and the
dynamic frame thus are not explicitly included in the aerodynamic expansion, but rather the expansion derivatives depend
on them implicitly. For simplification, \_RDpDr~\ will be abbreviated by \_R] unless ambiguities arise.
SEPTEMBER 1973
d6)
Expanding about the reference flight, i.e., for [ex] = [0] , yields
][ex]+(17)
4
The principle of material indifference states (see Noll) that
the physical process of generating aerodynamic forces is independent of spacial attitude. In other words, if [xr] is rotated
through \_R] the process or functional dependence is still the
same. The only difference being that the force has also been
rotated through \_R] , i.e.,
Downloaded by CARLETON UNIVERSITY LIBRARY on March 14, 2015 | http://arc.aiaa.org | DOI: 10.2514/3.50574
1249
(18)
['BB'] = [V][y"]
(22)
(24)
(25)
The equations of motion of atmospheric flight mechanics
assume the invariant structure
[V"] = [/J + [/,] + [/]
[V'] = KJ+K]
(26)
(27)
Where [/] represents the forces and [m] the moments relative
to the center-of-mass B, and the subscripts a, t, and g refer to
aerodynamics, thrust, and gravity forces, respectively. Both
Eqs. (26) and (27) are valid for the reference and perturbed
flights. To derive the linear momentum perturbation equation,
let Eq. (26) describe the perturbed flight
[y|] = [/j+[/(P]+[/j
(28)
(29)
] + M + Dsfl + M
(3D
The underlined terms are Eq. (26) for the reference flight, rotated
by [R]. They are satisfied identically. The gravitational term
can be rewritten using the fact that \_fgp\ = [/J
[fl = PO -[])[/,,]
(32)
Where [] is the unit tensor.
The perturbation equation of the angular momentum is
derived in the same way. Both equations are summarized as
follows :
(33)
(34)
These are the general perturbation equations of atmospheric
flight mechanics. No small perturbation assumptions have been
made as yet. They are expressed in an invariant form, i.e.,
they hold for all coordinate systems related by time-dependent
coordinate transformations. Two types of variables appear : first,
[pBr/] [^Br/]' tne linear and angular momenta of the reference
flight, known as a function of time; second, the component
perturbations, marked by a preceding e. Latter represent the
unknowns. The aerodynamic forces and moments will be the
subject of the following section. Evaluating the perturbational
thrust and gravity forces is straightforward and will not be
discussed.
The first terms on the left-hand side of Eqs. (33) and (34) are
the time-rate-of-change of linear and angular momenta, respectively; whereas the second terms account for unsteady reference flights. To gain insight into the structure of the
perturbation equations two special cases will be discussed.
Case I considers the perturbation equations of aircraft and
missiles in steady reference flight. Steady means that the reference
flight is unaccelerated and nonrotating. The body frame B is
chosen as dynamic frame D. During the reference flight the
body frame Br is an inertial frame /. With these substitutions
and the fact that [pBrBr~] = [0], [lBrBr~] = [0], the left-hand sides
of the perturbation equations become simply : \_@Br&pBBr~] and
OBr/BBr], where
(35)
= m[vBp]
[QIelBBr] +
] = [onJ + [onj
P. H. ZIPFEL
1250
Expressed in the ~]Bp coordinate system Eqs. (37) and (38) become
p
BpBr p BrI r
+[n
Y lp Y =
I = W-MM
(45)
(47)
Aerodynamic Forces
Downloaded by CARLETON UNIVERSITY LIBRARY on March 14, 2015 | http://arc.aiaa.org | DOI: 10.2514/3.50574
AIAA JOURNAL
(43)
The first group, separated by a semicolon, relates the interactions
between the body and dynamic frames. Its variables are implicitly
included in the expansion. In the case of a spinning missile,
e.g., the aerodynamic roll angle and its time derivative belong
to this group. For many applications, the method of averaging
by Bogoliubov and Mitropolsky5 can be employed for eliminating this type of dependence.
The second group includes the variables, whose s-perturbations are to be expanded, for instance
[scown] = [wwp?p] - [R] \wWrDf]
(44)
(48)
(50)
[Bri]Bp+'~
(51)
If this equation is normalized, and if p and pp can be approximated by pr and pr, the first three implicit variables are replaced
by the Mach and Reynolds numbers of the reference flight. Also
[wrB]Br' \yBYr> and [wrB]Br are the velocity components of the
reference flight, commonly denoted by ur, vr, and vv r ; [coBrAYr
represents the reference rotations pr, qr, and r r ; and [cow>Br]Br
contains the time-rate-of-change of the aerodynamic angles a r ,
f$r. For the simple case of the unaccelerated and nonrotating
reference flight, the partial derivatives become constants.
Equation (50) and a similar equation for the aerodynamic
moment perturbation, substituted into the linear and angular
moment Eqs. (33) and (34), constitute the general perturbation
equations of atmospheric flight mechanics.
Conclusions
The theory of perturbations, as presented in this report, is a
new approach to the difficult problem of mathematically
modeling the dynamics of atmospheric flight vehicles. It unifies
the methods employed for different types of vehicles and provides
greater insight into the physical aspects of their dynamics.
Because the equations are invariant under time-dependent coordinate transformations, they constitute the most general
formulation of the perturbation equations of atmospheric flight
mechanics. Furthermore, deriving the equations in the compact
tensor form saves time and avoids errors. Only the final result
need be written in component form for numerical calculations.
The new techniques developed in this paper have already been
successfully applied to the flight dynamics of Magnus rotors
(see Zipfel1). These provided the mathematical framework necessary to solve the unsteady lateral stability problem of Magnus
rotors in accelerated flight that cannot be solved by the classical
SEPTEMBER 1973
References
1
AIAA JOURNAL
Downloaded by CARLETON UNIVERSITY LIBRARY on March 14, 2015 | http://arc.aiaa.org | DOI: 10.2514/3.50574
SEPTEMBER 1973
1251
Introduction