Vous êtes sur la page 1sur 13

Prediction of Vortex Ring State Boundary of a Helicopter

in Descending Flight by Simulation

Pierre-Marie Basset
ONERA
Salon-de-Provence, France

Chang Chen
J. V. R. Prasad
Georgia Institute of Technology
Atlanta, GA

Sebastien Kolb
ONERA
Salon-de-Provence, France

A helicopter rotor in descending flight may encounter its own wake resulting in a doughnut-shaped ring around the rotor
disk, known as the vortex ring state (VRS). While VRS is a region of descending flight of unsteady flow through the
rotor, a determination of the precise boundary surrounding this region has posed a challenge to the researchers over the
years. Several criteria have been proposed in the literature for the determination of VRS boundaries. This paper considers
prediction of VRS boundaries using three different criteria, viz., zero transport velocity of rotor tip vortices, bifurcation of
equilibria, and zero heave damping. Two different inflow models previously developed at ONERA and the Georgia Institute
of Technology are used in the VRS boundaries predictions. It is shown that, within the accuracy of the inflow models used,
the VRS boundary predictions are significantly influenced by the criterion used.
Nomenclature
CT
C To
DT0
DTC
DTS
DTA
k
kx
kz
Nring
Nb
S
T
T
(Uhel , Vhel , Whel )
VH
VicMR , VisMR
Vih
VimMR , VimFAN
VTVDES

Vtv X
Vtv Z
VZ

VRS

rotor mean thrust coefficient


rotor thrust coefficient at hover
collective pitch angle
lateral cyclic pitch control
longitudinal cyclic pitch control
tail rotor collective pitch angle
empirical factor used in the Newman criterion
empirical factor used in the tip vortices criterion
empirical factor used in the Wolkovitch criterion
number of vortex rings
number of blades
rotor disk area
amplitude of rotor thrust fluctuation
mean rotor thrust
helicopter velocity vector in the helicopter
coordinate system
rotorcraft horizontal velocity in the Earth
coordinate frame
inflow gradients of the first harmonic inflow
model for the main rotor
rotor-induced velocity at hover
rotor mean induced velocity, MR for the main
rotor, FAN for the tail rotor fan in fin
nondimensional velocity of tip vortices and
adapted parameter for the tip vortices criterion
(Eq. (8))
in-plane component of tip vortex velocity
vertical component of tip vortex velocity
rotorcraft vertical speed in the Earth coordinate
frame (positive in climb)

x
z
WTV
WTVE

vortex ring state


empirical factor used in the tip vortices criterion
blade angle of attack
Z
descent angle, 2 arctan( V
)
VH
rotorcraft horizontal speed nondimensionalized by
the blade tip speed (advance ratio)
rotorcraft vertical speed normalized by
rotor-induced velocity at hover (positive in climb),
V Z / Vih
rotor-induced velocity normalized by induced
velocity at hover (positive in climb), Vi / Vih
air density
induced velocity used in Newman criterion
normalized by rotor-induced velocity at hover
rotorcraft horizontal speed normalized by
rotor-induced velocity at hover, V H / Vih
translational velocity used in the Newman
criterion normalized by rotor-induced velocity at
hover
vertical velocity used in the Newman criterion
normalized by rotor-induced velocity at hover
wake transport velocity used in the Newman
criterion normalized by rotor-induced velocity
at hover
effective wake transport velocity used in the
Newman criterion normalized by rotor-induced
velocity at hover
rotor rotational speed
bank and pitch angles
Introduction

Corresponding author; email: pierre-marie.basset@onera.fr.


Manuscript received June 2006; accepted November 2007.

Over certain ranges of forward speed and descent rate, a helicopter


rotor may encounter its own wake, resulting in a doughnut-shaped ring
139

140

P.-M. BASSET

around the rotor disk, known as the vortex ring state (VRS). Flight in VRS
condition can be dangerous as it may cause an uncommanded increase
in descent rate, excessive thrust and torque fluctuations, vibration, and
loss of control effectiveness. As simple momentum theory is no longer
valid for a rotor in VRS, modeling of rotor inflow in the VRS continues to daunt researchers. While routine operations of a helicopter in the
VRS are restricted, a better understanding of the VRS problem and the
ability to accurately predict the VRS boundaries provide certain advantages. For example, a detailed understanding of the VRS problem may
lead to the development of an automated system for VRS avoidance.
Also, the ability to accurately predict the VRS boundaries may lead to a
better utilization of the safe operational envelope to facilitate significant
noise abatement, e.g., through segmented steep approaches for civilian
helicopters.
Typical aerodynamic phenomena associated with VRS include unsteadiness of flow, excessive thrust and torque fluctuations, and a significant increase in vibration. Castles and Gray (Ref. 1) measured induced
velocity for multiple rotor configurations in VRS and observed considerable fluctuation in the distribution of induced velocity. Yaggy and Mort
(Ref. 2) studied the steady and oscillating rotor thrust in descending flight.
Their test results showed a loss in steady thrust and large oscillations of
rotor thrust in the VRS. Washizu et al. (Ref. 3) conducted experiments
to measure unsteady aerodynamic characteristics of a rotor operating
in the VRS. Empey and Ormiston (Ref. 4) tested a 1/8-scale model of
the AH-1G helicopter in a wind tunnel. Their data revealed significant
thrust oscillations in descent conditions. Wang (Ref. 5) applied the classical vortex theory in axial descent with a linear decay of circulation
of trailing vortices owing to the effect of fluid viscosity and interaction
of induced flow with opposite free-stream flow. The resultant induced
velocity curve from Wangs model matched well with the experimental
data from Ref. 1. Xin and Gao (Refs. 6, 7) conducted whirling beam
tests in both axial and inclined descent conditions. Remarkable fluctuations in both rotor thrust and torque were observed, especially in the
region of VZ /Vih = 0.6 to 0.8 (VZ /Vih is the rotor vertical speed
normalized by the induced velocity at hover). The loss in the mean rotor
thrust was also clearly indicated in the same region. Leishman et al. (Ref.
8) applied a time-accurate free-vortex wake scheme in their study. The
onset and development of the VRS was viewed as the result of spatial
and temporal wake instability. Brown et al. (Ref. 9) developed a vorticity transport model in their study of blade twist effects on a rotor in the
VRS.
Dynamic behaviors of a single-rotor helicopter operating in the VRS
condition include uncommanded drop in descent rate and loss of control
effectiveness. Basset and Prasad (Ref. 10) explored the use of nonlinear
analysis tools such as the bifurcation method in gaining an understanding of the flow dynamics associated with the VRS. In the investigations
reported in Refs. 11 and 12, a flight test campaign dedicated to the study
of VRS was conducted using a single main-rotor helicopter. The test program identified the main flight characteristics in the VRS condition as a
sudden increase in descent rate. Similar to the finding from Drees and
Hendal (Ref. 13), the VRS effects were not observed beyond forward
speeds of VH /Vih = 1.0 (VH /Vih is the helicopter forward speed normalized by the rotor-induced velocity at hover). Brand et al. (Refs. 14, 15)
conducted an extensive flight test program to evaluate VRS effects on
the V-22 tilt-rotor aircraft. While degradation of control effectiveness in
the vertical axis was often present for a single main-rotor configuration,
uncommanded roll response was more obvious in VRS encounters for a
tilt-rotor configuration.
Momentum theory has been successfully used for rotor inflow modeling in hover, climb, and even forward flight conditions. However, momentum theory breaks down in descending flight due to the collapse of
a smooth slipstream. As pointed out in Ref. 16, the region may begin

JOURNAL OF THE AMERICAN HELICOPTER SOCIETY

at a rate of descent equal to about 23% of the hover-induced velocity


and persist until the rate of descent exceeds 125% of the hover-induced
velocity. Nevertheless, researchers have developed various methods in
extending simple momentum theory for descending flight. One of the
earliest efforts can be traced back to Glauert (Ref. 17). Recent attempts
have come from He et al. (Ref. 18) and Jimenez (Ref. 19). They individually formulated parametric extensions of momentum theory in the
flow model to remove the modeling singularity in VRS and rendered
simulation models covering the full range of flight conditions. Perhaps
the most comprehensive parametric extension of momentum theory has
been Johnson (Ref. 20). On the basis of a broad review of available
wind-tunnel and flight test data for rotors in the VRS, a model was developed suitable for simple calculations and for real-time simulations.
Using the developed VRS model, Ref. 20 showed negative (unstable)
heave damping for a certain range of descent rates, resulting in the VRS
boundary being defined in terms of the stability boundary of the vehicle
flight dynamics. It is to be noted that models based on parametric extension of momentum theory need to further establish direct dependence
on critical rotor parameters such as rotor solidity and blade twist, and
so on.
VRS Boundaries
A number of criteria for arriving at VRS boundaries have been proposed over the years. These criteria include a region of roughness (Ref.
13), thrust or torque fluctuation (Refs. 3, 6, 7), mean thrust reduction
(Ref. 21), zero velocity of tip vortices (Ref. 22), blade-flapping fluctuation (Ref. 8), wake breakdown (Ref. 23), heave stability (Ref. 12), and
roll stability (Refs. 14, 15).
The most common way to display a VRS boundary is to use freestream velocity components, VH and VZ , normalized by hover-induced
velocity Vih . The area within the VRS boundary indicates the region
where the effects of VRS are significant in terms of a chosen criterion.
Another commonly used variable in the depiction of VRS boundaries is
descent angle (D ) with D = 90 representing axial descent. A summary
of several VRS boundaries from previous studies is included in Fig. 1.
What follows is a brief description of the various criteria that have been
used by the researchers in arriving at the VRS boundaries shown as
subplots in Fig. 1.
Region of roughness
From the investigation of a rotor in the VRS (Ref. 13), Drees and
Hendal identified a region of roughness. In this region, the rotor behavior
was rough with respect to attitude and control. Unexpected loss of altitude
and a large nose-down pitching motion were also observed. The region
of roughness ranged from VZ /Vih = 0.62 to VZ /Vih = 1.53 in axial
descent, extending in inclined descent to VH /Vih = 1.0.
Thrust fluctuation
Washizu et al. conducted an experiment for a rotor in descent conditions (Ref. 3). Unlike other wind-tunnel tests, Washizus experiments
were performed using a track system to minimize facility effects that were
typically associated with wind-tunnel experiments. During the tests, the
VRS boundary based on the magnitude of T /T was derived, where
T and T are the amplitude of fluctuation and mean value of thrust,
respectively. Two reference values of T /T were constructed: 0.15 and
0.30. When T /T = 0.15, the corresponding boundary extended from
axial descent to inclined descent with the forward velocity component
VH /Vih < 1. When T /T = 0.30, the corresponding boundary mainly

APRIL 2008

PREDICTION OF VORTEX RING STATE BOUNDARY OF A HELICOPTER IN DESCENDING FLIGHT

0.5

0.5

0.5

ih

V /V

1.5

2
0.5
1
V /V
H

2.5

ih

0.5
1
V /V
H

0
Peters

0.5

ih

1.5

2.5

ih

1.5

1.5

0.5
1
V /V
H

1.5

2.5

ih

1.5
2

0.5
1
V /V
H

1.5

2.5

ih

2
2.5
0

0.5
1
V /V
H

1.5

1.5

French Flight tests: Vz drop


Fluctuations increase
Vz stabilization
ONERA VRS model

2
2.5

V /V

V /V

ih

ih

VZ /V ih

0.5

1.5

0.5
1
V /V
H

ih

1.5

0.5
1
V /V

1.5

ih

Thrust fluctuations < 2.5%


Thrust fluctuations > 2.5%
Asymmetries
V-22 roll-off
Newman

2
2.5

Betzina (CT /CTo < 1)


Betzina (0.6 < CT /C To < 0.7)

0.5

Leishman, 10%
Leishman, 12%
Leishman, 15%
Leishman, 20%

1.5

ih

0.5

1.5

0.5
1
V /V

0.5

2
0

Wolkovitch upper limit


Wolkovitch lower limit

Xin and Gao

0.5
VZ /V

VZ /Vih

1.5

Washizu (T/T = 0.15)


Washizu (T/T = 0.30)

ih

2.5

2.5

1.5

1.5

V /V

VZ /Vih

ih

Drees

V /V

141

ih

0.5
V /V
H

1.5

ih

Fig. 1. A summary of VRS boundaries from the literature.


covered the inclined descent region with the descent angle ranging from
45 to 80 .
Wolkovitch criterion
To predict the VRS boundary in descent condition analytically,
Wolkovitch (Ref. 22) considered a flow model consisting of a slipstream
with uniform flow at any section of the rotor surrounded by a protective
tube of vorticity. The tube was made up of tip vortices leaving the rotor. It was assumed that unsteady vortex ring flow was associated with
a breakdown of this protective tube of vorticity. Consequently, VRS onset occurred when the net velocity of tip vortices became zero. With the
vertical convection speed of the tip vortices as the mean value between
the airspeed outside the wake (V Z ) and inside the wake (V Z + Vim ), the
condition for VRS entry corresponded to
VZ =

Vim
2

(1)

As for VRS departure, Wolkovitch utilized a coefficient k z to take into


account the distance above the rotor where the vorticity would accumulate, corresponding to
VZ =

k z Vim
, 1 kz 2
2

(2)

A value of 1.4 was used for k z by Wolkovitch. According to Ref. 12,


Wolkovitchs boundaries were close to those of experimental data at low
forward speed. Nevertheless, the boundaries were not consistent with
other experiments as VRS could be encountered at any forward speed.
However, Wolkovitchs ideas on critical vortex transport speed were later
extended by other researchers (e.g., Ref. 23).

Peters and Chen modified Wolkovitchs method by removing several


inconsistencies in the flow model and taking into account the wake skew
angle (Ref. 24). After modification, Peters boundary showed no VRS
for VH /Vih > 0.62 and predicted VRS over a wider range of VZ /Vih .

Torque fluctuation
Xin and Gao observed an irregular variation of the rotor torque at about
VZ /Vih = 0.28 (Refs. 6, 7). Torque fluctuations were more severe for
D = 60 and D = 75 than in axial descent. They further found that as
the descent angle decreased, the torque fluctuations also decreased and
finally disappeared below D = 40 .
On the basis of observations from the tests, Xin and Gao pointed out
that there were three problems associated with Peters VRS boundary.
First, Peters boundary showed that the rotor entered the VRS even for
small descent rates. Second, no occurrence of VRS existed for VH /Vih >
0.62. Third, VRS occurred at every descent angle. Xin and Gao thus
proposed an improved VRS boundary as shown in the subplot of Fig. 1.
This new boundary was more consistent with data from flight and model
tests.

Mean thrust reduction


Betzina observed a significant mean thrust reduction in VRS (Ref. 21).
As a result, the thrust ratio C T /C To was used as an indicator, where C T
and C To represented mean thrust coefficient and the thrust coefficient at
hover, respectively. It was shown that the lowest thrust ratio centered on
a descent angle of about D = 75 and VH /Vih = 0.3, and extended from
D = 60 to D = 90 .

142

JOURNAL OF THE AMERICAN HELICOPTER SOCIETY

P.-M. BASSET

Rotor Inflow Modeling

Blade-flapping fluctuation
Recently, a free-vortex wake method has been developed by Leishman
et al. (Ref. 8), which shows that in addition to thrust and torque fluctuation,
blade-flapping fluctuation may also be a concern as a result of unsteady
airloads found near or in the VRS. An excessive blade-flapping angle
(greater than 10% of the mean) may lead to piloting difficulties. As such,
contours of excessive blade-flapping fluctuation were used as indication
of VRS onset.
Newman criterion
Newman developed a wake transport criterion for VRS assessment
(Ref. 23) and defined an effective wake transport velocity, WTVE , as
follows:

WTVE = k 2 2x + (z + i )2
(3)
where k represents the relative effectiveness of the in-plane velocity component. A critical value of WTVE existed indicating an onset of flow
breakdown in the wake stream tube, denoted as WTVECRIT . The true wake
transport velocity at this critical condition was expressed with respect to
the critical effective velocity:

W T V = 2WTVECRIT + (1 k 2 )2x
(4)
On the other hand, the mean induced velocity i was represented as
i =

1
WTV

Thus, the boundary for the onset of flow breakdown was given by

z = 2WTVECRIT k 2 2x i

(5)

(6)

On the basis of experimental observations from Drees and Hendal (Ref.


13), Newman selected the values of the empirical factors as k = 0.65
and WTVECRIT = 0.74.

As simple momentum theory is no longer valid for a rotor in VRS,


modeling of rotor inflow in VRS becomes very challenging. Following
is a discussion of two different rotor inflow models: the ONERA inflow
model and the Georgia Tech ring vortex model.
ONERA inflow model
ONERA developed an inflow model (Ref. 19) based on experimental
data, mainly the Dauphin 365N flight test data (e.g., Ref. 11), as this
helicopter has been the one used for research at the French flight test
center in Istres. The purpose of this model was to provide an extension
of the momentum theory that could be applied in steep approaches, even
in case of VRS. The approach for building the model was first arrived
at for axial descent using the wind-tunnel data of Ref. 1 as sketched in
Fig. 2. The model was then further developed by using the Dauphin flight
test data. A first harmonic inflow model was built such that the power or
collective curves obtained by trim computations in descending flights at
different forward speeds matched the ones measured in flight tests. The
first harmonic inflow gradients were based on the Meijer-Drees model
with a modification of the longitudinal inflow gradient (VicMR ) in order to
avoid any discontinuity when the mean inflow became zero (during the
change between the helicopter and windmill modes).
The working steps of the ONERA inflow model are as follows: first,
at each forward speed below a critical advance ratio, an interpolation
is performed between the helicopter and windmill branches (for higher
forward speeds, the momentum theory can still be applied). From these
two branches, two points are identified (e.g., points 1 and 2 as shown in
Fig. 2). At each of the two points, two conditions are imposed to ensure
continuity and differentiability of the function Vim = f (V Z ). These four
conditions enable the calculation of the four coefficients of a third-order
interpolating polynomial. Second, the mean inflow given by the momentum theory is augmented by an extra term of induced flow. Indeed, most
of the wind-tunnel data (e.g., Ref. 1) show that the mean inflow is stronger
in descending flight than the inflow predicted from momentum theory.
During descent, tip vortices merge in a vortex ring at the periphery of the

Heave stability
During a recent ONERA flight test, several behaviors were observed,
especially in the VRS region (Ref. 12). During the phase leading to VRS,
the crew would first feel an increased level of vibration, followed by a
sudden increase in the rate of descent. Increasing the collective would not
prevent the helicopter from a further increase in its descent rate. During
the descent, the helicopter was very unstable and was difficult to control.
The flight tests also established that the VRS effects disappeared beyond
certain forward speed. The final VRS boundary was determined based
on the following criteria: (1) an increased level of vibration, (2) a sudden
increase in the descent rate, and (3) exiting from VRS due to stabilization
of the descent rate.
Roll stability
The goal of the testing of the V-22 tilt-rotor aircraft was to determine
the criteria for a quasi-steady state VRS boundary (Refs. 14, 15). With an
increase in descent rate from hover, the test team first observed an increase
of thrust fluctuations was first observed. As the situation degraded, the
pilots experienced an uncommanded roll response, which defined the
VRS boundary. Figure 1 shows Newmans VRS boundary superimposed
on the V-22 test data, indicating that the VRS boundaries for tilt-rotor
aircraft and for conventional helicopters were remarkably similar.

Fig. 2. Rotor-induced velocity variation with descent rate from the


ONERA inflow model (example of comparison in vertical descent
with experimental data from Ref. 1).

PREDICTION OF VORTEX RING STATE BOUNDARY OF A HELICOPTER IN DESCENDING FLIGHT

rotor, create air recirculation, and thus increase the induced flow. This
extra induced flow is represented in the model by a transition function
with three main parameters: a maximum magnitude term (A0 ), a vertical
speed (Vz0 ) at which the magnitude reaches its maximum (A0 ), and an
exponent on which depends the curvature of the added nonlinear function (Refs. 19, 25). Both A0 and Vz0 depend on forward speed. These
parameters are determined to match the flight test data. Moreover, the
model is formulated using nondimensional variables. Thus, the model
corresponds, in principle, to a generic two- to four-ton class helicopter.
One advantage of using flight test data is to avoid the uncertainties of
wind-tunnel measurements regarding the proximity of the tunnel walls
which may interact with the large airflow recirculating around the rotor in
the VRS (as is questionable in Ref. 1). However, any kind of experimental
test has its drawbacks. In flight tests, for example, one main difficulty is
to assess the induced flow through the rotor. One method is to deduce
mean inflow from the measured main rotor power after subtracting from
it an estimate of profile power. Another method is to make use of the
in-flight airflow measurements below the rotor as described in Ref. 12.

143

3.5
Test, VH /Vih = 0.0
Test, V /V
H

ih

= 0.23

Test, VH /Vih = 0.69


Test, V /V

ih

= 0.92

Test, VH /Vih = 1.16


Test, V /V
H

ih

= 1.85

RVM, VH /Vih = 0.0

2.5

RVM, V /V = 0.23
H

ih

RVM, VH /Vih = 0.69

RVM, V /V = 0.92
H

ih

ih

ih

RVM, V /V = 1.16

RVM, V /V = 1.85

Vi /Vih

APRIL 2008

1.5

0.5

Ring vortex model


0
2.5

One drawback of momentum theory is its inability to account for


the interaction between the rotor wake and the surrounding airflow in a
descending condition (Ref. 26). The effects of this interaction may be less
significant at hover or in climb. Nevertheless, as a helicopter increases
its descent rate, the interaction becomes more and more intense due to
larger velocity gradients between the upflow outside the wake and the
downflow inside the wake. To take these flow interactions into account,
several of the current authors proposed a new inflow model at the Georgia
Institute of Technology (Refs. 2729). The so-called ring vortex model
supposes that, due to the flow interaction, there exists a series of vortex
rings located at the rotor periphery. These vortex rings move downward
along the wake when the helicopter descends at a low rate. As the rate
of descent increases, the vortex rings tend to accumulate near the rotor
tip. As the rate of descent increases, the vortex rings move upward along
the wake. A new vortex ring is formed with every blade rotation, i.e.,
2 /
Nb second. Therefore, the locations of these discrete vortex rings
can be determined by the product of convection velocity of the vortex
rings and 2m/
Nb (where m is an integer representing the numbering
of vortex rings).
Each vortex ring induces additional normal velocity at the rotor disk.
The flow field of a vortex ring can be computed based on elliptic integrals
(Ref. 30). One advantage of utilizing vortex rings is that the effect of
vortex rings is nonuniform with respect to relative distance between the
rings and the rotor disk. The closer a vortex ring is to the rotor disk,
the larger the magnitude of normal velocity at the disk. The resulting
nonuniform effect conforms to what has been observed in test data. Also,
using a random number of vortex rings, it is possible to capture the scatter
in induced velocity measurements observed in experiment.
In the ring vortex model, downward velocity due to vortex rings is
integrated into the induced velocity calculated by momentum theory.
The concept works well in the VRS and windmill phases. Nevertheless,
in axial descent and inclined descent at low forward speed (VH /Vih up
to 0.6204), momentum theory fails to predict a transition phase between
the helicopter and the windmill branches. To address this problem, the
simple momentum theory equation for induced velocity is modified to

2

+ 2 + ( + )2 = 1
(7)
2.72(1 + 2 )
where , , and are the normalized values (normalized by Vih ) of
descent rate, induced velocity, and forward speed, respectively. The ad-

1.5

1
V /Vih

0.5

0.5

Fig. 3. Normalized inflow curves calculated with the ring vortex


model for a generic helicopter (test data source obtained from Ref.
20).

2
ditional term ( 2.72(1+
2 ) ) , which is analogous to the parachute drag term,
modifies equilibrium curves for inflow dynamics, creating a steady-state
transition between helicopter and windmill branches in axial and steep
descents. Its effect diminishes at other flight conditions.
To demonstrate the effectiveness of the ring vortex model, normalized inflow curves are calculated for a generic helicopter in Fig. 3 (test
data source from Ref. 20). Note that the forward speeds are kept constant during the test with Vx /Vh at 0.0, 0.23, 0.69, 0.92, 1.16, and 1.85.
Generally, results from the ring vortex model agree with the experimental data. At high forward speed (Vx /Vh = 0.92, 1.16, 1.85), predicted inflow curves from the ring vortex model tend to get closer with
simple momentum theory. In these conditions, vortex rings are further
swept away from the rotor disk and their influence on rotor inflow is
diminished.

Methodology
To identify by simulation the sensitive parameters for the VRS
phenomenon and their associated critical values (which may lead to
an entry into this regime), two complementary approaches can be
envisaged:
1) Equilibrium computations can be conducted to reproduce the conditions of steady descending flight by sweeping a large set of velocity
couples (V H , V Z ) and by changing the value of one parameter, such as
gross weight or other flight conditions.
2) Time simulations of different types of descending flight (constant
slope, segmented, or decelerated approaches) can be performed with a
rotorcraft flight dynamics model coupled with an automatic pilot module.
The present paper focuses on the first approach (equilibrium computation). The second one (time simulations) is addressed in Ref. 31. Indeed,
the objective of this paper is to compare three methods and associated criteria for the prediction of VRS boundaries in steady flight condition with
an unaugmented helicopter, i.e., without an automatic pilot or a stability
augmentation system.

144

P.-M. BASSET

JOURNAL OF THE AMERICAN HELICOPTER SOCIETY

The three methods discussed in detail hereafter are as follows:


1) Point-by-point method: a NewtonRaphson algorithm associated
with tip vortices criterion;
2) Bifurcation method: a continuation algorithm associated with bifurcation criterion;
3) Heave stability method: a NewtonRaphson algorithm associated
with heave stability criterion.

will encounter the VRS) can be assessed by using = 0.2. A wider risk
zone can be obtained by using = 0.3, for which all the test points are
inside the predicted domain. In the rest of the paper, is set to 0.25 as it
is the value for which the limits given by this criterion pass through the
flight test points and may be considered as the VRS onset boundaries

Point-by-point method

The principle of this method is presented in Ref. 25. Here, only the
main concepts are restated. In the study of nonlinear system dynamics,
the well-suited continuation algorithm (described, e.g., in Ref. 33) enables the determination of an equilibrium curve even in the presence of
folds (vertical tangents), which represent multiple equilibrium points for
the same set of parameters. Such complex patterns may occur depending
on the nonlinear terms within the system dynamics. Rotorcraft flight dynamics is well known for involving many nonlinear effects, for example,
the complex aerodynamics of rotor blades, its interactions with the other
components (airframe, tail rotor, etc.), inertial coupling, and so on. However, the application of nonlinear analysis to rotorcraft flight dynamics
remains rare (Refs. 10, 25, 3437).
The numerical implementation of the continuation algorithm in
ASDOBI (an ONERA program developed to study fixed-wing flight dynamics, Refs. 38, 39) is based on the repetition of four steps (see Fig. 4):
Step 1: finding a point on the equilibrium curve,
Step 2: predicting the direction of the tangent,
Step 3: predicting the next point, and
Step 4: correcting the predicted point.
Step 3 is based on an AdamsBashforth-like integration method,
whereas steps 1 and 4 are based on the NewtonRaphson scheme adapted
to the implicit problem of n equations with (n + 1) variables. Here,
n is the number of states, and the (n + 1)th variable is a parameter
or control, here designated by u. As can be seen, the continuation algorithm is more advanced compared with the basic NewtonRaphson
algorithm.
.
By denoting X as the state vector, X as its time derivative, U as the
.
control vector, X = F(X, U ) as the set of differential equations, and
(X 0 , U0 ) as an equilibrium solution, the stability property of each equilibrium point (X 0 , U0 ) can be characterized by calculating the Jacobian
matrix of the system dynamics:


Fi (X 0 , U0 )
(9)
D X F(X 0 , U0 )
xj

The helicopter overall simulation tool (HOST code created by Eurocopter, Ref. 32) makes use of the NewtonRaphson algorithm for trim
computation similar to most rotorcraft flight dynamics codes. The sixdegree-of-freedom rigid body dynamics are driven to steady state in an
iterative process searching for the values of the four controls (collective,
longitudinal cyclic, and lateral cyclic controls of the main rotor and collective control of the tail rotor), as well as two attitude angles (pitch and
roll angles). With this method, the trim is calculated point by point for
each flight condition.
With the HOST code, a series of trim sweeps are conducted for each
value of the studied parameters:
1) sweep of forward speed V H from 0 to 50 km/h with a step of 5 km/h,
and
2) for each forward speed, a sweep of descent rate V Z from 0 to
25 m/s with a step of 1 m/s.
Each series of trim sweeps contains 286 trim calculation points. Each
of these points corresponds to a steady descending flight condition with
a slope given by the (V H , V Z ) couple. At each equilibrium point, the tip
vortices criterion is used to determine whether this flight condition may
risk a VRS onset.
The tip vortices criterion is an extension of the Wolkovitch criterion
(Ref. 22). As mentioned earlier, the basic principle is that the rotor may
encounter a VRS situation when the convection speed of the main rotor
blade tip vortices is too small, resulting in vorticity accumulation around
the rotor and air recirculation. The vertical convection speed of the tip
vortices is assumed to be the mean value between the airspeed outside
the wake (V Z ) and the airspeed inside the wake (V Z + Vim ).
As pointed out earlier, the Wolkovitch criterion does not take into
account the effect of forward speed very well. As a result, this criterion
predicts that a VRS may occur at any forward speed (see Fig. 1). However,
such an occurrence is not physically feasible because above a certain
forward speed, the vortex wake is swept backward preventing the rotor
from flying in its own wake.
The extension of the Wolkovitch criterion has already been presented
with the consideration of wake skew angle (Refs. 12, 19).
of travel
 In terms

distance
with
respect
to
the
rotor
plane,
the
in-plane
V
and
normal
tv
X


Vtv Z components of the tip vortices velocity do not have similar effects. Therefore, the modified criterion has the following nondimensional
form:




Vtvz 2
Vtvx 2
VTVDES =
+

(8)
k x Vih
Vih

where Vtv X = (Vx,ROTOR )2 + (Vy,ROTOR )2 and Vtv Z = VzROTOR + V2im .
Comparisons with experimental data (wind-tunnel and flight test data)
have led to the following empirically tuned values. The weight coefficient
on the in-plane component k x is set at 4, while is determined to be 0.25
by giving an overall agreement with the experimental data. One note of
caution is that the ONERA VRS model in Fig. 1 adopted a value of
0.2. This parameter provides a way of bringing a certain margin with
respect to the VRS domain. Higher values of give rise to predictions
of wider VRS boundaries. The core VRS area (in which the aircraft

Bifurcation method

If one eigenvalue of the Jacobian matrix has a positive real part, the
system is unstable at this equilibrium point. In both Figs. 4 and 5, the
upper and lower parts of the equilibrium curve are stable, whereas the
middle part within the fold is unstable.

Fig. 4. Continuation algorithm for calculating an equilibrium curve.

APRIL 2008

PREDICTION OF VORTEX RING STATE BOUNDARY OF A HELICOPTER IN DESCENDING FLIGHT

145

loci are nearly identical in the classical (V H , V Z ) diagram shown as the


first subplot in Fig. 6.
Our first assumption seems, therefore, to be valid: the flapping and
lead-lag dynamics do not play a significant role in the VRS boundaries in
the (V H , V Z ) diagram, at least in the case of the helicopter studied. Moreover, the computational time required to calculate the bifurcation locus
with the blade element model is excessive compared with the analytical
rotor model. Hence, for the subsequent HOST and ASDOBI bifurcation
computations, the analytical rotor model with quasi-steady flapping was
selected as in Ref. 25.
Heave stability method
From the inflow velocity results in Figs. 2 and 3, it is noted that
there is not only an increase in the magnitude of induced velocity over
the prediction from momentum theory but also a steeper and varying
gradient of inflow curve.
According to Ref. 27, the derivative of the inflow curve from momentum theory can be obtained as follows:

Fig. 5. Calculation of bifurcation locus.

Bifurcation points are special equilibrium points that indicate a change


of system stability. On the equilibrium curve of both Figs. 4 and 5, the
black points between the stable and unstable parts are called turning
points. These points, also known as saddle nodes, belong to the family of
the real bifurcations associated with the nullification of a real eigenvalue.
The mathematical translation of this property involves adding one extra
equation to the n equations of the equilibrium problem. The equilibrium
point (X 0 , U0 ) is a real bifurcation (i.e., turning point), if the Jacobian
matrix has at least one zero eigenvalue. Thus, the extra equation, for
characterizing the real bifurcation points, is set by writing that the determinant of the Jacobian matrix is null. Hence, the bifurcation locus can be
found if the continuation algorithm is applied to this augmented system
of (n + 1) equations by considering two control parameters (u1 , u2 ) to
set implicit problem ((n + 1) equations for (n + 2) variables); see Fig. 5.
As described in Ref. 25, the rotorcraft flight dynamics simulation
code HOST was coupled with the ONERA ASDOBI nonlinear analysis
tool. The inflow gradients VicMR and VisMR of the first harmonic inflow
model are always included (in Ref. 25 and the present work). In the first
application of this method to the study of flight dynamics in descending
flight (Ref. 25), the first level of the rotor models available in HOST was
used, i.e., an analytical rotor disk model with quasi-steady blade-flapping
dynamics. Later, second-order blade flap and lead-lag dynamics were
considered by using the blade element model. As usual, the flapping ()
and lead-lag ( ) motions were described up to the first harmonics leading
to 12 more states:
X = {Uhel , Vhel , Whel , Phel , Q hel , Rhel , , ,
VimMR , VicMR , VisMR , Vim F AN ,

d
1

= + 
d
2 2 2 + 4

(11)

d
In the range from = 0 to = 1.5, the absolute value of d
from
momentum theory is always less than 1. This indicates that with momentum theory, a change in descent rate is associated with || < ||,
whereas with either the ONERA inflow model or the ring vortex model,
the result is || > || (for values of roughly in the range of 0.5
to 1.5).
The overall effect of a decrease in descent rate ( > 0) on blade angle of attack () and hence on heave damping ( T /) with momentum
theory is




d
  1 +
< 0 T / < 0
(12)
d

Here, the symbol stands for proportional to. With either the ONERA
inflow model or the ring vortex model, the overall effect is




d
> 0 T / > 0
(13)
  1 +
d
Thus for values of roughly in the range of 0.5 to 1.5, with momentum
theory, a decrease in descent rate results in a decrease in rotor thrust.
This is the case where the vehicle vertical dynamics has stable heave
damping, i.e., T / < 0. However, with the ONERA inflow model or
the ring vortex model, an increase in descent rate from a value of in the
above-mentioned range gives rise to a decrease in rotor thrust, resulting
in unstable heave damping, i.e., T / > 0.

(10)

0 , 1c , 1s , 0 , 1c , 1s , 0 , 1c , 1s , 0 , 1c , 1s }
The effect of the rotor model on the bifurcation loci is shown in Fig.
6. This is for a four-bladed generic helicopter with a gross weight of
3500 kg. The curves show the flight conditions for which there would
be a change of stability in descending flight. The bifurcation loci with
continuous and dotted lines were calculated with the analytical rotor disk
model (a continuous line for quasi-steady and a dotted line for first-order
flapping dynamics). These two results are nearly the same as using a
blade element model (shown as a dashed line in Fig. 6), except that it is
seen that the type of rotor model (either a rotor disk model or a blade
element model) plays a role in the prediction of the values of the trim
variables, i.e., trim control and attitude angles. However, the bifurcation

Results and Discussion


ONERA (Ref. 12) conducted flight tests of a helicopter in steep descents. The 11.93-m-diameter rotor has a solidity of 0.083 and a 10.2
blade twist.
VRS boundaries from tip vortices and bifurcation criteria
obtained by ONERA
For the prediction of VRS boundaries based on the tip vortices criterion and the bifurcation criterion, the examples of calculations presented
here by ONERA are for the cases of the helicopter in steady descending
flight at sea level (in International Standard Atmosphere condition) for

146

JOURNAL OF THE AMERICAN HELICOPTER SOCIETY

P.-M. BASSET

VZ m/ s

Quasi-steady blade flapping


Flapping velocity
Flap and lead-lag dynamics

VRS domain

DT0 DEG

Collective pitch

DTC DEG

5
12

16

0
DTS DEG

10

20

30
VH km/ h

Longitudinal cyclic pitch

30

2.4
2
1.6

10

DTA DEG

20

30
VH km/h

Tail rotor collective pitc h

DEG

10

20

4
3

10

20

30
VH km/h

20

30
VH km/h

20

30
VH km/h

Bank angle

1.2

0.8

0.4

30
VH km/h

10

20

10

VHEL M/S

WHEL M/S Vertical helicopter speed

DEG

UHEL M/S Longitudinal helicopter speed

20

10

30
VH km/h

Pitch angle

10

26

1.6
0

28

16

22

0.8

2.4

24

1.2

Lateral cyclic pitch

Analytical rotor disk model


Blade element model

30
VH km/h

Lateral helicopter speed

VIM-MR M/S Main rotor mean induced flo w

10

20

30
VH km/h

VIM-FAN M/S Tail rotor mean induced flow

28

48

12

24

44

20

40

16
4

10

20

30
VH km/h

10

20

30
VH km/h

36

10

20

30
VH km/h

Fig. 6. Comparison of the bifurcation loci calculated with different rotor models.
all these variations lead to an increase of the mean induced flow and thus
also to the increase of the critical descent rate for a VRS onset. However,
working with nondimensional parameters, i.e., dividing the speeds (V H ,
V Z ) by the theoretical mean inflow in hover (Vih ) (which is related to the
thrust coefficient CT ), tends to make the VRS domain less dependent on
gross weight, altitude, and temperature. This comes from the fact that Vih
and CT vary with gross weight, altitude, and temperature:

0
Tip vortices criterion ( = 0.25)
Max. slope,(M = 2850 kg): 27.22

0.2

Max. slope,(M = 3850 kg): 27.22


0.4

Vz /Vih

0.6
+: Flight tests

0.8


Vih =

GW
2 S

(14)

1.2
Bifurcation criterion
Max. slope,(M = 2850 kg): 45.07

1.4

Max. slope,(M = 3850 kg): 43.58


1.6

0.5

1.5

VH /Vih

Fig. 7. Comparison of the VRS boundaries computed with tip vortices


and bifurcation criteria.
two different gross weights (GW): 2850 and 3850 kg. The results are
shown in Fig. 7, together with the flight test data from Refs. 12 and 19.
According to the model, increasing gross weight has the same global
effect on the VRS domain as increasing altitude or temperature. Indeed

The discrepancy at the lower part of the VRS domain (high descent
rates and low forward speeds) in Fig. 7 is explained in Ref. 25. During
the flight tests, the pilot identified the exit of the VRS in terms of stabilization of descent rate in the windmill mode following the drop from
the helicopter mode. As sketched in Fig. 8, the folds of the equilibrium
curves at the bifurcations from the helicopter mode (circle in Fig. 8)
correspond naturally to higher descent rates than those at the actual bifurcations from the windmill mode (square in Fig. 8). HOSTASDOBI
results highlight that from the windmill branch the bifurcation toward
the helicopter branch occurs for lower descent rates due to the hysteresis
effect. On the other hand, the tip vortices criterion is adjusted by tuning
coefficients of the Wolkovitch criterion based on the flight test data. As a

APRIL 2008

PREDICTION OF VORTEX RING STATE BOUNDARY OF A HELICOPTER IN DESCENDING FLIGHT

147

Fig. 8. Why the bifurcation criterion predicts lower descent rates for the lower limit of the VRS boundary.

result, the boundary from the tip vortices criterion envelops these points
at the lower boundary.
Another discrepancy in Fig. 7 concerns the upper limit. For forward
speeds (V H /Vih ) approximately between 0.5 and 1, the VRS boundary
from the tip vortices criterion occupies higher portion of the figure over
the one from the bifurcation criterion. This difference plays a prominent
role in the determination of the predicted maximum descent slope. It
should be noted that the maximum slopes in Fig. 7 are large, but could
be lower in practice in a tail-wind condition.
With the tip vortices criterion, the descent ratio (V Z /Vih ) for a potential
VRS onset becomes lower in absolute value when the forward speed
increases. This tendency comes from the Wolkovitch criterion, the basis
for the tip vortices criterion (see the graph at the right upper corner of
Fig. 1).
In the applied tip vortices criterion, a condition is added onto the
in-plane velocity component (Vtv X ) such that a VRS onset is no longer
predicted above a certain forward speed. But the vertical component
(Vtv Z ) of the tip vortices velocity still depends on the half value of the
mean inflow (Vim /2), as mentioned in the discussion of the Wolkovitch
criterion. In the model, the mean inflow decreases with forward speed.
Hence, the critical value of the descent rate for a VRS onset becomes
lower in absolute value when forward speed increases. More precisely,
the minimum of the tip vortices criterion (VTVDES in Eq. (8)) is
reached at a lower descent rate when forward speed is increased (Fig. 9).
The derivative of this criterion with respect to the descent rate is null
when
d (VTVDES )
=0
d Vz

if

d Vim
= 2
d Vz

or

Vim = 2Vz

(15)

It can be seen in Fig. 9 that the minima of the tip vortices criterion are
reached when Vim = 2Vz . Besides, when forward speed increases, the
mean rotor inflow (Vim ) is lower and increases more slowly with descent
rate. Therefore, the inflow condition (Vim = 2Vz ) is obtained by the
model at a lower descent rate when forward speed increases. However,
this effect of forward speed on the variation of the minimum of the
criterion is reduced or truncated by using an parameter: the VRS is
predicted not at the minimum of VTVDES, but when its value becomes
lower than (see Fig. 9, VRS-MR = 1 if VTVDES < ). This is why the
upper limit of the VRS domain given by the tip vortices criterion remains
more or less horizontal. The corresponding critical descent rate for the

VRS entry may be slightly less (Fig. 7), mainly because of the decrease
of the mean inflow with forward speed as mentioned before.
The bifurcation criterion predicts the contrary: the higher the forward
speed, the higher the critical descent rate (from which a VRS may occur).
This prediction is not as easy to explain because it relies on a numerical
criterion rather than on an explicit analytical criterion such as the tip
vortices criterion. The bifurcation criterion is based on the change in
stability of the trim solutions, determined by computing the eigenvalues
of the Jacobian matrix of the helicopter flight dynamics. The discrepancy
on the upper limit may be explained by looking at the turning points of
the function Vz = f (DT 0) (as done for the lower limit differences in
Fig. 8). The bifurcation from the helicopter branch also corresponds to
the minimum of the required collective for trimming the helicopter when
the collective is decreased from a hover or level flight condition, (see DT 0
= f (Vz ) in Fig. 9). When forward speed increases, this turning point of
the Vz = f (DT 0) function or local minimum of the DT 0 = f (Vz ) function
occurs at a higher descent rate (see Fig. 9, the differences between the
trim curves for V H : 5, 20, and 30 km/h). The trim collective follows the
similar variation trend as the main rotor power (power coefficient CP-MR
in Fig. 9) and the average axial rotor inflow (Vim + Vz ). As indicated in
Ref. 25, because this function reaches the minimum in question when
d Vim
d (Vim + Vz )
= 0 if
= 1
d Vz
d Vz

(16)

the bifurcation criterion provides similar results in descending flights


as the heave stability criterion, the dominant dynamics of such flight
conditions.
In both cases, tip vortices and bifurcation criteria, the ONERA inflow
model is utilized. Yet the coefficients of the tip vortices criterion (kx ,
) are tuned to cover the VRS domain identified in the flight test data,
whereas the bifurcation criterion reflects the stability of the helicopter
flight dynamics model.
To illustrate this fundamental difference between the two criteria, time
simulations were performed from three different descending flight initial
conditions, as shown in Figs. 10 and 11.
At the beginning of the time simulations with the HOST code, a
minor decrease of the collective is applied from its trim value. From the
initial approach conditions (a) and (c) (Fig. 10), the minor perturbation
on the collective eventually leads to a large increase of the descent rate,
symptomatic of a VRS onset (Fig. 11). In case (a), a longer time is

148

JOURNAL OF THE AMERICAN HELICOPTER SOCIETY

P.-M. BASSET

Vh=5 km/h sweep on VZ


6

DT0 DEG

Vh=20 km/h sweep on VZ

Collective pitch

0.6

5 km/h

Vh=30 km/h sweep on VZ

VTVDES-MR S.U.

Tip vortices nondimensional speed (Eq. (8))

0.4

20 km/h
4

0.2

30 km/h
3
16

12

V +V m/s
14 im z

VZ m/s

Mean axial airflow through the rotor

0
16

VRS-MR S.U.

12

Tip vortices VRS criterion

VZ m/s

1 in VRS

0.8

12

10
0.4
8
6
16
4E4

12

CP-MR S.U.

VZ m/s

Main rotor power coefficient

0
16
30

VIM-MR M/S

0 out of VRS
12

Main rotor mean induced flow

VZ m/s

3.6E4
3.2E4

20

2.8E4
2.4E4
16

12

0
VZ m/s

10
16

12

VZ m/s

Fig. 9. Why the bifurcation criterion predicts higher descent rates for the upper limit of the VRS boundary in forward flight.
(a) : VZ = 6 m/s, VH = 20 km/h (b) : VZ = 6 m/s, VH = 30 km/h (c) : VZ = 10 m/s, VH = 30 km/h
-4

VZ m/s

(b)

(a)

-6

-8

(c)

-10

-12

-14

-16
0

10

15

20

25

30

35
VH Km/h

Fig. 10. Initial conditions for the subsequent tests in time simulation.

needed for the VRS onset compared to case (c). This is well predicted
by the bifurcation criterion considering that both points are close to the
bifurcation locus surrounding the VRS domain: case (a) is just outside
the domain, whereas case (c) is slightly inside (see Fig. 10). Thus, it is
logical that more time is needed for a VRS onset for case (a) than for case
(c). Variations of descent rate are also well predicted by the bifurcation
locus compared with the time simulations (larger Vz variation in case (a)
than in case (c)), although as explained in Fig. 8, the helicopter stabilizes

in the windmill regime at higher descent rates than the lower part of the
bifurcation locus (after a bifurcation from the helicopter mode).
For case (b), the initial approach condition is outside the bifurcation
locus, but inside the VRS domain given by the tip vortices criterion (see
Fig. 7). Even with a reduction of the collective twice the value applied
for cases (a) and (c), the helicopter remains stable and no VRS entry is
observed in time simulations.
Therefore, the bifurcation method gives a way to determine the conditions at which the rotorcraft would be naturally unstable (without stability
augmentation system or automatic pilot). More precisely, the bifurcation
locus surrounds the conditions for which the helicopter model is potentially unstable (high descent rate after a small collective pitch drop).
The bifurcation criterion strictly reflects the change in the stability of the
model. In contrast, the tip vortices criterion may predict VRS onset even
when the flight dynamics predicted by the model is stable. Thus, with
a perfect model, the bifurcation criterion would be preferable. But the
tuning coefficients of the tip vortices criterion may allow to compensate
the weakness of the model.
VRS boundary from the flight dynamics stability criterion
obtained by Georgia Tech
The VRS boundary obtained from the flight dynamics stability criterion was calculated separately at Georgia Tech. While the Georgia Tech
basic flight dynamics model is similar to that of ONERAs helicopter
model, the biggest difference is the adoption of the ring vortex model.

APRIL 2008

PREDICTION OF VORTEX RING STATE BOUNDARY OF A HELICOPTER IN DESCENDING FLIGHT

(a) Descent VZ =6 m/s - H = 0M ISA M = 3.5T VH = 20 km/h


VRS Onset
4 VZ m/s

4.34 DT0 DEG

21.6 VH km/h
21.2

4.32

149

20.8
12
20.4

4.3
16
4.28

40

80

DT0 DEG

t (s)

40

80

t (s)

(b) Descent VZ =6 m/s - H = 0M ISA M = 3.5T VH = 30 km/h


V m/s
0 Z

20
0

40

80

40

80

40

80

t (s)

31 VH km/h

0.8
4
0.4
30

12

-0.4
0

40

80

t (s)

40

80

t (s)

(c) Descent VZ =10 m/s - H = 0M ISA M = 3.5T VH = 30 km/h


V m/s
4 Z

3.58 DT0 DEG

VRS onset

29
0

t (s)

31 VH km/h

3.56

30

12
3.54
16
3.52

40

80

t (s)

40

80

t (s)

29
0

t (s)

Fig. 11. Tests in time simulation from the three initial conditions presented in Fig. 10.

Comparisons between the ONERA inflow model and the ring vortex
model can be summarized as follows.
The ONERA inflow model is a three-state first harmonic inflow model
(with the VRS effect globally taken into account for the mean inflow
term). The finite-state inflow model (Ref. 40), on the other hand, can
be selected as a baseline inflow module over momentum theory in the
ring vortex model. Effects from blade taper, blade twist, and forward
speed can be investigated with the finite state inflow model. The finite
state inflow model accounts for a three-dimensional unsteady wake for
a lifting rotor based on the unsteady potential flow theory (Ref. 40).
Similar to the adjustment associated with momentum theory, the mass
flow parameter (VT ) in the finite state inflow model can be modified
to accommodate a steady-state transition. It will be essential to include
a reasonable number of inflow states for future studies. In this study,
seven inflow states are chosen for both radial and azimuthal variation
up to first harmonic. Radial variation accounts for the spanwise distribution of inflow due to blade taper and blade twist, whereas azimuthal
variation takes care of the effect of the forward velocity. In the ring vortex model, the actual number of vortex rings is allowed to vary within a
given range in the simulation. By doing so, it is theorized that vortex rings
may either survive beyond a nominal age with a slightly benign aerodynamic environment at a particular moment or prematurely burst due to a
slightly adverse environment at another moment. With a random number
of vortex rings, magnitudes of induced velocities are expected to scatter
around their nominal values. This provides a pattern of data fluctuation,
which can be utilized to compare rotor thrust and torque distribution from
experiments. A different number of vortex rings also leads to different
VRS boundaries. In this study, the nominal number of vortex rings is
chosen as two, while three vortex rings are also considered as a worst
case.

The helicopter model used by Georgia Tech is an in-house rigid body


flight dynamics model coded in MATLABTM . It includes three translational motions as well as roll and pitch motions. The blade-flapping
motion and yaw motion of the vehicle are assumed to be quasi-steady.
The aircrafts gross weight was selected at 3500 kg. Notice this gross
weight is slightly different from ONERAs selection in Fig. 7 (the top
gross weight being 3850 kg). The NewtonRaphson algorithm is utilized
for trim purposes. The eigenvalue for heave dynamics is subsequently
identified from the linearized model.
The predicted heave mode values (positive values indicating unstable
heave mode) are shown in Figs. 12 and 13. Two different numbers of
vortex rings are given in these two figures. From the damping results, VRS
boundaries can be described based on stability characteristics of heave
mode, as given in Fig. 14. In Fig. 14, the smaller boundary corresponds
to Nring = 2, whereas the larger boundary is Nring = 3. Two other VRS
boundaries corresponding to the tip vortices and the bifurcation criteria
are also provided in Fig. 14 for a gross weight of 3850 kg.
When Nring = 2, the predicted VRS boundary based on the heave
mode criterion appears circling around inner points of the test data. When
Nring = 3, the corresponding VRS boundary embraces almost all the test
points. A stability strip is created between these two boundaries. Within
the strip, even if the heave mode is stable at one moment, it may become
unstable at another moment due to random variation in the number of
rings associated with the unsteady nature of flow in the VRS.
In a broader view, the boundary with Nring = 3 provides an advance
warning for a potential VRS occurrence. The boundary with Nring = 2
gives the most confined condition. Consequently, these two VRS boundaries may provide certain clues to the design of a VRS detection and
avoidance system, although such a system may involve more complicated safety issues.

150

JOURNAL OF THE AMERICAN HELICOPTER SOCIETY

P.-M. BASSET

Among the three VRS boundaries in Fig. 14, the boundaries with
the bifurcation criterion and heave stability criterion show similar trends
with forward speed (except at low forward speed for the lower part of
the boundary with the bifurcation criterion (see the previous explanation
in Fig. 8). Both boundaries differ only in magnitude. This is no surprise
as both boundaries are obtained from flight dynamic stability analyses,
whereas the boundary with the tip vortices criterion is obtained from
aerodynamics.

0.1
VH /Vih = 0.0

VH /V ih= 0.10
VH /Vih = 0.23
V /V = 0.30
H

ih

Damping of heave mode

VH /Vih = 0.40
VH /Vih = 0.50

V /V = 0.60
H

ih

Concluding Remarks
0.1

0.2
1.5

0.5

This paper provides a brief review of various criteria used in the


literature for arriving at VRS boundaries of rotors in descending flight.
Three different criteria are considered to study the impact of the choice
of a specific criterion on VRS boundary predictions. The first criterion
involves zero transport velocity of rotor tip vortices. It is rooted in the
rotor flow behavior, and as such requires a detailed modeling of the flow
environment of a rotor in descending flight. The second criterion involves
bifurcation of descending flight equilibria. The third criterion involves
the loss of heave damping. The second and third criteria are rooted in
the loss of descent rate damping, and as such they require detailed flight
dynamics modeling in descending flight.
Two different rotor inflow models for descending flight previously
developed at ONERA and the Georgia Institute of Technology are used
in this study. The ONERA model, arrived at from test data, provides
an empirical extension of the momentum theory to descending flight.
The model is incorporated into the HOST flight dynamic simulation
of a generic helicopter for its use in VRS boundary predictions. The
Georgia Tech ring vortex model provides a heuristic extension of the
finite state inflow model to descending flight by considering additional
induced velocity due to a random number of vortex rings. The model is
incorporated into a MATLABTM based in-house flight dynamic model of
a generic helicopter for its use in VRS boundary predictions.
Results obtained in this study show that the VRS boundary predictions are consistent with flight test data. More importantly, in addition
to this consistency, it is concluded that different criteria lead to different VRS boundary predictions. The similarities and discrepancies have
been analyzed in this paper. It is suggested that future VRS studies take
into account the respective effect of the inflow model and of the specific
criterion used.

V /V
Z

ih

Fig. 12. Heave mode damping with Nring = 2.


0.8
V /V = 0.0
H

ih

VH /Vih = 0.23

0.6

VH /Vih = 0.50
VH /Vih = 0.60
VH /Vih = 0.69

0.4
Damping of heave mode

VH /Vih = 0.80
V /V = 0.92
H

0.2

ih

0.2

0.4

0.6

0.8
1.6

1.4

1.2

0.8
V /V
Z

0.6

0.4

0.2

ih

Fig. 13. Heave mode damping with Nring = 3.


0

0.2

Acknowledgments

0.4

The second author would like to acknowledge the financial support


from the DSO National Laboratories, Singapore, for his graduate study
program at Georgia Tech.

0.6

VZ /Vih

0.8

References
1

1.2

1.4
Flight test: Vz drop

1.6

Flight test: V stabilization


z

Tip vortices criterion


Bifurcation criterion
Heave stability criterion, Nring = 2

1.8

Heave stability criterion, N

ring

0.1

0.2

0.3

0.4

0.5
V /V
H

0.6

0.7

0.8

0.9

=3
1

ih

Fig. 14. Comparisons of VRS boundaries with tip vortices, bifurcation,and heave stability criteria.

Castles, W., Jr., and Gray, R. B., Empirical Relation between Induced Velocity, Thrust, and Rate of Descent of a Helicopter Rotor as
Determined by Wind-Tunnel Tests on Four Model Rotors, NACA TN
2474, October 1951.
2
Yaggy, P. F., and Mort, K. W., Wind-Tunnel Tests of Two VTOL
Propellers in Descent, NACA TN D-1766, 1962.
3
Washizu, K., Azuma, A., Koo, J., and Oka, T., Experiments on a
Model Helicopter Rotor Operating in the Vortex Ring State, Journal of
Aircraft, Vol. 3, (3), MayJune 1966, pp. 225230.
4
Empey, R. W., and Ormiston, R. A., Tail-Rotor Thrust on a 5.5-foot
Helicopter Model in Ground Effect, American Helicopter Society 30th
Annual National Forum Proceedings, Washington, DC, May 1974.

APRIL 2008

PREDICTION OF VORTEX RING STATE BOUNDARY OF A HELICOPTER IN DESCENDING FLIGHT

5
Wang, S. C., Analytical Approach to the Induced Flow of
a Helicopter Rotor in Vertical Descent, Journal of the American
Helicopter Society, Vol. 35, (1), January 1990, pp. 9298.
6
Xin, H., and Gao, Z., An Experimental Investigation of Model Rotors Operating in Vertical Descent, 19th European Rotorcraft Forum
Proceedings, Cernobbio, Italy, September 1416, 1993.
7
Xin, H., and Gao, Z., A Prediction of the Vortex-Ring State Boundary Based on Model Tests, Transactions of Nanjing University of
Aeronautics and Astronautics, Vol. 11, (2), December 1994, pp. 159
164.
8
Leishman, J. G., Bhagwat, M. J., and Ananthan S., Free-Vortex
Wake Predictions of the Vortex Ring State for Single-Rotor and MultiRotor Configurations, American Helicopter Society 58th Annual Forum
Proceedings, Montreal, Canada, June 1113, 2002.
9
Brown, R. E., Leishman, J. G., Newman, S. J., and Perry, F. J., Blade
Twist Effects on Rotor Behavior in the Vortex Ring State, 28th European
Rotorcraft Forum Proceedings, Bristol, UK, September 1720, 2002.
10
Basset, P. M., and Prasad, J. V. R., Study of the Vortex Ring State
Using Bifurcation Theory, American Helicopter Society 58th Annual
Forum Proceedings, Montreal, Canada, June 1113, 2002.
11
Jimenez, J., Desopper, A., Taghizad, A., and Binet, L., Induced
Velocity Model in Step Descent and Vortex-Ring State Prediction, 28th
European Rotorcraft Forum Proceedings, Bristol, UK, September 1720,
2002.
12
Taghizad, A., Jimenez, J., Binet, L, and Heuze, D., Experimental
and Theoretical Investigation to Develop a Model of Rotor Aerodynamics
Adapted to Steep Descent, American Helicopter Society 58th Annual
Forum Proceedings, Montreal, Canada, June 1113, 2002.
13
Drees, J. M., and Hendal, W. P., Airflow through Helicopter Rotors
in Vertical Flight, National Aeronautical Research Institute, NLL Report
V. 1535, Amsterdam, December 1949.
14
Brand, A., Kisor, R., Blyth, R., Mason, D., and Host, C., V-22
High Rate of Descent (HOD) Test Procedures and Long Record Analysis, American Helicopter Society 60th Annual Forum Proceedings,
Baltimore, MD, June 710, 2004.
15
Kisor, R., Blyth, R., Brand, A., and MacDonald, T., V-22 LowSpeed/High Rate of Descent (HROD) Test Results, American Helicopter
Society 60th Annual Forum Proceedings, Baltimore, MD, June 710,
2004.
16
Prouty, R. W., Helicopter Performance, Stability, and Control,
Krieger Publishing Co., Malabar, FL, 1995, pp. 109115.
17
Glauert, H., The Analysis of Experimental Results in the Windmill
Bake and Vortex Ring States of an Airscrew, R. & M. No. 1026, British
A.R.C., 1926.
18
He, C. J., Lee, C. S., and Chen, W. B., Finite State Induced Flow
Model in Vortex Ring State, Journal of the American Helicopter Society,
Vol. 45, (4), October 2000, pp. 318320.
19
Jimenez, J., Experimental and Theoretical Study of the Helicopter
Behavior in Steep Descent: Modeling of the Vortex Ring State, Ph.D.
Dissertation, University of Marseilles, December 2002.
20
Johnson, W., Model for Vortex Ring State Influence on Rotorcraft
Flight Dynamics, AHS Fourth Decennial Specialists Conference on
Aeromechanics, San Francisco, CA, January 2123, 2004.
21
Betzina, M. D., Tiltrotor Descent Aerodynamics: A Small-Scale
Experimental Investigation of Vortex Ring State, American Helicopter Society 57th Annual Forum Proceedings, Washington, DC,
May 2001.
22
Wolkovitch, J., Analytic Prediction of Vortex-Ring Boundaries for
Helicopters in Steep Descents, Journal of the American Helicopter Society, Vol. 17, (3), July 1971, pp. 1319.
23
Newman, S., Brown, R., Perry, J., Lewis, S., Orchard M., and Modha,
A., Predicting the Onset of Wake Breakdown for Rotors in Descending

151

Flight, Journal of the American Helicopter Society, Vol. 48, (1), January
2003, pp. 2838.
24
Peters, D. A., and Chen, S. Y., Momentum Theory, Dynamic Inflow,
and the Vortex Ring State, Journal of the American Helicopter Society,
Vol. 27, (3), July 1982, pp. 1824.
25
Prasad, J. V. R., Chen, C., Basset, P.-M., and Kolb, S., Simplified
Model and Nonlinear Analysis of a Helicopter Rotor in Vortex Ring State
for Flight Simulation, 30th European Rotorcraft Forum Proceedings,
Marseilles, France, September 1416, 2004.
26
Johnson, W., Helicopter Theory, Princeton University Press,
Princeton, NJ, 1980, pp. 98102.
27
Chen, C., Prasad, J. V. R., and Basset, P.-M., A Simplified Inflow
Model of a Helicopter Rotor in Vertical Descent, American Helicopter
Society 60th Annual Forum Proceedings, Baltimore, MD, June 710,
2004.
28
Chen, C., and Prasad, J. V. R., A Simplified Inflow Model of a
Helicopter Rotor in Forward Descent, AIAA-2005-622, 43rd AIAA
Aerospace Sciences Meeting and Exhibit, Reno, NV, January 1013,
2005.
29
Chen, C., and Prasad, J. V. R., Ring Vortex Model for Descending
flights, American Helicopter Society 61st Annual Forum Proceedings,
Grapevine, TX, June 13, 2005.
30
Basset, P.-M., and Omari, A., A Rotor Vortex Wake Model for
Helicopter Flight Mechanics and Its Application to the Prediction of the
Pitch-Up Phenomenon, Paper no. H08, 25th European Rotorcraft Forum
Proceedings, Rome, Italy, September 1416, 1999.
31
Taghizad, A., and L. Binet, L., Assessment of VRS Risk during
Steep Approaches, American Helicopter Society 62nd Annual Forum
Proceedings, Phoenix, AZ, May 911, 2006.
32
Benoit, B., Dequin, A.-M., Basset, P.-M., Gimonet, B., von
Grunhagen, W., and Kampa, K., HOST, a General Helicopter Simulation Tool for Germany and France, American Helicopter Society 56th
Annual Forum Proceedings, Virginia Beach, VA, May 24, 2000.
33
Kubicek, M., and Marek M., Computational Methods in Bifurcation
Theory and Dissipative Structures, Springer-Verlag, Berlin, 1983, pp.
3942.
34
Sibilski, K., Bifurcation Analysis of a Helicopter Nonlinear Dynamics, 24th European Rotorcraft Forum Proceedings, Marseilles, France,
September 1517, 1998.
35
Buler, W., Sibilski, K., Winczura, Z., and Zyluk, A., Nonlinear
Dynamics of Helicopter with Slung Load Analysis by Continuation,
AIAA-2001-110, 39th AIAA Aerospace Sciences Meeting and Exhibit,
Reno, NV, January 811, 2001.
36
Bedford, R. G., and Lowenberg, M. H., Use of Bifurcation Analysis
in the Design and Analysis of Helicopter Flight Control Systems, 29th
European Rotorcraft Forum Proceedings, Friedrichshafen, Germany,
September 1618, 2003.
37
Bedford, R. G., and Lowenberg, M. H., Bifurcation Analysis of
Rotorcraft Dynamics with an Underslung Load, AIAA-2004-4947,
AIAA Atmospheric Flight Mechanics Conference and Exhibit, Providence, RI, August 1619, 2004.
38
Guicheteau, P., Bifurcation Theory: A Tool for Nonlinear Flight Dynamics, Philosophical Transactions: Mathematical, Physical and Engineering Sciences, Vol. 356, No. 1745, Nonlinear Flight Dynamics of
High-Performance Aircraft, Oct. 15, 1998, pp. 21812201.
39
Guicheteau, P., Stability Analysis through Bifurcation Theory
and Non-Linear Flight Dynamics, Nonlinear Dynamics and Chaos,
AGARD Lecture Series 191, 1993, pp. 3.13.10, 4.14.11, 5.15.13.
40
He, C. J., Development and Application of a Generalized Dynamic
Wake Theory for Lifting Rotors, Ph.D. Dissertation, Georgia Institute
of Technology, School of Aerospace Engineering, Atlanta, GA, August
1989.

Vous aimerez peut-être aussi