On The Modeling of Hydraulic Components in Rotorcraft Systems
On The Modeling of Hydraulic Components in Rotorcraft Systems
On The Modeling of Hydraulic Components in Rotorcraft Systems
Olivier A. Bauchau
Haiying Liu
School of Aerospace Engineering
Georgia Institute of Technology
Atlanta, GA
A physics based methodology for modeling of hydraulic devices within multibody-based comprehensive models of rotorcraft
systems is developed. The proposed models are developed in two stages. At first, models are developed for three basic
hydraulic elements: the hydraulic chamber, the hydraulic orifice and the pressure relief valve. These models consist of
nonlinear differential equations involving empirical parameters. Next, these basic elements are combined to yield device
models for linear hydraulic actuators, simple hydraulic dampers, and hydraulic dampers with pressure relief valves. The
proposed hydraulic device models are implemented in a multibody code and calibrated by comparing their predictions with
test bench measurements for the UH-60 helicopter leadlag damper. While predicted peak damping forces are found to
be in good agreement with measurements, the model does not predict the entire time history of damper force to the same
level of accuracy. The validated model of the UH-60 leadlag damper model is coupled with a comprehensive model of the
vehicle. Measured aerodynamic loads are applied to the blade and predicted damper forces are compared with experimental
measurements. A marked improvement in the prediction is observed when using the proposed model rather than a linear
approximation of the damper behavior.
Nomenclature
a, b
A
Aorf
Aprv1
A0 , A1
B
c
Cd
d
e
Fh
Fp
k
Km
m
p
p E0 , p E1
ps
Q orf
Q prvl
ti , t f
u
u k0 , u `0
uk , u`
u0
V
V0
,
1p
1t = t f ti
Subscript
()i
() f
()orf
()prvl
()
quantity at time ti
quantity at time t f
quantity associated with a hydraulic orifice
quantity associated with a pressure relief valve
derivative with respect to time
Introduction
175
176
O. A. BAUCHAU
prehensive rotorcraft model using a finite element based multibody formulation. The paper is organized in the following manner. The first section presents models for the basic hydraulic elements, and the second
section shows how these basic models can be combined to deal with
various hydraulic devices. Next, issues associated with the coupling of
hydraulic devices models with finite element based multibody formulations of rotorcraft dynamic simulation are discussed, with special focus
on the integration of the hydraulic equations. The modeling approach is
then validated using a number of numerical examples. Finally, conclusions of this work are offered.
Basic Hydraulic Elements
Hydraulic devices can be seen as an assembly of simple hydraulic
elements; in this work, three basic hydraulic elements will be presented:
hydraulic chambers, orifices, and pressure relief valves. These basic elements are described in the following sections. Of course, a variety of
other elements could be developed, such as hydraulic accumulators or
check valves.
Hydraulic chamber
The hydraulic chamber, shown in Fig. 1, is probably the most common
hydraulic component. The chamber, of volume V and cross-sectional area
A, is filled with a hydraulic fluid of bulk modulus B under pressure p.
Often, due to the presence of a piston, the length of the chamber can vary.
The change in length of the chamber, due to piston motion, is denoted
d. Finally, hydraulic fluid can flow into the chamber; Q denotes the net
volumetric flow rate into the chamber.
The evolution of the pressure in the chamber is governed by the following first order differential equation
p =
(Q A d)
V
(1)
(2)
The bulk modulus of the hydraulic fluid is a function of the fluid pressure;
an accurate approximation of this dependency is written as
B=
1 + p + p 2
+ 2p
(3)
where and are physical constants for the hydraulic fluid; see Ref. 1.
Hydraulic orifice
The hydraulic orifice, shown in Fig. 2, allows the flow of hydraulic
fluid through an orifice of sectional area Aorf . The orifice is connected
APRIL 2006
177
Hydraulic Devices
The hydraulic elements described in the previous section can be combined to form practical hydraulic devices such as linear hydraulic actuators, simple dampers, or dampers with pressure relief valves, which
are described below. More complex devices could be modeled using the
same technique.
(8)
The governing equations for the linear hydraulic actuator include equations for the pressures p0 and p1 in the two chambers, Eq. (1), and
equations for the flow rates Q 0 and Q 1 through the orifices, Eq. (4).
Most hydraulic actuators are also equipped with check valves that
connect the hydraulic chambers to the circuit background pressure when
the chamber pressure falls below the background pressure, in an effort to
avoid cavitation in the chamber.
(5)
When the net force acting on the poppet is smaller than the pre-load force,
i.e., when p0 A0 p1 A1 < Fp , the valve remains closed, x = x = 0. On the
other hand, when the net force is large enough to overcome the pre-load
force, the poppet opens and its motion is governed by Eq. (5). Once the
valve is open, fluid will flow through the valve at the following volumetric
rate
s
2|1p| 1p
Q prvl = Aprv1 Cd
(6)
|1p|
178
O. A. BAUCHAU
hydraulic dampers are also equipped with check valves, as discussed for
actuators. The hydraulic chambers are connected to a plenum with oil at
the circuit background pressure to prevent pressure drops in the chambers
and subsequent cavitation.
The governing equations for the hydraulic damper with pressure relief
valves include equations for the pressures p0 and p1 in the two chambers,
Eq. (1), one equation for the flow rate Q through the orifice, Eq. (4), two
equations of motion for the valve poppets, Eq. (5), and two equations for
the flow rates through the valves, Eq. (6). The flow area of the valves is
computed with the help of Eq. (7).
Finite Element Implementation
The various hydraulic devices described in the previous section interact with the dynamics of the mechanical system they are connected to.
For instance, a helicopter leadlag damper interacts with rotor blade dynamics; this effect is particularly pronounced on the fundamental blade
leadlag mode. This section describes the coupling of the hydraulic device model with a structural dynamics model, within the framework of
multibody system dynamics; see Ref. 5. The following sections describe
the coupling procedure in terms of the applied structural forces, their
time discretization, and the time integration scheme for the equations
governing the behavior of hydraulic devices.
Applied structural forces
In general, hydraulic devices generate hydraulic forces given by
Eq. (8) that are functions of the stroke d, through Eq. (2) and stoke
through Eq. (1). This stroking can be evaluated from the conrate d,
figuration of the device depicted in Fig. 7. In the initial configuration,
the end points of the device are at location u k0 and u `0 , respectively, with
respect to an inertial frame I = (i1 , i2 , i3 ). At those points, the device is
connected to a dynamical system of arbitrary topology. In the deformed
configuration, the displacements of the end points of the device are u k and
u ` , respectively. The relative position of the end points will be denoted
u 0 = u `0 u k0 and u = u ` u k in the initial and present configurations,
respectively. Note that the rotational degrees of freedom of the structure
at the connection points are not involved in this formulation, implying
the presence of spherical joints at these points.
The virtual work done by the hydraulic force is W = F h d, where
d = (kuk ku 0 k) is a virtual change in device length. This expression
then becomes
W = F h
u T u
= F h e T (u ` u k )
kuk
e
F = Fh
e
(9)
(10)
APRIL 2006
which are two forces of equal magnitude and opposite sign applied at the
connection points.
Time discretization of the structural forces
In a typical finite element implementation, the simulation of the system dynamics is discretized in time. The force generated by the hydraulic device will be assumed to remain a constant, Fmh , over the time
step and the work done by this force over the time step now becomes
1W = Fmh (d f di ). This expression is manipulated to become
1W =
Fmh 2
Fh
d f di2 = m u Tf u f u iT u i
2dm
2dm
(11)
u mT
(u f u i ) = Fmh emT (u f u i )
dm
(12)
T
U em e Tf
1 Fmh U em e f
Km =
2 dm U e e T
U em e Tf
m f
"
#
T
em e Tf
dFmh em e f
+
ddm em e Tf
em e Tf
(15)
and U is the 3 by 3 identity matrix. This expression requires the evaluation of the derivative of the hydraulic force with respect to the stroke,
dFmh /ddm , which could be computed from the governing equations for
the hydraulic device. However, this process is, in general, quite involved.
The following approximation was found to be suitable
dFmh
B A20
B A21
=
+
ddm
V0
V1
(16)
179
high stiffness of the hydraulic fluid (for typical systems, the bulk modulus of the fluid is about 1.5 GPa). Typically, this problem is overcome
by using a very small time step for the integration of the hydraulic equations; for instance, Welsh (Ref. 3) used a time step of 1t = 106 sec to
integrate the equations of a helicopter airoil strut. While this approach is
acceptable when dealing with the sole hydraulic equations, it is not practical to integrate both hydraulic and structural dynamics equations with
such a small time step because the computational effort would become
overwhelming. Consequently, it is imperative to decouple the integration
of the two systems: the structural dynamics equations are integrated with
a time step dictated by the frequency content of the structural response,
whereas the hydraulic equations are integrated with a much smaller time
step.
In this work, the following strategy was used: the structural dynamics equations are integrated with a time step 1t; energy decaying
schemes that guarantee nonlinear unconditional stability of the time integration process are used for this purpose (Refs. 711). This produces
a prediction of the stroking of the hydraulic device, di = ku i k d0 and
d f = ku f k d0 , the stroking rate has a constant value dm = (d f di )/1t.
This information was used to integrate the governing equations of the hydraulic device using a fourth order RungeKutta integrator (see Ref. 12).
The time step used in this integrator was h = 1t/N , i.e., N RungeKutta
steps are performed for each structural time step. Once the hydraulic
equations are solved, the pressures in the chambers are predicted and
hence the hydraulic device force. The nonlinear solution of the problem
is then obtained by iterating between the structural dynamics equations
and the hydraulic equations.
Numerical Examples
Hydraulic linear actuator
The first example deals with a hydraulic linear actuator that is used
to pitch a beam, as depicted in Fig. 8. The system consists of a flexible
beam of length L = 0.8 m with a 10 kg tip mass connected to a revolute
joint at point R. At point C, located at a distance d = 0.24 m from the root
of the beam, a flexible horn connects to the beam. Finally, a hydraulic
linear actuator is connected between the ground and the tip of the horn
at points S and D, respectively.
The physical properties of the beam and horn are as follows: axial stiffness, 5.7 107 N, bending stiffness, 4.275 103 Nm2 , shearing
stiffness, 1.80 107 N, mass per unit span, 2.4 kg/m. The configuration
of the hydraulic linear actuator is that depicted in Fig. 4. The physical
properties of the actuator are listed in Table 1.
The system was initially at rest. To simulate the actuators control
valves, the throttling areas of both orifices were linearly ramped up from
zero to their nominal value in 0.5 s. In the next 0.5 s, the throttling areas
were linearly ramped back down to zero. The time step for the structural
analysis was set to 1t = 1.0 104 s; for each structural step, 48 substeps were used for the integration of the hydraulic equations. These
180
O. A. BAUCHAU
A0 = A1
V0 = V1
pE0 = pE1
Aorf0 = Aorf1
Cd0 = Cd1
ps
B
Value
7.85 10 5 m2
9.42 10 6 m3
1.25 MPa
6.0 10 6 m2
0.611
1.0 MPa
1.53 MPa
Fig. 12. Time histories of the volumetric flow rates into the chambers.
Once the throttling areas vanish, the length of the actuator remains nearly
constant; the observed oscillations are due to vibrations of the beam-tip
mass system following actuation. The interaction between the hydraulic
device and the structure is further demonstrated in Fig. 11, which shows
the time histories of the pressures in the two chambers. Note the sudden drop in chamber 0 pressure at time t = 1 s due to the opening of
the check valve; the effects of this pressure spike are noticeable on the
device velocity and output forcesee Figs. 10 and 9, respectively. After
the closing of the throttling areas, large variations in chamber pressures
are observed resulting from structural vibrations. Finally, Fig. 12 depicts
the time histories of the volumetric flow rates into chamber 0 and chamber 1. The flow rates that are observed after the closing of the throttling
areas are flow rates through the actuator check valves. The very rapid
variations in chamber pressure and orifice flow rates are further evidence
of the very high stiffness of the system, and help explain the need for the
numerous sub time steps required to integrate the hydraulic equations.
Validation of the model of the UH-60 leadlag damper
Fig. 10. Time histories of the displacement (top figure) and velocity
(bottom figure) of the piston of the hydraulic actuator.
APRIL 2006
181
cient, have been set to their theoretical values. Other coefficients, such
as those appearing in Eq. (7), were selected based on indirect, uncertain measurements. Parametric studies performed with the present model
have demonstrated the great sensitivity of the predictions to the choice
of these coefficients.
Modeling the UH-60 blade and leadlag damper
Fig. 14. Time history of the leadlag damper force at a peak velocity
of 2 in/s.
The damper was tested under harmonic stroking conditions at a constant
circular frequency = 27.02 rad/s. Tests were run at various amplitudes
of the harmonic motion; Fig. 13 shows the experimentally measured peak
force in the damper as a function of peak velocity. This figure also shows
the predictions of the present model; good agreement is found between
measurements and predictions. The time history of the damper force at a
peak velocity of 2 in/s is shown in Fig. 14 for both model and experiment.
While peak loads are in good agreement, force time histories exhibit qualitative differences. First, the experimentally measured force dwells for a
short period when it reaches a value near zero; this phenomenon is not
predicted by the model and its physical origin is not known; possible explanations are discussed in the next paragraph. Second, the experimental
measurements exhibit a different behavior at peak positive and negative
forces. This dissymmetry is not present in the model and its physical
origin is also unclear.
The following conclusions can be drawn from this calibration effort.
The proposed model seems to predict damper peak loads with reasonable
accuracy, while the details of the force time history are not predicted to
the same level of accuracy. The probable cause of these discrepancies is
the highly idealized nature of the present model. Several components of
the device, such as the hydraulic accumulators and check valves, have
intrinsic characteristics that have not been modeled in the present effort.
Several coefficients of the model, such as the orifice discharge coeffi-
182
O. A. BAUCHAU
Fig. 17. Time history of the leadlag damper stroking (top figure)
and stroking velocity (bottom figure).
with a forward speed of 158 kts. A total of 75 rotor revolutions were simulated to allow all transients to die out and so to obtain a periodic solution.
The results presented in the figures below depict the last revolution of the
simulation, as a function of the azimuthal angle 9. A constant time step
size of 256 steps per revolution was used for the structural equations and
25 sub-steps were used for the integration of the hydraulic equations.
Figure 16 displays the time history of the predicted damper force. Note
the effectiveness of the pressure relief valves that limit the maximum
damping force to about 3200 lbs. The stroke and stroke velocity of
the damper are presented in Fig. 17. These quantities are the variables
forming the basis for the empirical models of the dampers: the output
force is assumed to be a nonlinear function of the instantaneous velocity.
In the present model, additional information is available that describes
the internal behavior of the device. The pressures in the two chambers of
the dampers are shown in Fig. 18. Next, the volumetric flow rate through
the orifice is shown in Fig. 19, and this flow rate tends to equilibrate the
pressures in the two hydraulic chambers. As expected, the shape of this
time history closely follows that of the damper force.
The role of the pressure relief valves is illustrated in Figs. 20 and
21, which depict the displacement and velocities of the valves, and the
flow rate through these valves, respectively. Valve 0 opens over the azimuthal range 9 [45, 100] then [125, 215] deg, whereas valve 1 opens
for 9 [240, 290] deg. These ranges are clearly correlated with the high
Fig. 19. Time history of the predicted volumetric flow rate through
the orifice.
Fig. 20. Time history of the predicted displacement (top figure) and
velocity (bottom figure) of pressure relief valves.
APRIL 2006
Fig. 21. Time history of the predicted volumetric flow rates through
the pressure relief valves.
183
1) A methodology allowing physics based modeling of hydraulic devices within multibody-based comprehensive models of rotorcraft systems was developed.
2) The new mathematical models of hydraulic devices were implemented in a multibody code and calibrated by comparing their predictions
with bench test measurements. While predicted peak damping forces
were found to be in good agreement with measurements, the model did
not predict the entire time history of damper force to the same level of
accuracy.
3) The validated model of the UH-60 leadlag damper model was coupled with a comprehensive model of the rotor system. Measured aerodynamic loads were applied to the blade and predicted damper forces were
compared with experimental measurements. A marked improvement in
the prediction was observed when using the proposed model rather than
a linear approximation of the damper behavior.
4) The proposed model also evaluates relevant hydraulic quantities
such as chamber pressures, orifice flow rates, and pressure relief valve
displacements. Hence, the present model could be used to design leadlag
dampers presenting desirable force and damping characteristics.
Acknowledgments
This work was sponsored by the National Rotorcraft Technology Center and the Rotorcraft Industry Technology Association under contract
WBS No. 2003-B-01-01.1-A1. Yung Yu was the contract monitor.
References
1
Viersma, T., Analysis, Synthesis and Design of Hydraulic ServoSystems and Pipelines, Elsevier, Amsterdam, 1980, Chaps. 1 and 2.
2
Canon, R., Dynamics of Physical Systems, McGraw-Hill, New York,
1967, Chap. 4.
3
Welsh, W., Simulation and Correlation of a Helicopter Air-Oil Strut
Dynamic Response, American Helicopter Society 43rd Annual Forum
Proceedings, St. Louis, MO, May 1820, 1987.
4
Welsh, W., Dynamic Modeling of a Helicopter Lubrication System, American Helicopter Society 44th Annual Forum Proceedings,
Washington, DC, June 1618, 1988.
184
O. A. BAUCHAU
5
Bauchau, O., Bottasso, C., and Nikishkov, Y., Modeling Rotorcraft
Dynamics with Finite Element Multibody Procedures, Mathematical
and Computer Modeling, Vol. 33, (1011), 2001, pp. 11131137.
6
Cardona, A., and Geradin, M., Modeling of a Hydraulic Actuator
in Flexible Machine Dynamics Simulation, Mechanisms and Machine
Theory, Vol. 25, (2), 1990, pp. 193207.
7
Bauchau, O., and Theron, N., Energy Decaying Scheme for NonLinear Beam Models, Computer Methods in Applied Mechanics and
Engineering, Vol. 134, (12), 1996, pp. 3756.
8
Bauchau, O., and Theron, N., Energy Decaying Schemes for Nonlinear Elastic Multi-body Systems, Computers and Structures, Vol. 59,
(2), 1996, pp. 317331.
9
Bauchau, O., Computational Schemes for Flexible, Nonlinear
Multi-Body Systems, Multibody System Dynamics, Vol. 2, (2), 1998,
pp. 169225.
10
Bauchau, O., and Bottasso, C., On the Design of Energy Preserving
and Decaying Schemes for Flexible, Nonlinear Multi-Body Systems,
Computer Methods in Applied Mechanics and Engineering, Vol. 169,