Solving the Equations of Motion

Chapter 5

Lockheed C-5 (AFFTC History oflice photo)

Flying Qualities Testing 5-i

Solving the Equations of Morion
Chapter 5 Contents

Solving the Equations of Motion .. ...

5.1 Selecting an Airplane and Equations ................................. 1

5.2 Signconvention .............................................. 3

5.3 Solving the Longitudinal Equations of Motion . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

5.3.1 Longitudinal Digital Simulator Solution .................... ~ ... 14
5.3.2 Measuring the Longitudinal Modal Parameters .................... 15
5.3.3 The Phugoid Mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
5.3.4 The Short Period Mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
5.3.5 Solving the Two Degree of Freedom Longitudinal Equations of Motion . . . . 20
5.3.6 A Footnote on the Phugoid Mode and the Equations of Motion . . . . . . . . . . 24

5.4 Solving the Lateral-Directional Equations of Motion ...................... 25

5.4.1 The Dutch Roll Mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
5.4.2 The Roll Subsidence Mode . . . . . . . . . . . . . . . . . . . . . . . . . '. . . . . . . 29
5.4.3 The Spiral Mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
5.4.4 Completing the Lateral-Directional Solution ..................... 30
5.4.5 Lateral-Directional Digital Simulator Solution . . . . . . . . . . . . . . . . . . . . 31
5.4.6 Measuring the Lateral-Directional Modal Parameters . . . . . . . . . . . . . . . . 33
5.4.7 Concluding Note . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

5.5 Waypoint: Solving the Equations of Motion ........................... 37

54 Flying Qualities Testing

 Chapter 5

Solving the Equations of Motion

The equations of airplane. motion we developed in Chapter 4 enable us to penetrate, analytically, the
origins of airplane motion, stability, and control, and to build piloted and unpiloted simulators. This
insight can spell the difference between successful and unsuccessful flight control system design and
flying qualities test programs. Most often, the insights we reap from the equations of motion do not
require solving them. Instead, these insights are usually gathered from the unsolved equations of
motion, written in the s domain. For this reason, we ordinarily have little interest in solving the
differential equations of airplane motion. When a solution is necessary, we commonly avoid the
analytical drudgery by turning to a simulator. Nevertheless, in this chapter we will solve the
equations of motion so that we may illustrate several important points regarding the characteristic
modes of airplane motion. We will follow a three step procedure. First, we will solve the equations
of motion analytically and examine the characteristic modes of airplane motion revealed by the
solution. Next, we will simulate the airplane motion, solving the same equations of motion
numerically, using a digital computer, just as is done in flight test simulators. Finally, we will
compare the analytical and simulator solutions and draw some conclusions about simulators.

5.1 Selecting an Airplane and Equstions We will choose as our subject for solving the equations
of motion an F-16AB flying at 1.2 Mach number at 31,000 feet. The F-16 has a flight control
system that cannot be turned off. But we have not yet developed the tools needed to handle a flight
control system, so we will look only at the natural, or unaugmented, airframe motion (often referred
to as the bare airframe dynamics). When we say "unaugmented" we mean that no help is provided
by a flight control system or auto-pilot. The bare airframe, or unaugmented motion is the motion
that would be exhibited if the flight control system were turned off.

Flying Qualities Testing 5-1

Solring the Equations of Motion
5.1 Selecting a n Airplane and Equations

.. k
The F-16 is longitudinally unstable over much of its flight envelope. However, at the flightcondition
we have selected, 1.2 Mach number and 31,000 feet, the airplane is naturally stable, so we will be
able to identify the classical modes of airplane motion. Later, when our toolbox includes suitable
control system analysis techniques, we will look at unstable airframe dynamics.

Next, we must identify the equations of motion we will use. Because we can solve only linear
differential equations analytically, we must choose from Equations (4-103), (4-104), and (4-105),
which we developed in section 4.8 in Chapter 4. Equations (4-103) are the linearized, small
perturbation equations of motion. These equations are linear, but complex and more general than we
need. If we trim the airplane in a turn and use these equations, the longitudinal and lateraldirectional
modes of motion will couple, meaning that pitching motion will produce lateraldirectional motion
and vice versa. This makes it more difficult to separate and identify the modes of motion in time
histories of the airplane response to control inputs.

Equations (4-104) are the decoupled, linearized, small perturbation equations of motion. These
equations are l&ar and almost as uncomplicated as we can make them, but they restzict us to straight
and level trim conditions. In practice this is not a burdensome restriction, and it serves the very
useful purpose of decoupling the longitudinal and lateral-directional modes of motion. In the service
of analytical insight, Equations (4-104) are tried and proven.

Equations (4-105) are the five degree of freedom equations of motion that result when the X, axis
speed U is constrained to be constant (that is, when rl=O). When acceleration along the X, axis is
prohibited, the long-term, or phugoid mode of longitudinal motion disappears. This result is
frequently useful, but for now we do not want to exclude the phugoid mode.

Since we wish to learn as much as we can for the least amount of work, we will use Equations (4-
104), the decoupled, linearized, small perturbation equations of motion. After we use these equations
to identify the phugoid and short period modes of longitudinal motion, we will demonstrate that the
phugoid mode does indeed disappear when we use Equations (4-105).

A nice feature of the decoupled, small perturbation equations of motion, Equations (4-104), is that
they allow us to separate the longitudinal and lateral-directional equations of motion. If you look
back at Equations (4-104) in section 4.8.3 in Chapter 4, you will see that no lateraldirectional motion
variables appear in the J , & , and 4 equations, and that no longitudinal motion variables appear in
the , p , and f equations. This reduces our workload substantially, allowing us to separate an
eighth order characteristic polynomial into two fourth order characteristic polynomials.

We will take full advantage of this decoupling feature and solve first the longitudinal equations of
motion, then the lateraldirectional equations of motion.

Before we embark on solving the equations of motion, it will save many headaches if we first update
and summarize the sign convention we will use. (As you know, many errors are traceable to sign
convention and units.) You may find it helpful to refer to this sign convention summary from time

5-2 Flying Qualities Testing

 Solving the Equations of Motion
5.2 Sign Convention

to time as we solve the equations of motion and as you progress through your study of flying quaiti&
testing.

5.2 Sign Convention We have already assembled a flying qualities sign convention for forces,
moments, angles, and angular rates that is rather widely used. But we have not yet established a sign
convention for control surface deflections and controller deflections. By "controller" we mean the
control stick or wheel, rudder pedals, and any other device used by the pifot tocontrol the airplane.

The Air Force Flight Test Center cannot impose an immutable flying qualities sign convention. It
is inevitable that different companies will use different conventions. On one test program the
contractor defined positive aileron to be that deflection which rolls the right wing down, while on
another test program the contractor defined positive aileron in exactly the opposite way. Neither
convention is wrong, so long as the signs of C,,., C, , and C accord with the convention being
8. a'.
used. I

What is important to you, as flight test aircrew and engineers, is to know and understand the
controller and control surface sign convention that is used in your test program.

The control surface sign convention we will use during our study of flying qualities testing is widely
recognized and fairly widely used. This convention is presented in Table 5-1 and illustrated in Figure
5-1.

Surface Symbol Sign Direction Result

elevator iie - trailingedgeup +AM. + A a *

aileron iia + ... +AL,+ A +
rudder 6, + trailiagedgeleft -AN, +AP

* where. A denotes incremental change

Table 5-1 Sign conventionfor control sur$ace deflections.

We did not define a direction for aileron deflection because there are two ailerons and they move in
opposite directions. Our convention will be to define trailing edge down as positive deflection for
each aileron. We define total aileron deflection 6a as the following combination of individual aileron
deflections:

Flying Qualities Testing 5-3

Solving the Equations of Motion
5.2 Sign Convention

where is right aileron deflection and is left aileron deflection.

When pitch control is provided by independent elevators (or stabilators) that may deflect either
together to produce pitching motion or differentially to produce rolling motion, we define elevator
deflection ae as the following combination of individual elevator (or stabilator) deflections:

We have not yet considered control inputs by the pilot, but we will establish a sign convention now,
in anticipation of the need. The controller sign convention we will adopt is widely recognized and
almost universally used. This convention is presented in Table 5-2 and illustrated in Figure 5-1. The
subscript cs in F, and 8 , denotes "elevator stick." The subscript (IS in F, and &- denotes "aileron
stick." The subscript ip in Fn and aV denotes "rudder pedal.''

I Control

pitch stick force

Symbol

Fea
S i

+
Direction

aft
Reslln

+AM. *Aa
*
pitch stick deflection 8- + aft + A M , *Aa
lateral d c k / w h l force F, + +A, +Ab

lateral dck/wheel deflection 8- + right +A,

+A+

rudder pedal force FrP + right +d,

-AB

radder pedal weaim hrp + right +d,

-AB

We will define total rudder pedal force and deflection as the following combinations of right and left
forces and deflections:
F V = PV I -FB,

where positive force and deflection are defined to be in the forward direction for both pedals.

Note that the control surface deflections shown in Figure 5-1 are produced by the positive controller
(stick and rudder pedal) forces and deflections shown. It is worth your time now and again to review
the sign convention summarized in Tables 5-1 and 5-2 and in Figure 5-1. It will keep you out of

5-4 Flying Qualities Testing

 Solving the Equations of Motion
5.2 Sign Convention

Figure 5-1 Flying qualifiespight rest sign convention.

Flying Qualities Testing 5-5

Solving the Equah'ons of Motion
5.3 Solving the Longitudinal Equations of Motion

trouble while you are a student at the Test Pilot School and, for the most part, will stand you in good - 1

stead during your work in flying qualities testing.

5.3 Solving the Longitudinal Eqnations of Motion We begin by extracting the decoupled,
linearized, small perturbation, longitudinal equations from Equations (4-104), in section 4.8.3, and
renumbering them as follows:

(5-la)

(5-lb)

Because these equations contain three degrees of freedom, they are often referred to as the
longitudinal three degree offreedom equations of motion. The three degrees of freedom are linear
(or translational) motion along the Xb axis ( u ) , linear (or translational) motion along the 2, axis ( w .
or a ) , and rotational motion about the Yb axis ( 4 . or e).

There are four motion variables in Equations (5-1), u , a , q , and 8 , and only three equations of
motion. Either we must augment Equations (5-1) with a fourth equation or we must eliminate one
of the motion variables. The customary practice is to adopt the latter course of action because it
reduces the number of motion variables and the number of equations to three, which simplifies the
solution. Adding a fourth equation would require us to evaluate fourth order determinants. Third
order determinants require less work.

Because we have restricted ourselves to trimmed straight and level flight, q = 8 . You may satisfy
yourself that this is so by reviewing Equations (4-61) in section 4.5.6 in Chapter 4, for the case of
wings-level flight, when 41=O degrees. Hence we may substitute 6 for q and e for 4 in Equations
(5-1). This reduces the number of motion variables, and the number of required equations, to three.
The motion, or state variables are now u , a , and 0 .

You may recall from Chapter 3 that the Laplace transform technique is a convenient way to solve a
system of linear, constant coefficient, differential equations. Since the airplane is trimmed, the initial
conditions of the perturbation variables are zero. Performing the Laplace transform of Equations (5-
1) we obtain

5-6 Flying Qualities Testing

 Solving the Equations of Motio+#
5.3 Solving the Longitudinal Equations of Motion

where, to mini& clutter, it is understood that u , a , 8 , and ae are functions of S: e.g., u(s),
a@), and so on. Note that we have substituted d for q and s28 for 4 .

In practice, it is difficult to separate Cm, from C based on flight test data. For this reason, C*, is
a*
usually combined with C,
* . In effect, this means that we combine the two terms Cm4&+C q into a
single term C%.,)q. However, to preserve notational compactness, we usually write C q with the
5
understanding that C,,+,,q is implied. Combining Cm, and C in this manner and rearranging, we
'I
recast Equations (5-2) into the following form: %

Putting these equations into a matrix format we have

Examining Equation (5-3), we see that the bare airframe dynamics are determined entirely by the
flight condition (4, V,, ao, eo), airplane physical properties ( I n , m , c , S), and aerodynamics, or
stability derivatives (which also vary with flight condition, as we will see in Chapter 7). Of these,
the most influential players are the stability derivatives. This is why, in the past, when flight control
systems were relatively primitive, stability derivatives were often treated as classified information.

Flying Qualities Testing 5-7

Solving the Equations of Motion
5.3 Solving the Longitudinal Equations of Motion

..
Stability derivatives, together with rough estimates of airplane physical properties, could tell an
analyst a great deal about performance and flying qualities. Today, this is not as true as it once was.
Modern flight control systems can turn seemingly intractable aerodynamics into a flyable airplane.
Without the flight control system, there is little an analyst can conclude about flying qualities from
the aerodynamics alone. Airplane performance is still a direct function of aerodynamics, but the
aerodynamics of modem airplanes are so nonlinear that performance characteristics in the middle of
the flight envelope can no longer be reliably extrapolated to the edges of the envelope.

We will solve Equation (5-3)for the response of pitch attitude angle e to an elevator input, 6,. From
your review of differential equations in Chapter 3, recall that Cramer’s rule is a convenient way to
do this. Applying Cramer’s rule to Equation (5-3). we obtain

(5 -4)

All we need now are a flight condition, airplane physical properties, and stability derivatives. In
Table 5-3 we provide these data for an F-l6A/B flying at Mach 1.2 and 31,000 feet. In the interest
of full disclosure, we must note that because F-l6A/B speed derivatives were not available, we used
Douglas A 4 values for CN. and Cma. We estimated the speed derivative Ccmbased on available F-
16AB data. Note that Cm, and C are zero. These derivatives are difficult to obtain from flight
%
testing: C is usually negligibly small and Cm, is nearly impossible to separate from Cm,. As we
6%

noted earlier, flight test measurements of C

5
are treated as the combination derivative CmQ..,.

5-8 Flying Qualities Testing

 Solving the Equations of Motioh
5.3 Solving the Longitudinal Equations of Motion

Mach number, M 1.2

Altitude, h 31,000 A
-airspeed.y, 1188 ft/sec (704hots)
calibrated airspeed. v, 469 knot3
Dynamic P==u=> 5 605 lbSlft2
Trim angle of attack, arc 2.24 deg (0.0391 rad)
Trim pitch attitude angle, 0, 2.24 deg (0.0391 rad) -
Trim elevator ddlection, 6, -2.63 deg (-0.0459 tad)
Wiw3area.S mft2
Mean aerodynamic chord, e 11.32 ft
Whgspau,b 3oft
center of gravity, cg. 35%
Wei& w m,000 Ibs
Mass,
Z, = 9,850 --it2
622 slugs

In = 51,605 slug-fl
.
Z, = 59,170 slug-ft2 I, = 345 slug-ft2

Coo = 0.0437 Cn. = -O.OlSO/dCp C5 = -O.OSaO/rad

C0. = 0.0450 = '-0.859/rad ca, = -o.soO/rad
= O.oOll30/deg = -3.80/nd Cs. = 0.oooS50/deg
cc. c-,
= 0.246/rad c=a, = -o.o1m/deg = O.O544/d

c., = 0 = -O.aSS/rad c., = -O.ocQ8Oo/&g

CNo .= -0.0500 = -O.rnlO/deg = -0.0458/rad

CN* = 1.24 = -1.2O/tad CI, = -0.00160/deg

CN. = O . W / d e g c yo.
= -O.oozoo/&g = -0.0917/rad
= 4.81/rad = -0.115/rad C$ = -O.soO/rad
CN,. = O.O106/deg c% = o.m/deg c4 = o.os00/rsd
= o.m/rad = O.IlS/rsd c1. = O.o0100/deg
c- = 0.00200
cn,
= O.o043o/deg = O.O573/rad
C- = -0.0566 = o.z46/nd = o.m/deg
Cf.,
= O.O172/rad

Table 5-3. Trim conditiom, aircraftphysicalpropenies, and sfabitifyderivativesfor an F-lM/B af 1.2 Mach number
and 31,Wfeef.

Flying Qualities Testing 5-9

Solving the Equations of Motion
5.3 Solving the Longitudinal Equations of Motion

I .

Note also in Table 5-3 that the values of all but the speed derivatives and rotational rate derivatives
are given in both degree and radian measure. In flight testing we use degree measure, but when we
solve the equations of motion we must use radian measure because the equations are written in radian
measure.

Substituting the airplane physical properties, flight conditions, and stability derivatives from Table
5-3 into Equation (5-4), we obtain
s + .M21

e(s) =
-27.4
s + ,0221 71.8 46.4s + 322

.000513 s + 1.18 -s + .Gal06

a0379 33.9 s(s + .721) .

Evaluating the numerator and denominator determinants, we get
-27.4(sZ + 1.02s - .0133)
e(s) = a,@) (5 -5)
s4 + 1.93s' + 3 4 . 9 +
~ 9~2 5 s + .420
The ratio of polynomials in Equation (5-5) is called a transferfuncrion. The transfer function in
e
Equation (5-5) is denoted -(s), and describes the pitch attitude motion of the airplane in response
a*
to an elevator input. You will become more familiar with transfer functions when we discuss flight
control system in Part III.

e
The numerator of the -(s) transfer function is often denoted N: , while the denominator is often
ae

denoted by the symbol Ah, where the subscript long denotes "longitudinal. " Hence you might see
e
e
-(s) represented by 3.
In a similar manner, the roll rate to aileron deflection transfer function,
ae Areas
N,'
E @ )might
, be represented by 2 ,where the subscript Id denotes "lateral-directional. "
aa AM

The denominator polynomial, Ah, in Equation (5-5) is called the characteristic polynomial. It gets
its name from the fact that the factors of this polynomial characterize, or describe, the longitudinal
modes of motion of the airplane. In this case, the characteristic polynomial factors into two quadratic
terms, which are the phugoid and short period modes of longitudinal airplane motion.

5-10 Flying Qualilies Testing

 Solving the Equations of Motion
5.3 Solving the Longitudinal Equations of Motion

Abq = s 4 + 1 . 9 3 +~34.9s'
~ + 925.9 + .420

= (s2 + .oL59s + .0121)(s2+ 1.90s + 34.8)

The first quadratic factor is the phugoid, or long-term mode of longitudinal motion. This factor has
the general form
A p = s 2 + 2 C 0 s + w q2 p
p "P

where the subscript p denotes the phugoid mode. We refer to the phugoid damping ratio CP and the
phugoid natural frequency o as modal parameters because they completely characterize the phugoid
*>
modal response. Using this generalized modal form we may determine the natural frequency and
damping ratio of the phugoid mode in the following way:
Ap = s2 + 0259s + ,0121

- w
%
= @EZ= ,110 radians/second
2Cpwnp= ,0259

- c=:- 0259
20
"r
,0259 - ,118
2(.110)

After we have completed our analytical solution of the equations of motion, we will solve them with
a simulator and measure CP and GI from simulator time histories. We will compare the measured
"P

values of Cp and w with the analytically derived values we have just calculated and see what
%
conclusions we may draw.

The second quadratic factor in the characteristic polynomial of Equation (5-5) is the short period, or
short-term, mode of longitudinal motion. Using the procedure outlined above, we may determine
that the short period natural frequency and damping ratio are ons=5.9o radians per second and
Cv=0.161. The subscript sp denotes "short period." Later, we will measure ,C and o
"P
from
simulator time histories and compare the results with these analytically derived values.

Now that we have identified the phugoid and short period modal parameters, CP, o ,C and o
"P ' nP '
e
we may express the -(s) transfer function given in Equation (5-5) in the shorthand notation we
&e

introduced in section 3.2.3 in Chapter 3. This gives us

e
-(s) =
-27.4(-.0129)( 1.03) radians
6< [.118,.110][.161,5.9o] radian

We should pause here to note that the longitudinal characteristic polynomial doesn't always factor into
two quadratic terms. Ocassionally, one of the second order terms will factor into two first order

Flying Qualities Testing 5-11

Solving the Equations of Motion
5.3 Solving the Longitudinal Equations of Motion

terms, or modes of motion. For example, when the short period mode is unstable, there will be one
stable first order term and one unstable first order term. The X-29 had two first order short period
terms, or modes of motion: one stable and the other unstable.

The next step in our pursuit of a solution is to select an input, or forcing function, ae(s), Impulse
functions are good analytical forcing functions because they highlight the characteristic, or modal,
responses of a system and because they are easy to work with. For airplanes, impulse functions are
doubly useful because they are unlikely to change the trim condition of the airplane very much. But
while impulses are analytically useful, they are physically unrealizable, so in a real airplane or a
simulator we can only approximate them. Sharp doublets offer some of the advantages of impulse
functions, and they are physically realizable, but they are cumbersome to use analytically. On
balance, it will pay us to use an impulse forcing function for our analytical solution.

We will adopt as our forcing function a one degree, or ,0175 radian trailing edge up elevator
impulse. Recall from Table 5-1 that a trailing edge up elevator deflection is negative: hence our
elevator deflection will be -.0175 radians. The Laplace transform of an impulse function is simply
the magnitude of the impulse. For a -.0175 radian elevator impulse, Equation (5-5) becomes
-27.4(s2 + 1.02s -0.0133)
e(s) =
(s’ + .0259s + ,0121) (s’ + 1.90s + 34.8)

- -27.4(s2 + 1.02s -0.0133) (-,0175)

(s’ + ,0259s + ,0121) (s2+ 1.90s + 34.8)

.478(s2 + 1.02s -0.0133)

e(s) =
(s’ + .0259s + ,0121) (sz + 1.90s + 34.8)

We wish now to solve this equation for e . As we noted in the opening paragraph of this chapter,
and also in Chapter 3, in flying qualities testing we are ordinarily not very interested in solutions of
the equations of motion. When a solution is necessary, we usually turn to a simulator. Because we
have no great practical interest in solving the equations of motion, we will not engage in the
analytical drudgery attendant on a Heaviside partial fraction expansion of Equation (5-6), followed
by an inverse Laplace transformation into the time domain. Instead, we will present the results.
Equation (5-6) may be expanded and rearranged to give
. 0 1 3 7 ( ~- ,0249) - .0137(s - 33.1)
e(s) = (5 -7)
(s + .0129)’ + .1@ (s + .950)’ + 5.82’

The following Laplace transform pair may he used to inverse transform each of the terms on the right
hand side of Equation (5-7) into the time domain.

5-12 Flying Qualities Testing

 Solving the Eqwrions of Motim
5.3 Solving the Longitudinal Equations of Motiod

.* 2

(5-8)
whae $l-tau-n-'--b
a -a

Because we are working in radian measure, the phase angle 4 in this transform pair is measured in
radians. Using this transform pair, we find that the inverse Laplace transform of Equation (5-7) is
e(?) = .0145e-~01"'sin(.109t+ 1.90) - .0813e-.m'sin(5.82t ~ 2 . 9 7 )mmhns (5 -9)
To express 6 in degree measure we multiply this result by 180/r, which gives us
-
6(t) = .831e-"'29'sin(.109r+ 1.90) 4 . 6 6 ~
- . m r ~ 5 . 8 2+t2.97) okgrees (5-10)
Equation (5-10) is the pitch attitude time response of an unaugmented F-l6A/B (that is, an F-16AB
w'thout a fright corurol
sysreqz) when it is excited by a
-1 degree (or a -.0175 radian)
elevator impulse at Mach 1.2
and 31,ooOfeet. Solutionsfor a
and u may be obtained by
following the same procedure.

Short-term and long-term time

histories of Equation (5-10)
are presented in Figures 5-2
and 5-3. Note the two very
-,- distinct modes of motion. The
Figure 5-2 A~lyticoNyderived short prriad response ofpitch anitude angle 9
to a -I degree elevator impulse. phugoid is a very slow, long-
period mode of motion that is
lightly damped. The short
period is a much faster mode
of motion, and also lightly
damped in this case.

The short period and phugoid

motion shown in Figures 5-2
and 5-3 is not at all evident to
the pilot of an F-16. This is
because the flight control
system modifies these lightly
o io a0 80 u) M) BO TO BO m iw
damped bare airframe modes
Time, BBcouLd.
Figure 5-3 AMbticaiiy derivedphugoid response of pitch arnkie angle 9 io a into a more desirable set of
-I degree elevator impulse. responses. For example, at
the trim condition we selected,

Flying Qualities Testing 5-13

Solving the Equah'ons of Motion
5.3.1 Longitudinal Digital Simulator Solution

.. ..
the F-16AB flight control system increases the short period damping ratio from 0.161 to 0.9. When
the short period damping ratio is 0.9, the pitch attitude response does not oscillate at all. We will
see how a flight control system may be used to modify the short period damping ratio when we study
flight control systems in Part III.

5.3.1 Lougitudiual Digital Simulator Solution Equation (5-10) and the time histories in Figures
5-2 and 5-3 are an exact solution of Equations (5-1). In simulation, the solutions are never exact
because numerical integration must be used instead of analytical integration. (The analytical
integration we performed was embedded in the Laplace transform techniques we used to solve the
equations of motion.) Numerical integration can give results that range from crude to excellent, hut
it can never exactly match analytical integration for the equations we are working with.

To illustrate the kind of results we may

expect to get from a simulator, we built a
digital simulator that solves Equations (5-1).
The Fortran code for this simulator is
presented in Appendix 5A. We used a fixed-
step, fourth order, Runge-Kutta numerical
integrator running at 100 samples per second.
This integrator is often used in real-time
.piloted simulators, such as those used at the
Air Force Flight Test Center. Because a
simulator cannot duplicate an analytical
impulse, we used a one second long, f 2
degree, triangular elevator doublet to excite
the airplane response. The resulting short
period and phugoid time histories of 6, a ,
and u are presented in Figures 5 4 and 5-5.
We have added a time history of flight path
angle y , which is defined as y = 0 -a when an
Figure 5-4 Simulator shon period response of u. a,e. ond airplane is in wings-level flight. When the
7, to o f2 degree niangulor elevator doublet. Simuhor wings are level, the flight path angle is the
s w l e rare is 100 somples per second.
angle of the velocity vector relative to the
horizon.

We should note that we are putting simulation techniques to a difficult test with the dynamics of an
unaugmented F-16 at 1.2 Mach number at 31,000 feet. Lightly damped modes, such as the short
period and phugoid modes we are working with, are difficult to simulate accurately. We will briefly
explain why this is so in a moment, and explain again more fully in Part 111, when we study flight
control systems.

5-14 Flying Qualilies Testing

 Solving the Equations of Motion
5.3.2 Measuring the Longitudinal Modal Parameters

5.3.2 Mearmring the Longitudinal Modal Parameters Bearing in mind that we used an impulse
forcing function for our analytical solution and a triangular doublet in the simulator, the simulator
responses presented in Figures 5 4 and 5-5 appear reasonably similar to the analytical responses
presented in Figures 5-2 and 5-3. To check on appearances, let's use the simulator time histories to
measure the modal parameters CP, on,, (, ,and osr. To do this we will isolate the simulator time
histories of e, repeating them in Figures 5-6 and 5-7.

We begin with the simulator short period response shown in Figure 5-6. We will use the log
decrement method to calculate ,( from the time history measurements. The log decrement method
is explained in the shaded box in this section. Marking and measuring the envelope half-amplitudes
+, and, two cycles of motion later, 3,we get +,= 1.7 degrees and +=.Ndegrees. For n=2 cycles,
we use the log decrement method to calculate (,=.14. The elapsed time between +, and 3 is about
2.15 seconds, so the damped short period frequency is two cycles divided by 2.15 seconds, or 0.93
cycles per second, which is 5.8 radians per second. The short period natural frequency is
w
0 =- dm , or 5.9 radians/second. These results are in reasonably good agreement with the

analytically derived values of (,=.I61 and o"r-5.90 radiadsecond.

Flying Qualities Testing 5-15

Solving the Equalions of Morion
5.3.2 Measuring the Longitudinal Modal Parameters

. ...

5-16 Flying Qualities Testing

 7
Solving the Equafions of Motion
5.3.2 Measuring the Longitudinal Modal F’iuameters

- ---- -
z
i
we may make similar measuri?-G
-
z for the phugoid mode from the long-
- =-- term simulator time history of pitch
-.
r
-. attitude shown in Figure 5-7. Doing
-=- so, we find that for x.,= ,016 degrees,
- f x1= ,009 degrees, and n = 1 cycle, the
- b -
-_ -
2 log decrement method produces
Cp = .09. The elapsed time between
% and x, is about 58 seconds, so the
damped phugoid frequency is about
0.11 radians per second. Because
the phugoid is so lightly damped, the
~ natural frequency of the phugoid is
0 I 2 3 6
also about 0.11 radians per second,
ELAPSEO‘TIPlE 1:ECI
to two significant digits. These
pigum 56 Shonpenod response ofpuch amtude 8. markedfor
measuremem of Mnrral frequency and dnmprng rano. Sirnubtor sample measurements are nearly identical to
rate IS I00 samples per second. the analytically derived values of
Cp=.118 and 0~,=.110 radians per
second.

____ -. -_
.. - _.__
__ - .-
- . -
- ...
_.__.I__ -___
.._ ~

- - .._
-.
._.____ - - ,..
-=.-
-0
-.
======
----
.__-
~~

-.-
.-- -
__

-- -__
-- .
1
a 0 za a 50 io f0 10 90 100
” LAPSED riw ~ S E C ~
Figure 5-7 Simularor phugoid response of pirrh am’nuie 8. marked for measurement of rpand up.
Simulator sample rare is 100 samples per second.

These simulator results are quite respectable. But note that the measured damping ratio of both the
short period and phugoid modes is slightly lower than the analytically derived values. This reduction
in damping in the simulator response is caused by the time delay associated with digital computers.
In this case, the time delay is 11100 seconds, or .01 seconds, or 10 milliseconds. When we run the
simulator at 50 samples per second, the time delay increases to 1/50 seconds, or 20 milliseconds, and
the short period damping ratio falls from C,,,=.14 to C,=.09. We see then, that time delay reduces

Hying Qualiiies Testing 5-17

Solving the Equations of Motion
5.3.2 Measuring the Longitudinal Modal Parameters

damping ratio. When we study flight control systems in Part 111, we will see why this happens and
..
why lightly damped modes are most affected. We will also learn that this unwanted but unavoidable
effect applies not only to digital simulators, but to digital flight control systems as well.

Simulating airplanes with lightly damped modes is a difficult task. Had we selected as our subject
an airplane with more heavily damped modes, or with a flight control system designed to increase
the bare airframe damping, we could reduce the simulator sample rate and still come quite close to
matching the analytical results. ~

----- To demonstrate this, we increased the short

-- -
period damping ratio of our example airplane
_ . ---
from C,=.161 to (,=.40 and ran the simulator
F----=---
I_- --
- 7-

at 50, rather than 100, samples per second. The

-- - resulting pitch attitude time history, marked for

-
--
-
--.
__
_- ____
F-
. .- measuring damping ratio, is presented in Figure
-. 5-8. For x,,= 1.09 degrees, x, :.1 degrees, and
_.-
-
__
L R = 1, the log decrement method yields <,=.35.
_-
__ --- .
Bearing in mind that the effect of time delay can
never be eliminated, this result is quite
-- -
0.1*--
__-
-
a
satisfactory. While the 12.5 percent error in
damping ratio may seem large, in practice it is
- unlikely that a pilot would detect the difference.

' 0 2 J.
a
3ZzE
1.16 8 5 2
From a handling qualities perspective, flying
qualities simulators are most useful in the so-
called short period region, where pilots are
E-APSEO'TIME ~SIECI likely to experience the greatest difficulty flying
F i r e 5-8 Simuhror shon period response when <v is an airplane having poor handling qualities. This
increased to 0.40 and the sirnuhior sanple rare i.T reduced short period region Of dynamics extends to
10 50 sampks per second. frequencies of 10 LO 12 radians per second (and
sometimes more), and includes the longitudinal
short period mode of motion and two of the three lateraldirectional modes of motion we will discuss
shortly. It is the region where the most important piloting tasks are performed, including take-off,
landing, tracking, refueling, formation, and so on. Airplane dynamics in this region are usually quite
noticeable to pilots, and are sometimes difficult to control. Motion in the region of the phugoid
mode, on the other hand, is usually so slow that pilots may not notice it at all, and when they do
notice it, they control it easily, even subconsciously.

Recognizing that flying qualities simulators are most useful in the so-called short period region is
important because it allows us to concentrate on improving simulator quality in this region, where
return on investment is greatest. The quality of a simulator is measured by its ability to accurately
match the airplane dynamics, particularly in the short period region. We often speak of "fidelity"

5-18 Flying Qualities Testing

 Solving the Equot'ons of Motion
5.3.4 The Short Period Mod&

when we characterize the quality of a simulator. Just as audiophiles are drawn to "hi-fi" soun&
equipment, flying qualities engineers and pilots are drawn to high fidelity simulators.

It is evident that simulators can do a good job of solving the equations of motion. It is equally
evident that the quality of the simulator solution depends on several factors, including the damping
ratio of the modes being simulated, the quality of the numerical integrator being used, the sample rate
of the simulator, and attention to the nuances of human sensitivity to motion.

It is unfortunate that most military pilots' experience in simulators is limited to training simulators.
In the past, training simulators have not done a good job of simulating flying qualities, and pilots
detected this instantly. The result is a large reservoir of pilots who hold simulators in low regard.
As we say, this is unfortunate because good flying qualities simulators, such as those used at the Air
Force Flight Test Center and at contractor facilities, are accurate and extremely useful in flight
testing. We will be discussing and using simulators throughout our discussion of flying qualities
testing.
f

5.3.3 The Phugoid Mode Note in Figure 5-5 that the phugoid mode is marked by varying airspeed,
flight path angle, and altitude, while angle of attack changes very little. This phugoid motion is
produced by an interchange of kinetic and potential energy. You may observe this exchange in
Figure 5-5. As the nose of the airplane rises in a shallow climb, the airplane gathers altitude and
potential energy while airspeed and kinetic energy are slowly dissipated. As speed wanes during the
climb, the nose of the airplane slowly pitches over into a shallow dive. During the dive, speed and
kinetic energy increase, while altitude and potential energy are lost. Finally, the nose begins slowly
to rise again and the cycle is renewed. This interchange of potential and kinetic energy is attenuated
principally by airplane drag. But the influence of drag is a weak one, so the phugoid mode is usually
very lightly damped.

The period of the phugoid mode is typically rather long, generally between 30 and 90 seconds.
Motion this slow is easily controlled by the pilot. For some airplanes, the pilot may not be aware
of the phugoid motion at all, unless the controls are released for an extended time while airspeed and
altitude are closely watched. The phugoid motion shown in Figure 5-5 is so small that it would be
surprising if the pilot noticed it.

For other airplanes, the phugoid motion may be more pronounced, possibly to the point of taxing the
stomachs of passengers and even seasoned aircrew if allowed to go uncontrolled for extended periods
of time. In modern airplanes, control of the phugoid is usually delegated to the flight control system
or an auto-pilot.

5.3.4 The Short Period Mode In contrast to the phugoid mode, short period motion is marked by
essentially constant airspeed, flight path angle, and altitude. Only angle of attack and pitch attitude
vary during short period motion, and the variation of these two angles is virtually identical. This

Flying Qualities Testing 5-19

Solving the Equalions of Mdion
5.3.5 Solving the Two Degree of Freedom Longitudinal Equations of Motion

..
may be seen in Figure 5-4. Considering the short duration of this mode, it is not surprising that
airspeed, flight path angle, and altitude do not change appreciably.

The period of the short period mode typicaiiy ranges from about one to three seconds, which is to
say that the short period frequency usually ranges from about two to six radians per second.
Exceptions are not uncommon however. The F-15,for example, has a bare airframe short period
frequency in excess of nine radians per second at certain flight conditions.

5.3.5 Solving the Two Degree of Freedom Longitudinal J3quations of Motion Of the two
longitudinal modes of motion, it is usually the short period mode that commands the greater interest.
This is because the faster motion associated with the short period mode is usually more difficult for
humans to control than the slower, low frequency dynamics associated with the phugoid mode of
motion. (Of course, when there is no auto-pilot the phugoid motion may also prove troublesome on
occasion, as when the pilot’s attention is diverted from altitude control, under instrument flying
conditions, to comply with Air Traffic Control requirements.)

Because the short period mode is usually of greater interest than the phugoid mode in flying qualities
work, the five degree of freedom equations of motion, Equations (4-105),are often used. You may
recall that the restricted six degree of freedom equations, Equations (4-104)in section 4.8.4of
Chapter 4, are reduced to five degrees of freedom by constraining the X, axis small perturbation
speed u to be constant. Removing this speed degree of freedom leaves only two degrees of freedom
for longitudinal motion. This effectively eliminates the phugoid mode because kinetic energy can’t
change if speed can’t change. Since the interchange of kinetic and potential energy is the engine that
powers the phugoid mode, eliminating this interchange eliminates the mode.

5-20 Flying Qualilies Testing

 Solving the Equations of Motioh
5.3.5 Solving the Two Degree of Freedom Longitudinal Equations of Motion

I .

Let's check our assertion that eliminating the X, axis speed degree of freedom will eliminate the
phugoid mode of motion. The best way to do this is to solve the two 'degree of freedom equations
for pitch attitude motion and compare the result with the three degree of freedom solution given in
Equation (5-10) in section 5.3.

Extracting the longitudinal equations of motion from Equations (4-105) in section 4.8.4 of Chapter
4, we have the two degree of freedom longitudinal equations:

(5-ll~)

(5-llb)

Equations (5-1 1) are sometimes called the short period approximation equations of motion because
the phugoid mode has been eliminated. If we substitute 9 for q and 8 for Q in Equations (5-1 1) and
Laplace transform the result, we get

(5-1%)

(5-1s)

where, once again to reduce the clutter of notation, it is understood that the motion variables a and
".
6 are functions of s : e.g., a b ) and e@). Combining Ca,with C for the reasons we adduced in
section 5.3, and rearranging Equations (5-12) into a matrix format, we have

(5-13)

It is apparent that computational simplicity is an attractive feature of the two degree of freedom
equations. A glance back at Equation (5-3) in section 5.3 bears this out. We will see in Chapters
7 and 8 that the simplicity of Equation (5-13) allows us to establish some useful insights into short
period airplane motion.

Applying Cramer's rule to solve Equation (5-13) for pitch attitude response to elevator deflection
ae, we obtain

Flying Qualities Testing 5-21

Solving the Equations of Motion
5.3.5 Solving the Two Degree of Freedom Longitudinal Eguations of Motion

..

(5-14)
~

Substituting the airplane physical properties, flight conditions, and stability derivatives from Table
5-3 into Equatipn (5-14), we get
.
e(s) =
I s + 1.18

34.2
-.149

-27.4
a,&)

I s + 1.18

34.2
-s + .00106

s(s + .721)

Evaluating the numerator and denominator determinants yields

-27.4(s + ,996)
e(s) = a*@)
s3+ 1 . 9 0 ~
+ 35.1s
~ - .0362
-
- -27.46 + .996)
(s' + 1.90s + 35.1)(~- .00103)

The ratio of polynomials in Equation (5-15) is the pitch attitude to elevator transfer function, -(es).
6*
It describes the airplane pitch attitude response to an elevator input when we eliminate the degree of
freedom associated with the X, axis perturbation speed u .

Whereas in the three degree of freedom transfer function we had two quadratic factors which
corresponded to the phugoid and short period modes, in the two degree of freedom transfer function
given in Equation (5-15) we have one quadratic and one first order factor. The quadratic factor is
the short period mode. Calculating the natural freqency and damping ratio of the short period mode,
we have

A* = s2 + 1.90s + 35.1

- 0
"r
radians
=J33i=5.92 -
second

5-22 Flying Qualities Testing

 Solving the Equations of Motioh
5.3.5 Solving the Two Degree of Freedom Longitudinal Equations of Motioh

.. ..
2p,* = 1.90

- 1.90
C,=-=.161
2(5.92)

The two degree of freedom short period frequency, o =5.92 radians per second, is nearly identical
%
to the three degree of freedom value, wC=5.90 radians per second. The two degree of freedom short
period damping ratio, <,=.161, is the same as the three degree of freedom value. In a practical
sense, there is no significant difference between the two and three degree of freedom short period
modes in this case.

Using these calculated values of short period frequency and damping ratio, we may write the two
e
degree of freedom version of -(s) using shorthand notation:
6s

The first order unstable root at s=.00103 in the transfer function denominator is a negligible remnant
of the phugoid mode. This root causes an imperceptibly slow divergence of pitch attitude over time.
As we will see, it is not a significant player in the pitch attitude response.

Although in flying qualities testing we are not ordinarily interested in solving the equations of motion
analytically, we will do so in this case so that we may present analytical time histories of the pitch
attitude response. To excite the response, we will use a -1 degree (or -.0175 radian), trailing edge
up, elevator impulse. Substituting -.0175 for 6e in Equation (5-15) and rearranging, we obtain the
partial fraction expansion

e(s) =
- .0136(s - 33.3) + .0136
(5-16)
(s + .%2)2 + 5.842 -
s .rmo3

To inverse Laplace transform the first term into the time domain, we may use the same transform
pair we used in section 5.3 to solve the three degree of freedom equations. We repeat that transform
pair here.

wbcn +=tali'-- b
a -a

+
Recall that because we are working in radian measure, the phase angle in this transform pair is
measured in radians. To perform the inverse Laplace transform of the first order real pole we use
the generalized transform pair

Flying Quoli!ies Testing 5-23

Solving the EqrrOr'ons of Motion
5.3.6 A Footnote on the Phugoid Mode and the Equations of Motion

..

where the time constant T = l is a modal parameter that characterizes the response of first order
a
modes. You will learn more about time constants in Part 111, Chapter 9, when we discuss flight
control system.
~

Using these Laplace transform pairs, we find that the inverse Laplace transform of Equation (5-16)
is

e(t)= -.0809c~~9n'~h(5.842+2.97) rdau

+ .0136~.~'~'

To express 0 in degree measure we multiply this result by 180/r, which gives us

e(t) = -4.64e-mcsin(5.84r +2.97) + .779e.m1mt&pees (5-18)
Quation (5-18) is the two degree of freedom time response of pitch attitude to a -1degree elevator
impulse. Short-term and long-term time histories of this response are presented in Figures 5-9 and
5-10.

A comparison of the two and

three degree of freedom
solutions, Equations (5-18)
and (5-lo), shows that the
short period responses are
virtually identical. We see
then, that removing u , the X,
axis speed degree of freedom,
left the short period mode
intact and eliminated all but an
insignificant remnant of the
phugoid mode. This remnant
-semnds is the very slow divergence of
Figure 5-9 AnalyticaIty derived nvo degree offreedom shon period responre IO
a -I degree elevator impulse. pitch attitude, manifested in
the third term of Equation (5-
18). This divergence is only
barely detectable even after 100 seconds, as you may see in Figure 5-10.

These results suggest that it is indeed safe to use the two degree of freedom small perturbation
equations of motion when we wish to focus our attention on the longitudinal short period mode.

5.3.6 A Footnote on the Phugoid Mode and the Equations of Motion Perhaps you have noticed
an apparent conflict between what we have said and what we have demonstrated regarding the
dependence of the phugoid mode on speed changes. We said that speed must vary if there is to be

5-24 Flying Qualities Testing

 Solving the Equafions of Motion
5.4 Solving the Lateral-DirectionalF.quations of Motion

I _ .
an interchange of kine& and
potential energy to drive the
phugoid mode. But you may
3 recall that we constrained true
airspeed, V,, to be constant
1.
-0
m"
when we were developing and
simplifying the & and
equations of motion. This
f constraint is reflected in' the
restricted six degree of
4
a ,, u a ,, u freedom equations of motion,
nm%S-XUIB which we presented as
Figure 5-10 Analytically derived two degree offeedom long-term response to a Equations ( 4 4 9 , in section
-1 degree eIevator impulse. 4.8.1 of Chapter 4. Since we
have not lifted this
requirement that V, be constant, how is it that the three degree of freedom5quations of longitudinal
motion could produce the speed changes necessary to sustain a phugoid mode?

When we set the restriction, we pointed out that constraining V, to be a constant was a convenient
way to simplify the & and equations by eliminating the k, t e r n . But V,=d-, and we
did not impose a constraint on U , V, and W,'which means that these terms, and by implication V,,
are free to change. We saw the evidence of this in the phugoid motion presented in Figures 5-3 and
5-5.

The upshot is that constraining V, to be constant will eliminate the phugoid motion in real life, but
not in the restricted six degree of freedom equations as we have written them. Requiring that V, be
constant serves only to simplify the equations of motion. Ultimately, it is the requirement that u ,
v , and w be small perturbation values that resolves the apparent conflict between constant V, and the
presence of the phugoid mode in the longitudinal three degree of freedom equations of motion. So
long as we observe the small perturbation limitation, small changes in U , V,and IY can coexist with
the assumption of constant v,.

5.4 Solving the Lateral-DirectionalEquations of Motion Perhaps you recall that a useful feature
of the linearized, decoupled, small perturbation equations of motion, is that these equations decouple
longitudinal and lateral-directional motion. This permits us to analyze longitudinal motion separately,
as we have done in the preceding sections of this chapter, and also to analyze lateraldirectional
motion separately, as we will do in the sections that follow. Taking advantage of this decoupling
feature, we extract the lateraldirectional equations from Equations (4-104), which we presented in
section 4.8.3 of Chapter 4.

Flying Qualities Testing 5-25

Solving the Equalions of Motion
5.4 Solving the Lateral-Directional J3quatiom of Motion

..

(5-19b)

(5-1%)

Just as, in section 5.3, we discovered that we had four longitudinal motion variables and only three
+
equations, we also discover that we have four lateraldirectional motion variables ( p , , p , and r )
and only three equations. As we did in the longitudinal case, we must choose between adding a
fourth equation or expressing one of the motions variables in terms of the others. The established
convention is to choose the latter course because it simplifies the solution. Evaluating third order
determinants is easier than evaluating fourth order determinants.

Recall from Equations (4-55) in section 4.5.5 that

p=&-+sine

When an airplane is trimmed for straight and level flight, Ip =O and 8 is small, so that we may write
P=&. For small perturbation motion this becomes p = & . Hence, we may substitute & and 4 for
p and p in Equations (5-19). This leaves us with three motion variables ( p , 4 , and r ) and three
equations of motion.

We will follow the same procedure in solving the lateraldirectional equations of motion that we
followed in solving the longitudinal equations of motion. Laplace transforming Equations (5-19). we
have

(5-20b)

(5-2Oc)

Note that s@ and s2+ have replaced p and p.

Rearranging Equations (5-20) and placing them in a matrix format we obtain

5-26 Flying Qualities Testing

 Solving the Equations of Motion
5.4 Solving the Lateral-Directional Equations of Motion

(5-21)

We will use Cramer's rule to solve Equation (5-21) and find the response of bank angle 4 to an
aileron input &a. In section 3.2.3 in Chapter 3 we reviewed the use of Cramer's nile when there are
two or more inputs, as there are in Equation (5-21). To determine the response to we first set
&,=O. 'This has the practical effect of eliminating the second column of the matrix on the right hand
side of Equation (5-21). Now, to determine the bank angle response'to &a we substitute the
remaining column from this matrix for the second column of the numerator determinant (the column
associated with 4). This gives us

Substituting the airplane physical properties, flight conditions, and stability derivatives from Table
5-3 in section 5.3 into Equation (5-22). we get

Hying Qualities Testing 5-27

Solving the Equations of Motion
5.4.1 The Dutch Roll Mode

1 1
..
~+.292 -.0278

50.2 31.4 -.0350~- .345

-22.5 4.97 s + . w
4(s) = a&)
s + .292 -.0127~- 0270 1

50.2 s(s + 2.76) -.0350~- .345

-22.5 ~ ( - . 0 0 5 8 3+~,0575) s + .0460

Evaluating the numerator and denominator determinants, we obtain

31.6(s2 + .848s + 30.5)
4(s) = a&)
' + 3 . 5 1 +~ 2~5 . 6 +~ 67.0s
s ~ + .413

- 31.6(s2 + A485 + 30.5)

a s = (5 -23)
(s2 + .679s +23.7)(s +2.83)(s + .OOalS)

4
The ratio of polynomials in Equation (5-23)is the bank angle to aileron transfer function, -(s) . It
&a
describes the airplane bank angle response to an aileron input. We see in Equation (5-23)that the
lateraldirectional characteristic equation factors into .a quadratic and two first order terms. The
quadratic term is the lateral-directional short period mode, which is called the durch roll mode of
motion. The first order term with the short time constant, TR=-,1 is called the roll mode, or the
2.83
roll subsidence mode, and the first order term with the long time constant, Ts=- , is called the
.00618
spiral mode.

5.4.1 The Dutch Roll Mode The name "dutch roll" is said to have come from the characteristic
side-to-side motion exhibited by skaters on the wintry, frozen ponds of the Netherlands. The dutch
roll mode is an often lightly damped rolling and yawing response with a natural frequency in the
vicinity of one to six radianslsecond. The magnitude of the ratio of the bank angle to sideslip
response during the dutch roll motion is called the "phi to beta" ratio, or 14/18 1. The 14/18I ratio has
two components: a dynamic component denoted l+/pld, and a steady-state component denoted
I+/IpI,. The dynamic component, 14/18Id, is the magnitude of the ratio of 4 to 18 during the dutch
roll oscillation. The steady-state component, 14/plm, is the magnitude of the ratio of 4 to 18 after the
dutch roll oscillations have subsided. When the value of l@/pId is large, the rolling motion is more
noticeable to the pilot. Hence, the motion associated with a dutch roll mode having a large 14/18Id
is often described as "rolly." When the value of 14/181d is small, the yawing motion is more
noticeable to the pilot. The motion associated with a dutch roll mode having a small 14/18Id is often
described as "snakey," because of the predominantly side-to-side motion.

5-28 Flying Qualities Testing

 Solving the Equations of Motion
5.4.2 The Roll Subsidence Mode

Lightly damped dutch roll modes were a problem in the early swept wing jets, prompting a flbbd of
. -.
pilot complaints. F-86 pilots were upset with the lightly damped dutch roll motion of that airplane.
During the Korean war, quite a hit of work was devoted to building a yaw damper for the F-86,
because pilots felt the snaking bullet pattern caused by the dutch roll motion was costing them MiG
kills. Today, flight control systems rein in the dutch roll mode, rendering it more docile.

From the quadratic dutch roll factor in Equation (5-23), we may calculate the dutch roll damping
ratio and natural frequency for a bare airframe F-16AiB at 1.2 Mach number and 31,000 feet:
A& = s a + ,679s + 23.7

- o "=,m = 4 . 8 6 rudiam/second

21,0,, = .679

=
In section 5.4.6 we will compare these analytically derived values of dutch roll frequency and
damping with measurements taken from a simulator time history, just as we did for the phugoid and
short period modes of longitudinal motion. We will also compare analytical and measured values of
the roll mode and spiral mode time constants.

Now that we have determined i, and a , we may use shorthand notation to express the transfer
,
function in Equation (5-23). The result is

4s)
-( = -=$ 31.6 [.0768,5.52]
80 AM (.ao618)(2.83)[.oa98.4.86]

where the subscript Id in A M denotes the lateraldirectional characteristic polynomial.

The bare airframe F-16AiB dutch roll damping ratio, G=,0759, is fairly representative of modem
fighter airframes at supersonic speeds. Well designed flight control systems take care of this lightly
damped hare airframe response so that pilots may perform demanding tasks relatively easily. For
example, at 1.2 Mach number and 31,000 feet, the F-16AiB flight control system increases the dutch
roll damping ratio to a value of C=, .45.

5.4.2 The Roll Subsidence Mode The roll subsidence mode, more commonly referred to as the
roll mode, figures prominently in the rolling performance of an airplane. It determines the time
required for an airplane to reach a steady state roll rate for a given aileron input. The shorter the
time constant (that is, the larger the root in the factored characteristic equation), the faster an airplane
can reach a steady-state roll rate.

Flying Qualities Testing 5-29

Solving the Equations of Motion
5.4.4 Completing the Lateral-Directional Solution

5.4.3 The Spiral Mode The spiral mode is usually an unstable, or divergent rolling response that
has a very long time constant. The spiral mode characterizes the tendency of an airplane to slowly
roll off on one wing. In the 1920's and 30's. when flight instruments were few and relatively
primitive, this slow roll divergence was called the "graveyard spiral" because, at night or in weather,
many pilots flew into the ground when they failed to detect the insidious build-up of bank angle, and
the resulting increase in dive angle, caused by the spiral mode.

The divergence of unstable spiral modes is usually so slow that, like the phugoid mode, pilots can
easily control it, usually without thinking about it, provided they have adequate visual or instrument
references. But the spiral mode can be an irritation when the flight control system or auto-pilot isn't
working and the pilot wants to trim the airplane for long-term, hands-off, wings-level flight. When
the spiral mode is unstable, any disturbance, no matter how small, will cause bank angle to begin a
divergent march. How many pilots have heaped obloquy on the T-33 for its notorious spiral
instability at cruising altitude? Today, the spiral mode, like the dutch roll, short period, and phugoid
modes, is brought to heel by well designed flight control systems.

Interestingly, the bare airframe spiral mode of the F-l6A/B at this flight condition is stable.

5.4.4 Completing the Lateral-Directional Solution Before we can complete the solution of
Equation (5-23), we must select an input, or forcing function. We will use an impulse aileron
deflection of 1 degree, or ,0175 radians. Substituting 6n=.0175 radians into Equation (5-23) and
performing a partial fraction expansion, we obtain

4(s) = - '
0147(s + 8.71) --+ .238 .252
(s + 339)' + 4.8@ s + 2.83 s + .00618

To obtain the time response, we use the same Laplace transform pairs we used in section 5.3.5 to
solve the two degree of freedom longitudinal equations of motion for pitch attitude. Repeating these
transform pairs we have

where 4 =tat-'- b
a -a

b
-
s+a
- be-"'

where the phase angle @ in the first transform pair is in radian measure. Using these transform pairs
to perform the inverse Laplace transformation of Equation (5-24), we get
@ ( t )= -.0291e-.3"0'sin(4.86t+ ,525) - .238e-2-83'
+ . 2 5 2 ~ - . radianr
~~~'
1so
To obtain the solution in degree measure, we need only multiply by -. Doing so, we obtain
lT

5-30 , Flying Qualities Testing

 Solving the Equations of Motion
5.4.5 Lateral-Directional Digital Simulator Solution

@(f) = -1.67e-.3"'sin(4.86f + .525) - 13.6e-2.83'+ 14.4e-.W618'

degrees (5 -25)

where the phase angle in the sin(4.86f+ ,525) term is in radian measure. Equation (5-25) is the hank
angle time response of a bare airframe F-l6A/B when it is excited by a one degree aileron impulse
at Mach 1.2 and 31,000 feet.

Using the same procedure, we may also solve Equation (5-21) for p and r in response to an aileron
e,
input, or we may solve for p , and r in response to a rudder input.

2.
A time history of Equation (5-
25) is presented in Figure 5-
11. Each of the characteristic
15
lateral-directional modes of
m
m motion is clearly visible in
ki le
-3
this time history. Bank angle
6
quickly increases to about 14
degrees, right wing down, and
5 oscillates as it begins a very
slow return to wings level.
The gradual return to wings
e
8 1 7. 3 5 b 7 9 10 level occurs only because, in
Time, seconds this case, the spiral mode is
Figure 5-11 Analytically derived response of bank angle to a one degree stable. More typically, the
impulse aileron deflection.
bank angle - would slowly
diverge to larger and larger
values. The lightly damped oscillation is the dutch roll response, and the initial, rapid rise of hank
angle is governed by the roll mode.

The lightly damped dutch roll motion in Figure 5-11 is not experienced by F-16 pilots because the
flight control system increases the dutch roll damping, making the airplane much easier to handle.
We will see how this is done when we study flight control systems in Part 111.

5.4.5 Lateral-Directional Digital Simulator Solution Equation (5-25) is an exact solution of

Equations (5-19). We have already observed, in section 5.3.1, that a simulator can never exactly
match an analytical solution because of the limitations of numerical integration techniques. To
illustrate the kind of results you may expect to get from a simulator, we built a digital simulator that
solves Equations (5-19). The Fortran code for this simulator is presented in Appendix 5A. We used
a fixed-step, fourth order, Runge-Kutta, numerical integrator running at 1000 samples per second.
This unusually high sample rate was necessary because of the very lightly damped dutch roll mode.
Recall from our longitudinal simulator experience in sections 5.3.1 and 5.3.2 that digital time delay
reduces damping and that lightly damped modes are most affected. If the bare airframe dutch roll
mode of the F-l6A/B at 1.2 Mach number and 31,000 feet were more heavily damped, or if we

Flying Qualities Testing 5-31

Solving the Eqwlions of Motion
5.4.5 Lateral-Directional Digital Simulator Solution

._ - .-.. I

. ,.-

. .-. ... .. . . . ..

0 , I J ~LAPSEll’TInE t & C l
9 IO

Figure 5-12 Simulator time histories of 13, @, p, and r in response to a 20 degree oileron pulse. Simulator
sample rate is loo0 samplesper second.

included the flight control system in our simulator, we could run at 50 samples per second and
achieve good results.

Because a simulator cannot match an analytical impulse function, we approximated the analytical
input with a twenty degree, 0.1 second, triangular aileron pulse. Perhaps you are surprised that a
twenty degree pulse is needed to approximately match the response produced by a one degree impulse
function. This is because an analytical impulse contains more energy than a triangular pulse, as you
will learn when you study flight control systems in Part III.

Simulator time histories of p , 41, p , and r responding to a 20 degree aileron pulse are presented in
Figure 5-12. Bearing in mind the difference in inputs, the simulator bank angle response appears to
be quite similar to the analytical response. It is interesting to note that p and r exhibit clear dutch
roll responses, but no obvious roll or spiral mode components. In fact, the dutch roll, roll, and spiral
modes are present in the response of every lateraldirectional motion variable, but the amplitudes of
the modes (that is, the magnitudes of the residues) varies greatly. For example, the analytical
solution of yaw rate response to a one degree aileron impulse is

5-32 Flying Qualities Testing

 Solving the EquntiOns of Motim
5.4.6 Measuring the Lateral-Directional Modal Parameters

.. &
. ..
r(t) = 4 . 1 4 ~ - ~ ~ s i n ( 4 . 8 61.67)
1 + + .64Qc'2.B' + .386~-.~~~*
second
Note that each mode is present in this solution, and that the frequency, damping ratio, and time
constants are identical to those in the bank angle impulse response, Equation (5-25). The only
difference between this result and the bank angle impulse response is the amplitudes of the modes and
the phase angle of the dutch roll mode. The roll mode amplitude for the bank angle response is 12
times the roll mode amplitude for the yaw rate response, so it is not surprising that the roll mode
response is difficult to detect in yaw rate.

5.4.6 Measuring the Lateral-Directional Modal Parameters To determine how similar the
simulator and analytical solutions are, we will isolate the simulator time history of bank angle
presented in Figure 5-12 and repeat it in Figure 5-13. We will use this time history to measure the
dutch roll frequency and dampiing ratio and the roll and spiral mode time constants. We will compare
the measured values with the analytically derived values and see what conclusions we may draw.

Figure 5-13 Tim history of bank angle responding to a 20 degree aileron pulse, markcdfor determination
of modal paromters. Simulator sample rate loo0 samples per second.

Let's look first at the oscillatory dutch roll response in Figure 5-13. Using the log decrement
method, we can determine C, from measurements of the envelope amplitudes % and, four and a half
cycles later, x , ~ . For +,=2.0 degrees, xu=.70 degrees, and n=4.5 cycles, the log decrement method
yields a damping ratio of about .0371, compared to the analytically calculated value of C,=.O697.

Flying Qualilies Testing 5-33

Solving the Equations of Morion
5.4.6 Measuring the Lateral-Directional Modal Parameters

We determine the natural frequency by first calculating the damped frequency, od,. Noting that it -. . ..
takes 5.8 seconds for 4.5 cycles of motion, we calculate a damped natural frequency of .77 Hz or

4.86 radians per second. The natural frequency is on&=%, which gives us a dutch roll natural
/i-z
frequency of 0,=4.86 radians per second, which is identical to the analytically calculated value.

The dutch roll mode is so lightly damped that, even at loo0 samples per second, the-digital time
delay has an effect on the simulator solution. But we want to be wary of misinterpreting the
difference between the simulator and analytical solutions. While a 51 percent error in damping ratio
seems large, the damping ratio itself is so small that a 51 percent error doesn't make much of a
difference in the time response. It is almost certain that a human pilot would not detect the difference .
between a damping ratio of .0371 and 3697. The simulator solution for the dutch roll mode is
actually quite good, especially for a mode so lightly damped. Still, we must keep it in mind that we
ran the simulator at loo0 samples per second to achieve these good results. Usually, the flight
control system increases the dutch roll damping to levels that allow mucb slower Simulator sample
rates. Fifty samples per second is a more typical rate.

We may approximate the roll and spiral mode time constants from the bank angle time history in
Figure 5-13. When we study flight controls you will learn that a first order time constant is the time
required for a system respon:e to reach 63 percent of its steady-state value. If we fair a line through
the middle of the oscillation and estimate the peak bank angle to be about 14 degrees, the roll mode
time constant is the time required for the bank angle to reach .63(14) =8.8 degrees. From Figure 5-13
we see that .43 seconds is required for the bank angle to reach 8.8 degrees. Hence the roll mode
1
time constant is T, = .43 seconds. The analytically derived value is T, = - =0.35 seconds.
2.83

Our measured value of roll mode time constant is more accurate than it seems. Because digital
simulators cannot produce analytical impulses, we approximated an impulse with a 0.1 second
triangular pulse. But bank angle does not respond as quickly to pulses as to impulses. For this
reason, measured time constants will always be smaller than real time constants when the
measurement is taken from a bank angle response to a pulse input. We get around this in flight
testing by using step aileron inputs and measuring the roll mode time constant on a time history of
roll rate. If we use a 2 degree step aileron input in our simulator, we get the roll rate response
shown in Figure 5-14. Fairing through the dutch roll response, the steady-state roll rate is about 28
degrees per second. The time needed for roll rate to reach 63 percent of steady state, or 17.6 degrees
per second is about T,=.37 seconds, which compares well with the analytically derived value of
T,=.35 seconds.

To measure the spiral mode time constant, we will use the time history of bank angle responding to
an aileron pulse, presented in Figure 5-13. We begin by fairing a straight line through the dutch roll
oscillation and estimating the slope of bank angle versus time as bank angle returns slowly to zero.
Next, we extend the slope of decaying bank angle backwards to zero seconds. At zero seconds, the

5-34 Flying Qualilies Testing

 Solving the Equations of Motioia
5.4.6 Measuring the Lateral-Directional Modal Parametem

PI

extended value of bank angle is about 14.5 degrees. Then we mark the extended slope at ten
seconds, at which point it has a value of about 13.6 degrees. The slope is 14.5-135 =.wdegrees
10
per second. This slope is the rate of decay, or subsidence, of bank angle and is governed by the
spiral mode. Using this rate of decay, we calculate the time required for bank angle to return 63
percent of the way to wings level. Sixty-three percent of the peak bank angle, about 14 degrees, is
8.8 degrees. At the measured rate of decay, it will take T S = E=98 seconds for bank angle to return
.09
63 percent of the way to wings level. Considering the problems of fairing so fine a slope through
the dutch roll oscillation and measuring such small values on so coarse a scale, this is not too
different from the analytically calculated value of Ts=--162 seconds.
.00618

The long subsidence time of the spiral mode is similar to the long period of the phugoid mode. They
are usually such slow motion modes that pilots control them easily, even subconsciously, and often
don't notice them at all. For this reason, the practical difference between the analytically derived
spiral mode time constant and the simulator value that we measured is almost certainly not detectable
by a pilot engaged in a flying task.

In section 5.4.1 we remarked on 14/18 Id. the ratio of roll to sideslip in the dutch roll mode. We may
calculate 14/18 I d analytically using the 4 and 18 responses to an aileron or rudder input. We already
have the bank angle response to a one degree aileron impulse. The sideslip response is

p(f) = -.890~-~'sin(4.86f
+ .108)+ .0572~-*."~
+ .00751e-.m'8r degrees (5-26)

Flying Qualilies Testing 5-35

Solving the Equasions of Motion
5.4.6 Measuring the Lateral-Ddonal Modal Parameters

To calculate /@/@Id, we merely divide the amplitude of the dutch roll mode for the bank angle
response by the amplitude of the dutch roll mode for the sideslip response. From Equation (5-25),
the amplitude of the dutch roll mode for the bank angle response is -1.67 degrees. From Equation
Id
(5-26) the amplitude of the dutch roll mode for the sideslip response is -.890 degrees. The I@/@

ratio is -1.67 = 1.88.

-.E90

We may also measure I@/@ Id from the simulator time histories in Figure 5-12. In Figure 5-15 we
repeat the simulator time histories of I$ and p that were presented in Figure 5-12. We begin by
selecting a time at which to make the measurements. Next, we sketch envelopes around the dutch
roll responses of both bank angle and sideslip in the vicinity of that time. Then we measure the
envelope amplitudes and calculate the ratio. For example, at time t = 4 seconds, the amplitude of the
bank angle envelope is about 1.75 degrees and the amplitude of the sideslip envelope is about .93
degrees. The measured Id ratio is 1.75 = 1.88, which is the same as the analytically derived
I@/@
.93
%
value.

- ...

I 2 1 9 IO
~ L A P S i O S T I M E (tEC1
Figure 5-15 Time histories of4 and B responding to an aileron pulse, markedfor measuring the 4 to B ratio.
Simulator sMlpre rate is IOW samples per second.

The lightly damped lateraldirectional modes we are working with are a severe test of simulator
fidelity. Judging from the comparisons we have just made, it appears that simulators can do a good
job of solving the lateraldirectional equations of motion. As we have already pointed out, and
confirmed here, the fidelity of a simulator depends on the damping ratio of the modes being
simulated, the quality of the numerical integrator used, and the simulator sample rate.

5-36 Flying Qualities Testing

 Solving the Equations of Motion
5.5 Waypoint: Solving the Equations of Motion

5.4.7 Concluding Note We conclude by noting that the modal parameter measurements we ;nacik
..
using simulator time histories could just as easily have been made using flight test time histories.
However, when a flight control system is present, we get the augmenred modal parameters rather than
the bare airframe modal Parameters. Often, the augmented response is so heavily damped that it is
difficult to measure the modal parameters accurately. To counter this difficulty we use the equations
of motion and a model of the flight control system, together with flight test measurements, to estimate
the modal parameters. We will introduce you to these simple but powerful techniques in Part IV,
when we study flying qualities test methods. ~

5.5 Waypoink Solving the Equations of Motion We solved the longitudinal and lateraldirectional
equations of motion twice: first analytically, and then with a digital simulator. We also introduced
the characteristic modes of airplane motion, identified the modal parameters analytically, and
measured the modal parameters using time histories of simulator (or airplane) motion.

We began by analytically solving the longitudinal three degree of freedtm equations of motion,
revealing in the process the phugoid and short period modes of motion. We pointed out that the short
period mode tends to dominate pilot attention and workload, and hence flight control design and flight
testing. This is because dynamics in the short period region can be the most difficult for human
pilots to handle. In contrast, the phugoid mode is usually a slow motion response that is easier to
control.

We noted that the phugoid mode is fueled by an interchange of kinetic and potential energy, so that
if speed is held constant the phugoid mode disappears. We confirmed this by eliminating the x, axis
speed degree of freedom and analytically solving the resulting two degree of freedom equations of
longitudinal motion. In this two degree of freedom solution, the short period mode was left intact
and unchanged while the phugoid mode disappeared, as we predicted.

Next we undertook to solve the lateraldirectional equations of motion analytically. We discovered

that lateraldirectional motion also has a characteristic short period mode, called the dutch roll mode.
In addition to the dutch roll mode, there are two first order characteristic modes of lateraldirectional
motion. One of these is the roll subsidence mode, usually abbreviated to roll mode, which is a
relatively fast mode of motion that governs roll acceleration. The other is the spiral mode, which
is a very slow mode of rolling motion. Usually the spiral mode is unstable, resulting in the slow
divergence of bank angle over time. This mode makes it impossible to trim an airplane for long-
term, hands-off, wings-level flight, unless a flight control system or an auto-pilot is engaged.

We observed that, today, well-designed flight control systems and auto-pilots modify these five modes
of motion, making the airplanes easier to fly.

We also used a digital simulator to solve both the longitudinal and lateraldirectional equations of
motion, using a popular numerical integrator. By comparing the simulator and analytical solutions,
we demonstrated that simulators can do a surprisingly good job of solving the equations of motion.
We witnessed the adverse effect of digital time delay on lightly damped modes of motion and noted

Flying Qualities Testing 5-37

Solving the Equnrions of Motion
5.5 Waypomt: Solving the Equations of Motion

..
that time delay is also a problem in digital flight control systems. You will learn the simple
explanation for this when we study flight control systems in Part 111.

The question naturally arises, how are the modal parameters we calculated and measured used? One
very important use is in guiding the design of flight control systems. Another is to characterize
flying qualities. In Chapter 6, we will briefly introduce the Military Standard for "Flying Qualities
of Piloted Aircraft," MIL-STD-1797 (USAF). This document is a compendium of several decades
of flying qualities experience and research, and is a principal source of guidance in flight control
design. In Chapter 6 we will show you how to use MIL-STD-1797 to characterize the flying qualities
of the bare airframe F-16AB.

5-38 Flying Quoliries Testing