DYNAMIC MODELING AND STABILITY

ANALYSIS OF MODEL-SCALE HELICOPTERS

WITH BELL-HILLER STABILIZING BAR
R. Cunha

and C. Silvestre

Instituto Superior Tecnico,

Institute for Systems and Robotics,
Lisbon, Portugal.
This paper presents an accurate self-contained helicopter dynamic model, derived
from rst-principles, and that is specially tailored for model-scale helicopters. The
simulation model includes the rigid body, main rotor apping, and Bell-Hiller stabi-
lizing bar dynamics. Particular emphasis is placed on the analysis of the stabilizing
bar and on the evaluation of its impact on the overall helicopter dynamics. The
model is parameterized for the Vario X-Treme acrobatic helicopter, and solutions
for a set of trimming trajectories are identied and discussed. Dierent simplica-
tions, needed to derive models for control system design, are presented, compared,
and their inuence on the resultant dynamics evaluated. The eect of changing the
physical parameters of the stabilizing bar is also assessed. An LQ state feedback
controller is synthesized to stabilize the vehicle in forward ight. Simulation results
obtained with the full nonlinear dynamic model and the forward ight control system
are presented and discussed.
Introduction
Over the last few years, model-scale unmanned he-
licopters have become a major topic of research. The
wide and valuable range of applications due to the
helicopter high maneuvering capabilities, together
with the recent availability of increasingly accurate,
reliable and miniaturized sensors justify the exten-
sive research and development eort undertaken by
numerous groups. Dynamic modeling of helicopters
plays an essential role in this eort. Without accu-
rate and reliable models, the design of guidance and
control systems for autonomous ight is seriously un-
dermined.
The main contribution of this paper is the presen-
tation of a simple yet accurate self-contained heli-
copter dynamic model, whose derivation was based
on theoretical principles, and that has the particular

Ph.D. candidate, Department of Electrical Engineering

and Computer Science, and Institute for Systems and Robotics
(ISR), Instituto Superior Tecnico, Av. Rovisco Pais, 1, 1046-
001, Lisboa, Portugal. rita@isr.ist.utl.pt

Assistant Professor, Department of Electrical Engineering

and Computer Science, and Institute for Systems and Robotics
(ISR), Instituto Superior Tecnico, Av. Rovisco Pais, 1, 1046-
001, Lisboa, Portugal. cjs@isr.ist.utl.pt
feature of including a detailed description of the Bell-
Hiller stabilizing bar. Its equations of motion, which
explicitly include the inertial, gyroscopic and aero-
dynamic eects, are incorporated in the helicopter
model, taking into account the geometry of the Bell-
Hiller mixing device. An analysis and explanation
of the Bell-Hiller system stabilizing eect is devel-
oped, based both on the analytical expressions and
on linear models obtained by linearization of the full
nonlinear model about specic trimming conditions.
The impact of changing the physical parameters of
the stabilizing bar is also assessed.
As a result of the work presented hereafter, a he-
licopter dynamic model simulator, named SimMod-
Heli ,
5
implemented in Matlab, using Simulink and C
MEX-le S-functions, will be made freely available
for the scientic community. This simulator is com-
pletely parameterizable and describes the dynamics
of helicopters with any number of blades, with or
without Hiller or Bell-Hiller stabilizing bar. The sim-
ulation model includes the rigid body, main rotor
apping, and stabilizing bar dynamics.
Several reference books on the theory of heli-
copter ight dynamics can be found in the litera-
ture. The reader is referred to Padeld
11
, Johnson
7
,
1
American Institute of Aeronautics and Astronautics
AIAA Guidance Navigation and Control Conference, Texas, USA, 2003
Bramwell et al.
2
, and Prouty
12
, for a thorough and
extensive coverage of full-scale helicopters aerody-
namic and dynamic modeling. However, none of
them covers the modeling of the Bell-Hiller system.
Only Bramwell et al.
2
give a very brief and simplied
explanation of the Bell stabilizing device that resem-
bles the Bell-Hiller stabilizing bar, but does not use
aerodynamic paddles nor does it include a mixing de-
vice. In addition, the analysis presented covers only
the longitudinal plane dynamics.
Although there are a few publications that ex-
plicitly address the topic of model-scale unmanned
helicopter modeling, including the Bell-Hiller system
(see, for example,
6, 10
and references therein), these
are mostly based on a black box system identication
approach. Results from linear frequency identica-
tion and time-domain analysis of ight-data are used
to develop the model. In general, the obtained mod-
els are restricted to particular ight regimes, such
as, hover or forward ight, and use very crude ap-
proximations to describe the fundamental eects on
helicopter behavior. Only Kim et al.
9
present a
modeling approach based on rst-principles and ba-
sic aerodynamics, however it is not clear how is the
stabilizing device included in the overall helicopter
dynamics. To the best of the authors knowledge, no
stability analysis of the Bell-Hiller system eect on
helicopter dynamics has been presented so far.
The paper is organized as follows. The rst sec-
tion introduces a general model for the dynamics
of helicopters, presents the main rotor apping dy-
namics, and describes in detail the modeling of the
Bell-Hiller stabilizing bar dynamics. The second sec-
tion presents the trimming results obtained for the
Vario X-Treme model-scale helicopter followed by a
stability analysis of the stabilizing bar impact on the
helicopter dynamics. The third section focuses on
the design and implementation of a forward ight
control system for the Vario X-Treme helicopter, and
presents the simulation results obtained with the full
nonlinear dynamic model. The last section sum-
marizes the contents of the paper and points out
directions for future work.
Helicopter dynamic model
This section presents the dynamic model of a sin-
gle main rotor and tail rotor helicopter equipped with
a Bell-Hiller stabilizing bar, as the one depicted in
Fig. 1. A comprehensive study of the helicopter dy-
namic model can be found in
3
. For in depth coverage
of helicopter ight dynamics, the reader is referred
to Johnson
7
and Padeld
11
.
Fig. 1 Vario X-Treme helicopter
The dynamics of the helicopter can be described
using a 6 DoF rigid body model driven by forces
and moments that explicitly include the eects of
the main rotor, Bell-Hiller stabilizing bar, tail rotor,
fuselage, horizontal tailplane, and vertical n.
To derive the equations of motion, the following
notation is required:
{U} - universal coordinate frame;
{CM} - body-xed coordinate frame, with ori-
gin at the vehicles centre of mass;
p =
_
x y z

T
- position of the vehicles center
of mass, expressed in {U};
=
_

T
- Z-Y-X Euler angles that
parametrize locally the orientation of the vehicle
relative to {U};
v =
_
u v w

T
- body-xed linear velocity vec-
tor;
=
_
p q r

T
- body-xed angular velocity vec-
tor.
Fig. 2 captures the general structure of the heli-
copter model that can be written as
_

_
v = v + [f (v, , u) +f
g
()] /m
= I
1
( I) +I
1
n(v, , u)
p = R() v

= Q()
, (1)
where m is the vehicle mass, I is the tensor of inertia
about the {CM} frame, u is the command vector, f
and n are the vectors of external forces and moments
respectively along the same frame, f
g
is the gravita-
tional force also expressed in {CM}, R is the rotation
matrix from {CM} to {U}, and Q is the transforma-
tion from angular rates to Euler angle derivatives.
The command vector u =
_

0

1c

1s

0t

T
com-
prises the main rotor collective input
0
, main rotor
2
American Institute of Aeronautics and Astronautics
AIAA Guidance Navigation and Control Conference, Texas, USA, 2003
Fig. 2 Helicopter model - block diagram
and ybar cyclic inputs
1c
and
1s
, and tail rotor col-
lective input
0t
. The total force and moment vectors
account for the contributions of all helicopter com-
ponents, and can be decomposed as
f = f
mr
+ f
tr
+ f
fus
+ f
tp
+ f
fn
n=n
mr
+n
tr
+n
fus
+n
tp
+n
fn
, (2)
where subscript mr stands for main rotor, tr for tail
rotor, fus for fuselage, tp for horizontal taiplane, and
fn for vertical n.
As the primary source of lift, propulsion and con-
trol, the main rotor dominates helicopter dynamic
behaviour. The Bell-Hiller stabilizing bar improves
the stability characteristics of the helicopter. The
tail rotor, located at the tail boom, provides the mo-
ment needed to counteract the torque generated by
the aerodynamic drag forces at the rotor hub. The
remaining components have less signicant contri-
butions and simpler models as well. In short, the
fuselage produces drag forces and moments and the
horizontal tailplane and vertical n act as wings in
forward ight, increasing ight eciency.
The following sections present mathematical mod-
els for the main rotor and Bell-Hiller stabilizing bar.
Detailed models for the remaining components can
be found in Padeld
11
and Prouty
12
.
Main rotor
In rotary-wing aircraft, the main rotor is not only
the dominant system, but also the most complex
mechanism. It is the primary source of lift, which
counteracts the body weight and sustains the heli-
copter on air. Additionally, the main rotor generates
other forces and moments that enable the control of
the aircraft position, orientation and velocity. This
section presents a simplied rotor dynamic model,
whose main building blocks are depicted in Fig. 3.
The main rotor actuation system consists in chang-
ing the blade pitch angle , along each blade revolu-
tion. Through this variation, control over the blade
aerodynamic loads, which ultimately determines the
main rotor force and moment contributions (f
mr
and
n
mr
), is attained. Without the Bell-Hiller system
Fig. 3 Main rotor block diagram
and neglecting the servo actuators dynamics, the
blade pitch angle is given by
() =
0
+
1c
cos() +
1s
sin(). (3)
where = t is the blade azimuth angle and
is the rotor speed. In systems equipped with the
Bell-Hiller stabilizing bar, only the collective input

0
is directly applied to the main rotor. The cyclic
inputs are mixed with the motion of the bar to de-
termine the actual cyclic components (
1c
and
1s
)
applied to blade pitch links. The equations govern-
ing the motion of these components (represented by
the Bell-Hiller block in Fig. 3) are described in the
next section.
The aerodynamic forces, generated at the surface
of the rotating blades, produce the rotor thrust that
is responsible for sustaining and propelling the heli-
copter. In reaction to this thrust force, by conserva-
tion of momentum, the air is accelerated downwards.
The generated oweld can be approximated by the
mean component perpendicular to the rotor disk,
usually called induced downwash, see Fig. 3. The
present model uses momentum theory combined with
blade element theory
7
to compute the induced down-
wash, in the normal working state (climb, hover and
low rates of descent) and windmill brake state (high
rate of descent). In the vortex ring and turbulent
wake states, where momentum theoty is not applica-
ble, empirical approximations are used.
4
Rotor blade loads are not uniquely determined by
the applied inputs. They also depend on helicopter
velocities, induced downwash velocity, and on the
motion of the blades themselves, which besides rota-
tion about the hub and blade pitch angle variation,
also includes ap and lag bending and pitch torsional
motions.
7
The model adopted to describe rotor
blades is standard and assumes that these are rigid
and linked to the hub through ap hinge springs,
with stiness k

.
11
The dynamic behaviour is thus
conned to the apping motion (Fig. 3), described
by vector =
_

0

1c

1s

T
, where
0
denotes the
collective mode (also denominated coning), and
1c
and
1s
the longitudinal and lateral cyclic modes,
respectively. This vector, which corresponds to the
3
American Institute of Aeronautics and Astronautics
AIAA Guidance Navigation and Control Conference, Texas, USA, 2003
constant and rst-order harmonics of the Fourier Se-
ries expansion of (), comprises the fundamental
components of the apping motion.
The apping equation of motion for a single blade
is obtained from the moment equilibrium expressed
on a frame attached to the rotating and apping
blade, mounted on the moving helicopter. To derive
the aerodynamic contributions to this equilibrium,
expressions for the air velocity at each blade sec-
tion were established. Using standard notation in
helicopter theory, the helicopter velocities are nor-
malized and expressed in the main rotor wind frame,
with and
z
denoting the forward and vertical ve-
locities, respectively, and p and q the roll and pitch
rates, respectively.
11
The induced downwash is also
normalized and decomposed into constant
0
and si-
nusoidal components
1c
and
1s
.
The resulting equation of motion, expressed in the
wind-aligned frame, can be expanded in Fourier Se-
ries and the constant and rst-order harmonic terms
extracted to obtain the following second-order sys-
tem of dierential equations

+ A

()

+
2
A

() =
2
B

()
_
_

1c

1s
_
_
+

2
B

()
_
p
q
_
+
2
B

()
_
_

z

0

1c

1s
_
_
.
(4)
It should be noted that, for control system design
purposes, the apping motion as described by (4)
preserves a high degree of accuracy, while rendering
a much more tractable system. For instance, the
coecient matrices in (4) depend solely on the heli-
copter forward velocity.
4
In steady-state, equation (4) reduces to
= [A

()]
1
_
_
B

()
_
_

1c

1s
_
_
+B

()
_
p
q
_
+B

()
_
_

z

0

1c

1s
_
_
_
_
,
(5)
with coecient matrices given by
[A

()]
1
=
1
S
2

+ 1

4
4
_

_
_
S
2

+1

4
4
_

8
S

+1
0 0

4
3

8
S

+1
8

_
1 +

2
2
_

4
3

2
2

8
S

+1
8

_
1

2
2
_
8

_
.
(6)
B

() =

8
_

_
1 +
2
0
4
3

0 1 +
2
0
8
3
0 1 +
3
2

2
_

_, (7)
B

() =
_

8
_
2
3

_
0
2

8

8
2
_

_, (8)
and
B

() =

8
_
_
4
3
0
2
3

0 1 0
2 0 1
_
_
. (9)
The Lock number and Stiness number S

are
structural adimensional constants that give the ra-
tio of aerodynamic to inertial forces, and the ratio
of stiness to aerodynamic moments, respectively.
They embed all the physical characteristics of the
rotor relevant for an accurate description of its ap-
ping behaviour.
The major eects in apping motion become more
perceptible if the conditions on helicopter motion are
further constrained to include no forward velocity at
the hub ( = 0). Then, expressions (6) and (7) show
that
0
commands the coning mode
0
and that the
cyclic modes
1c
and
1s
are determined by an S

-
dependent combination of the cyclic inputs.
In the case of articulated rotors (k

= 0 and thus
S

= 0), with no forward velocity at the hub ( = 0),

A

becomes the identity matrix and expression (5)

reduces to
_

0
=

8
_

0
+
4
3

z

4
3

0
_

1c
=
1s
p +
16

q +
1s

1s
=
1c
+
16

q + q
1c
. (10)
The cyclic modes are now fully decoupled and cor-
respond to the solution of a second order dynamic
system excited at the resonant frequency (maximum
magnitude amplication, 90
o
input-output phase
shift).
Using either the dynamic or the steady-state so-
lution for the apping, the main rotor forces and
moments at the hub (last block in Fig. 3) can be
written as
f
mr
=
n
2
_
_
Y
1s
Y
1c
2Z
0
_
_
+
n
2
_
_
Z
1c
Z
0
0
Z
1s
0 Z
0
0 0 0
_
_
_
_

1c

1s
_
_
,
(11)
and
n
mr
= n
_
_
0
0
N
0
_
_
+
n
2
_
_
N
1c
N
0
k

N
1s
k

N
0
0 0 0
_
_
_
_

1c

1s
_
_
.
(12)
4
American Institute of Aeronautics and Astronautics
AIAA Guidance Navigation and Control Conference, Texas, USA, 2003
The Y
(.)
, Z
(.)
, and N
(.)
terms, in (11) and (12),
represent the force and moment components gener-
ated by the blades. These quantities are functions of
the helicopter state variables and main rotor inputs
(see
3
, for further details).
The main rotor thrust and torque, Z
0
and N
0
re-
spectively, have dominant out-of-plane components
(along the hub z axis), and smaller in-plane compo-
nents, which are due to the main rotor tilt. Terms
Z
1c

0
and Z
1s

0
represent the in-plane contribu-
tions of the blade lift forces due to the rotor coning,
while Y
1c
and Y
1s
account for the in-plane contribu-
tions of the drag forces acting on the blades. In (12),
the spring moments, due to the cyclic ap angles, are
explicitly given by k

1s
for the roll moment and
k

1c
for the pitch moment.
Bell-Hiller stabilizing bar
The Bell-Hiller stabilizing bar, a mechanical blade
pitch control system that improves helicopter stabil-
ity, is currently a standard component of model-scale
helicopters. From a control point of view, the stabi-
lizing bar can be interpreted as a dynamic feedback
of the roll and pitch rates. The system consists of a
so-called ybar, a teetering rotor placed at a 90
o
ro-
tation interval from the main rotor blades and tipped
on both ends by aerodynamic paddles, see Fig. 4.
Paddle
Flybar
Paddle
Fig. 4 Bell-Hiller stabilizing bar
With this mechanism, the blade pitch cyclic com-
mands do not go directly from the swashplate to the
blade pitch links. Instead, the cyclic commands are
applied to the ybar whose apping motion deter-
mines the blade pitch angles. The system derives
from a combination of the Bell stabilizing bar, t-
ted with a mechanical damper and weights at each
tip, and the Hiller stabilizing bar, which instead of
weights uses small airfoils with incidence commanded
by the cyclic inputs.
2
In the Hiller system, the blade
pitch angle is determined by the ybar apping only.
The Bell-Hiller system introduces the mixing device
that allows some of the swashplate input to be di-
rectly applied to the blades.
The ybar and main rotor apping motions are
governed by the same eects, namely the gyroscopic
moments due the helicopter roll and pitch rates. But
unlike the main rotor, the ybar is not responsible
for providing lift or maneuvering ability. Thus, it can
be designed to have a slower response and provide
the desired stabilization eect. The ybar response
can be optimized by varying the ratio of aerody-
namic to inertial loads on the paddles. Changing
the shape, weight or distance between the paddles
are all straightforward ways of tailoring the system.
The notation used to describe the Bell-Hiller sys-
tem is presented in Fig. 5.
Fig. 5 Bell-Hiller system with angular displace-
ments
Due to the Bell-Hiller system, the ybar apping
and blade pitching angles are physically constrained
to satisfy
_

0
()

1
()
_
=
l

l
1
l
1
+l
2
. .
c
1
_
(l
4
h

()) /l

1
()
_
+
l

f
l

l
2
l
1
+l
2
. .
c
2
_
0

f
_
+

2
_
_
,
(13)
where
1
is the dierential pitch input, given by

1
() =
1c
cos() +
1s
sin(). (14)
5
American Institute of Aeronautics and Astronautics
AIAA Guidance Navigation and Control Conference, Texas, USA, 2003
Likewise,
f
has no coning mode, since the ybar, as
a teetering rotor, can only describe see-saw apping
motions. Equation (13) shows that the collective
pitch,
0
(), depends only on the vertical displace-
ment of the swashplate, given by h

(), and that

the dierential pitch,
1
(), results from a linear
combination of
1
() and
f
( +/2). The coe-
cients are constant values that depend on the ratios
between the lever arms involved in the system.
Besides satisfying this algebraic relation, blade
pitching and ybar apping are also constrained to
verify the moment equilibrium given by
n

f
_
+

2
_
=
l

l
2
l
1
+l
2
n

() . (15)
Expressions (13) and (15) can be used to combine
the ybar apping and blade pitching equations of
motion. This combination showed that the contribu-
tions from the blade pitching dynamics are negligible
with respect to the ybar apping,
3
yielding the fol-
lowing result
_

1c

1s
_
+ A

1c

1s
_
+
2
A

1c

1s
_
=

2
B

1c

1s
_
+ B

_
p
q
_
+
2
B

_
_

z

0

1c

1s
_
_
,
(16)
where
A

=
_

f
/8 2
2
f
/8
_
, (17)
A

=
f
/8
_
0 1
1
2

2
1
1
2

2
0
_
, (18)
B

=
1
c
2

f
8 _
0 (1+c
1
)+
1
2
(3+c
1
)
2

2
(1+c
1
)
1
2
(1c
1
)
2

2
0
_
,
(19)
B

=
1
c
2
_

f
/8 2
2
f
/8
_
, (20)
and
B

=
1
c
2

f
8
_
2
2
0 1
0 1 0
_
. (21)
According to (16), the blade pitching response to
helicopter shaft rotations is determined by c
1
, c
2
,
dened in (13), the constant
2
given by

2
= R
2
R
2
2
R
2
1
R
4
2
R
4
1
, (22)
and the ybar Lock number dened as

f
= c
f
a
0
f
_
R
4
2
R
4
1
_
/ I

f
, (23)
where is the air density, c
f
the paddle chord, a
0
f
the paddle lift curve slope, and I

f
the ybar mo-
ment of inertia. Therefore, there are several dierent
means of adjusting the blade pitching response to
helicopter shaft rotations, which can be divided into
two sets, one concerning the ybar and the other
concerning the Bell-Hiller system. Computing the
eigenvalues for (16) with = 0, yields a very simple
expression for the real part, which is given by
=

16
. (24)
According to this result, an increase in the ybar
aerodynamic factors (the paddle chord, c
f
, the lift
curve slope, a
0f
, the ybar radius, R
2
, or the paddle
span, given by R
2
R
1
), has the eect of speeding
up the ybar response, and decreasing the overall
stabilization eect. Conversely, the use of heavier
blades has the eect of increasing the stabilization
eect. In the Bell-Hiller system, increasing l
1
, the
lever arm that is usually designed to be adjustable,
increases B

and decreases B

and B

, giving more
weight to the commands and decreasing the stabi-
lization eect.
The stability analysis of the helicopter linearized
models, presented in the following section, reinforce
the conclusions described above.
Trimming and linearized systems
analysis
Having established the equations of motion for
the helicopter, including the main rotor apping and
blade pitching dynamics, it becomes possible to nd
the solutions under which the helicopter describes
trimming ight trajectories and linearize the system
about these trajectories. The eigenstructure analysis
of the linearized systems can provide further insight
into the eect of the Bell-Hiller stabilizing bar on
helicopter stability.
To perform this analysis, the full nonlinear he-
licopter model was implemented in Matlab, using
Simulink and C MEX-le S-functions to increase the
computational speed. The results presented here-
after were obtained with the model parameterized
for the Vario X-Treme helicopter.
4
Consider the helicopter equations of motion pre-
sented in (1), and let v
c
,
c
, p
c
,
c
, and u
c
denote
the trimming values of the state and input vectors.
At trimming, these vectors satisfy
_
0 =
c
v
c
+ [f (v
c
,
c
, u
c
) +f
g
(
c
)] /m
0 = I
1
(
c
I
c
) +I
1
n(v
c
,
c
, u
c
)
.
(25)
6
American Institute of Aeronautics and Astronautics
AIAA Guidance Navigation and Control Conference, Texas, USA, 2003
From (25), it can be shown that
_
_
_
u
c
= 0

c
= 0

c
= 0
. (26)
Given the dependence of the gravitational terms on
the roll and pitch angles, only the yaw angle can
change without violating the equilibrium condition.
However,
c
satises
_
_
0
0

c
_
_
= Q(
c
)
c
, (27)
and thus the yaw rate,

c
, is constant. As shown
in
4
, trimming trajectories correspond to helices that
can be described by

c
=
_
_
0
0

c
_
_
, p
c
=
_
_
V
c
cos(
c
) cos(

c
t +
0
)
V
c
cos(
c
) sin(

c
t +
0
)
V
c
sin(
c
)
_
_
, (28)
where V
c
= v
c
is the linear body speed,
c
the
ight-path angle, and
0
the helix initial condition.
The trimming problem was solved for a set of
straight line trajectories at dierent speeds, using
the X-Treme helicopter conguration. The set com-
prises horizontal trajectories (
c
= 0

c
= 0) with
forward speed V
c
varying from 0 to 14 m/s. Figs. 6-7
shows the trimming values obtained for the inputs.
As forward speed increases, the change in the main
rotor collective is due to the combination of two ef-
fects: decrease in induced downwash and increase in
fuselage drag. The tail rotor collective follows a sim-
ilar curve, reecting the fact that it is counteracting
the main rotor torque.
The trimming values for the helicopter roll and
pitch angles are depicted in Fig. 8. The slight side-
ward inclination provided by the negative roll is
needed to balance a combination of sideforces, dom-
inated by the tail rotor thrust. The pitching angle is
approximately given by the helicopter drag to thrust
ratio, which naturally increases with speed. Fig. 8
shows the expected nose-down pitch evolution as
function of forward velocity. The trimming solutions
presented in Fig. 6-8 are fairly consonant with the re-
sults for the Helisim simulation model, presented in
11
, which contributes for the validation of the present
simulation model.
Linearized models describing disturbed motion
about the given set of trimming conditions were ini-
tially computed for the full 6DoF model with no
0 2 4 6 8 10 12 14
0
1
2
3
4
5
Forward velocity (m/s)
C
o
l
l
e
c
t
i
v
e

c
o
m
m
a
n
d
s

(
d
e
g
)

0t
Fig. 6 Trimming values for the collective com-
mands
0 2 4 6 8 10 12 14
1
0.5
0
0.5
1
1.5
Forward velocity (m/s)
C
y
c
l
i
c

c
o
m
m
a
n
d
s

(
d
e
g
)

1c

1s
Fig. 7 Trimming values for the cyclic commands
0 2 4 6 8 10 12 14
2.5
2
1.5
1
0.5
0
0.5
Forward velocity (m/s)
E
u
l
e
r

a
n
g
l
e
s

(
d
e
g
)

Fig. 8 Trimming values for the helicopter orien-

tation
7
American Institute of Aeronautics and Astronautics
AIAA Guidance Navigation and Control Conference, Texas, USA, 2003
ybar. Fig. 9 shows evolution of eigenvalues for
the X-Treme helicopter linearized dynamic model as
a function of forward speed, and identies the as-
sociated modes that characterize helicopter motion
under the small perturbation assumption.
80 60 40 20 0
40
30
20
10
0
10
20
30
40
Re(s)
I
m
(
s
)
hover
8 m/s
10 5 0
6
4
2
0
2
4
6
roll subsidence
pitch subsidence
Dutch roll
phugoid
heave and yaw
subsidences
Fig. 9 Eigenvalues for X-Treme helicopter model
linearized at dierent forward velocities
To assess, in terms of stability, the impact of the
Bell-Hiller system on the overall helicopter dynamic
model, the following analysis focuses on the unstable
phugoid mode and on the comparison between high
and reduced order models. As shown in Fig. 10,
the introduction of the ybar has a stabilizing ef-
fect on the phugoid mode, whose damping and fre-
quency factors change from (0.50; 1.30 rad/s) to
(0.61; 0.89 rad/s). These eigenvalues are identi-
cal, considering either the rst or the second-order
ybar apping dynamics (or main rotor blade pitch-
ing dynamics). In fact, the rst-order system, which
assumes that the inertial terms in (16) are negligi-
ble, constitutes an adequate approximation, since the
only mode that is being neglected is fast (322 rad/s)
and does not couple with the fuselage body motions.
The same is not true for the slower ybar apping
mode, which couples with the roll and pitch subsi-
dence modes (see Fig. 10).
A standard approach followed in helicopter sta-
bility analysis assumes that the coupling between
lateral and longitudinal modes can be neglected,
and that the helicopter motion can be described
by two independent lower-order systems. In heli-
copters tted with the Bell-Hiller ybar, this par-
titioning proves to be inadequate. An unstable 8th-
order model may correspond to two stable reduced-
order models. This is due to the fact that lat-
eral/longitudinal couplings increase the instability of
80 60 40 20 0
400
300
200
100
0
100
200
300
400
Re(s)
I
m
(
s
)
flybarless
flybar (1st order)
flybar (2nd order)
2 1 0 1
1
0
1
Fig. 10 Eigenvalues for systems with and without
Bell-Hiller stabilizing bar (hover)
the long-period oscillatory modes, see Fig.11.
3 2 1 0 1
1
0.5
0
0.5
1
1.5
Re(s)
I
m
(
s
)
6DoF (no flybar)
6DoF (flybar)
Long.+Lat. (no flybar)
Long.+Lat. (flybar)
Fig. 11 Comparison between 6DoF, longitudinal
and lateral models (hover)
Another relevant issue, addressed in this analysis,
consists in quantifying the impact of the physical
parameters of the Bell-Hiller system on the over-
all helicopter stability. The obtained results are in
agreement with expressions for the blade pitching
dynamics given in (16)-(21), conrming that an in-
crease in paddle weight m
f
improves stability, while
increasing the ybar radius R
2
or the lever arm l
1
have the opposite eect, see Fig. 12.
From the conducted analysis, it can be concluded
that, for an adequate description of the Bell-Hiller
stabilizing eect, the model should include both the
rst-order blade pitching dynamics and the couplings
between longitudinal and lateral modes.
8
American Institute of Aeronautics and Astronautics
AIAA Guidance Navigation and Control Conference, Texas, USA, 2003
0.4 0.45 0.5 0.55 0.6 0.65
0.6
0.65
0.7
0.75
0.8
Re(s)
I
m
(
s
)
l
1
: 0.018> 0.030 (m)
R
2
: 0.365> 0.620 (m)
m
f
: 0.029> 0.045 (Kg)
(l
1
= 0.030 m)
(R
2
= 0.0620 m)
(m
f
= 0.045 Kg)
Fig. 12 Variation of Bell-Hiller stabilizing bar
parameters (hover)
Control system design and simulation
This section focuses on the design, implementa-
tion and simulation of a forward ight control sys-
tem for the Vario X-Treme helicopter. The lin-
ear state feedback controller was required to meet
the following design specications: i) Zero Steady
State Error, achieve zero steady state error in re-
sponse to constant input commands in the vector
e = [z
c
z,
c
, u
c
u, v
c
v]

, four ex-
tra integrators were added, one to each channel in
e; ii) Actuator Bandwidth Requirements, the control
loop bandwidth for all actuators should not exceed
30 rad/s to ensure that the main and tail rotor com-
mand servos are not driven beyond their normal
actuation bandwidth.
The forward ight controller was obtained by re-
sorting to the solution of the standard continuous
time Linear Quadratic Regulator problem,
1
where
the state and control weighting matrices Q and R,
respectively, were selected as to achieve a reasonable
tracking performance for the channels in e without
violating the actuator bandwidth requirements.
K
Tz
z1
z1
Tz

iE E E
i
c
T
E
'
E E
E
+
+

x
e
u
Fig. 13 Controller implementation with an anti-
windup mechanism.
The controller was discretized using a sampling
frequency of 50 Hz and the actuators were saturated
at 8
o
to avoid blade stall. The implementation of the
resulting discrete time controller, was done by using
the D-methodology,
8
which guarantees the following
fundamental linearization property: the linearization
of the nonlinear feedback control system about each
equilibrium trajectory preserves the internal as well
as the input-output properties of the corresponding
linear closed loop designs. This methodology moves
all integrators to the plant input, and adds derivators
where they are needed to preserve the transfer func-
tions, making straightforward the implementation of
anti-windup schemes, see Fig. 13. Furthermore, the
input trimming values are naturally provided by the
integrator block, which is a major issue in this ap-
plication where the constant terms present in model
have to be compensated. In the gure, e represents
the state variables that are required to achieve good
tracking performance in steady state, vector x the
helicopter state variables including the main rotor
blade pitching, and u = [
0
,
1s
,
1c
,
0t
]

the heli-
copter actuation vector.
The results of the simulation, presented in Figs. 14
through 20, were obtained with the full nonlinear
closed loop system that includes the nonlinear dy-
namic model of the Vario X-Treme helicopter and
the D implementation of the controller. The maneu-
ver was performed about a horizontal forward ight
trimming condition (V
c
= 2 m/s,
c
= 0,

c
= 0) and
can be divided in ve stages: i) keep the helicopter
in level ight during one second; ii) track a positive
ramp in altitude, Fig. 15; iii) keep the new altitude
for three seconds; iv) track an increase in forward
velocity, Fig. 14; v) maintain the helicopter at the
new equilibrium (V
c
= 2.6 m/s,
c
= 0,

c
= 0).
0 2 4 6 8 10 12
1.9
2
2.1
2.2
2.3
2.4
2.5
2.6
2.7
Time (s)
F
o
r
w
a
r
d

v
e
l
o
c
i
t
y

(
m
)
v
c
v
Fig. 14 Forward velocity v
c
and v
9
American Institute of Aeronautics and Astronautics
AIAA Guidance Navigation and Control Conference, Texas, USA, 2003
0 2 4 6 8 10 12
2
1
0
Time (s)
A
l
t
i
t
u
d
e

(
m
)
z
c
z
Fig. 15 Altitude z
c
and z
0 2 4 6 8 10 12
2.5
3
3.5
4
4.5
5
Time (s)
C
o
l
l
e
c
t
i
v
e

p
i
t
c
h

c
o
m
m
a
n
d
s

(
d
e
g
)

0t
Fig. 16 Collective actuation commands
0 2 4 6 8 10 12
0.2
0
0.2
0.4
0.6
0.8
1
Time (s)
C
y
c
l
i
c

p
i
t
c
h

c
o
m
m
a
n
d
s

(
d
e
g
)

1c

1s
Fig. 17 Cyclic actuation commands
Between the rst and third seconds of the maneu-
ver, the actuation variable
0
, that corresponds to
the main rotor collective, increases to impart the
desired ascending rate to the vehicle, Fig. 16. The
remaining actuation variables, the longitudinal and
lateral cyclics,
1c
,
1s
respectively, and the collective
tail rotor,
0t
, react as to compensate for the model
coupling. In the third stage of the maneuver, the ac-
tuation acquires the initial trimming values, required
to keep the vehicle in the commanded altitude.
0 2 4 6 8 10 12
1.2
1
0.8
0.6
0.4
0.2
0
0.2
Time (s)
L
i
n
e
a
r

v
e
l
o
c
i
t
i
e
s

(
m
/
s
)
v
w
Fig. 18 Linear velocities
0 2 4 6 8 10 12
0.06
0.04
0.02
0
0.02
0.04
0.06
Time (s)
A
n
g
u
l
a
r

v
e
l
o
c
i
t
i
e
s

(
r
a
d
/
s
)
p
q
r
Fig. 19 Angular velocities
In the fourth stage, the controller responds to the
commanded increase in forward velocity by lowering
and raising
1s
, which has the eect of tilting the he-
licopter forward (decrease in pitch angle , Fig. 20)
and redirecting the thrust vector to increase forward
speed. As the vehicle enters the nal stage of the
10
American Institute of Aeronautics and Astronautics
AIAA Guidance Navigation and Control Conference, Texas, USA, 2003
maneuver, the reverse action takes place and the ac-
tuation acquires the trimming values associated with
the new equilibrium condition.
0 2 4 6 8 10 12
3
2
1
0
1
2
3
Time (s)
E
u
l
e
r

a
n
g
l
e
s

(
d
e
g
)

Fig. 20 Euler angles

Conclusions
The paper presented an accurate self-contained he-
licopter dynamic model, derived from rst-principles,
and that was specially tailored for model-scale heli-
copters. Particular emphasis was placed on the anal-
ysis of the stabilizing bar and on the evaluation of
its inuence on the overall helicopter dynamics. The
trimming trajectories were identied and discussed.
Dierent approximations, needed to derive simplied
models for control system design, were presented,
compared, and their impact on the resultant dynam-
ics evaluated. The model was parameterized for the
case of the Vario X-Treme acrobatic helicopter, and
the eect of changing the physical parameters of the
stabilizing bar on the helicopter dynamics was as-
sessed. An LQ state feedback controller was designed
and implemented, and its performance evaluated in
simulation along a typical maneuver. Future work
will focus on adjusting and validating the model so
that it can be used to exploit the particular dy-
namic characteristics of model-scale helicopter in its
whole ight envelope. Extra eort will be placed on
studying, developing, and testing advanced control
strategies to achieve good performance characteris-
tics in highly demanding maneuvers.
Acknowledgements
This work was supported by the Portuguese FCT
POSI programme under framework QCA III and by
the POSI/SRI/41938/2001 ALTICOPTER project.
The work of R. Cunha was supported by a PhD
Student Scholarship, SFRH/BD/5034/2001, from
the Portuguese FCT POCTI programme.
References
1
B. D. O. Anderson, J. B. Moore, Optimal Con-
trol, Linear Quadratic Methods, Prentice Hall, New
Jersey, 1990.
2
A. R. S. Bramwell, G. Done, D. Balm-
ford, Bramwells Helicopter Dynamics, 2nd Edition,
Butterworth-Heinemann, Oxford, Great Britain,
2001.
3
R. Cunha, Modeling and control of an au-
tonomous robotic helicopter, MSc thesis, Depart-
ment of Electrical and Computer Engineering, Insti-
tuto Superior Tecnico, Portugal, 2002, in english.
4
R. Cunha, C. Silvestre, Modeling and simu-
lation of model-scale helicopters, Internal Report,
Institute for Systems and Robotics, Portugal, 2003.
5
R. Cunha, C. Silvestre, SimModHeli: A Dy-
namic Simulator for Model-Scale Helicopters, 11th
Mediterranean Conference on Control and Automa-
tion MED03, Rhodes, Greece, June 2003.
6
V. Gavrilets, B. Mettler, E. Feron, Nonlin-
ear Model for a Small-Size Acrobatic Helicopter,
Proc. of the AIAA Guidance, Navigation, and Con-
trol Conference, Montreal, Quebec, Canada, August
2001.
7
W. Johnson, Helicopter Theory, Dover Publica-
tions, New York, USA, 1994.
8
I. Kaminer, A. Pascoal, P. Khargonekar, E. Cole-
man, A Velocity Algorithm for the Implementa-
tion of Gain-Scheduled Controllers, Automatica,
31(8):11851191, 1995.
9
S. K. Kim, D. M. Tilbury. Mathematical
Modeling and Experimental Identication of an Un-
manned Helicopter with Flybar Dynamics, Submit-
ted to the Journal of Robotic Systems, 2001.
10
B. Mettler, M. Tischler, T. Kanade, System
Identication Modeling of a Model-Scale Helicopter.
Technical report CMU-RI-TR-00-03, Robotics In-
stitute, Carnegie Mellon University, Pittsburg, PA,
USA, January 2000.
11
G. D. Padeld, Helicopter Flight Dynamics:
The Theory and Application of Flying Qualities and
Simulation Modeling, AIAA Education Series, Wash-
ington, USA, 1996.
12
R. W. Prouty, Helicopter Performance, Sta-
bility, and Control, Krieger Publishing Company,
Florida, USA, 1995.
11
American Institute of Aeronautics and Astronautics
AIAA Guidance Navigation and Control Conference, Texas, USA, 2003