See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/267717455

Dynamics of a Controlled Flexible Multibody Model of a 2 MW Wind Turbine

Conference Paper · May 2012

DOI: 10.13140/2.1.1061.7282

CITATIONS READS

6 2,429

3 authors, including:

Roman Rachholz Ch Woernle

University of Rostock University of Rostock
27 PUBLICATIONS   98 CITATIONS    110 PUBLICATIONS   377 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

DynAWind View project

MBS/BEM-Co-Simulation for Contact View project

All content following this page was uploaded by János Zierath on 04 November 2014.

The user has requested enhancement of the downloaded file.

 The 2nd Joint International Conference on Multibody System Dynamics
May 29–June 1, 2012, Stuttgart, Germany

Dynamics of a Controlled Flexible Multibody Model

of a 2 MW Wind Turbine

Roman Rachholz∗ , Christoph Woernle∗ , János Zierath#

∗ #
Department of Mechanical Engineering W2E Wind to Energy GmbH
and Marine Technology, 18055 Rostock, Germany
University of Rostock [email protected]
18051 Rostock, Germany
roman.rachholz, [email protected]

ABSTRACT

For a wind turbine prototype with a novel drivetrain concept multibody models are developed by means
of the program systems MSC.Adams and SIMPACK. Of special interest are simulations of failure cases
where the interaction between mechanical system and controller is important. For this purpose the multi-
body models are coupled with the original controller of the real wind turbine by means of an appropriate
interface. As an example a design load case based on the rules and guidelines of Germanischer Lloyd (GL)
for certification is simulated.

1 INTRODUCTION

The ongoing development of wind turbines aims both at higher efficient energy conversion and at reduced
purchase and operational costs. To reach these goals the size of wind turbines has been grown during the
last three decades. For example the Wind to Energy GmbH (W2E) developed in 2006 a 2.5 MW wind
turbine with a tower height of 160 m. Thus all components of a wind turbine as well as its overall dynamic
behavior have to be optimized closer to the limits than before [2].

Figure 1. a 2 MW wind turbine W2E-93/2.0 (hub height 98.2 m, rotor diameter 93.2 m), erected in
Tarnow, Mecklenburg Vorpommern, b Multibody model built up in SIMPACK
Flexible multibody models have been proven to be suitable as they describe well the interactions between
elastic body deformations of, e.g., blades and large rigid body motions. In addition extensive validations
are necessary in order to get an approval from classification societies [1, 8]. Further the dynamic behavior
of a wind turbine is influenced by the operational mode that is mainly governed by the control system. For
example the natural frequencies of wind turbines are influenced by the pitch angle of the blades and thus
by the pitch control. To accompany the ongoing development of a new generation of wind turbines with a
novel drivetrain concept, a simulation environment based a flexible multibody models is built up.

To describe the dynamic behavior of wind turbines specific simulation codes like FLEX5 were developed
about 20 years ago and continuously adapted to specific requirements in industrial wind turbine develop-
ment. These tools are computationally efficient, and the results obtained have been validated by measure-
ments and other simulations. However the topology of the implemented models is fixed and cannot be
extended within the given program structure. Therefore a strong demand for more general simulation tools
has been grown in the wind energy industry. Above all multi-purpose multibody simulation programs like
MSC.Adams or SIMPACK are suitable to meet the requirements. The depth of the mechanical model can
be successively enhanced according to the design process [4]. However it is necessary to provide submodels
for specific components of wind turbines like drivetrain elements, aerodynamic models to generate external
loads, the generator, or controllers.

In order to compare specific properties of the multi-purpose simulation programs MSC.Adams and SIM-
PACK models of the wind turbine W2E-93/2.0 shown in Fig. 1 have been implemented in both software
packages. Within these models the novel drivetrain concept Larus Compact according to Fig. 2 is integrated,
where the loads acting on the gearbox are reduced, and by this, the lifetime is extended. These improve-
ments are achieved by a fixed rotor bearing in one plane of support, an elastically compliant shaft bearing,
and a mounting of the gearbox at its center of gravity on the nacelle frame. Moreover an interface to inte-
grate the original controller is developed in order to analyse the dynamics of the controlled wind turbine.
The blades and the tower are defined as flexible bodies. Aerodynamic loads are applied by the simulation
code AeroDyn.

Figure 2. Drivetrain Concept Larus Compact, developed by W2E GmbH, Rostock

2 MULTIBODY MODELS OF A 2 MW WIND TURBINE IN MSC.Adams AND SIMPACK

For simulation of the previously described wind turbine the models in MSC.Adams and SIMPACK com-
prise a specific multibody topology, an interface for the original controller and the controller itself, flexible
components, and the applied aerodynamic loads. In the following subsections these model components are
described in more detail.

2.1 Multibody model of the overall wind turbine

The overall MSC.Adams multibody model of the wind turbine is shown in Fig. 3.

Figure 3. MSC.Adams model of the prototype W2E-93/2.0

The MSC.Adams and SIMPACK multibody models of the wind turbine are tree-structured according to
Figure 4 [13]. The main mechanical components are defined as rigid or flexible bodies that are connected by
joints with different degrees of freedom and force elements. As a wind turbine of a certain type can typically
be provided with different nacelles, blades, tower, or drivetrains in order to meet specific requirements of
the installation location dynamic simulations are to be performed for different configurations. Therefore
these main components are defined as substructures within the multibody models.

The tower assembly consists of a ground body with three rheonomically prescribed translational motions
to simulate optional earthquake excitations, the basement connected with a three-dimensional bushing to
the ground in order to simulate a viscoelastic ground, and the elastic tower body. The nacelle is mounted
on the tower by a rheonomically driven revolute joint for azimuth control. The main shaft supporting the
rotor is mounted on the nacelle in the rotor bearing that is modelled as a revolute joint. A revolute joint
with a torsional spring between rotor hub and main shaft describes the torsional stiffness of the rotor shaft.
The three rotor blades are connected to the hub by revolute joints for pitch control. Aerodynamic loads on
the blades are generated by the program AeroDyn and applied on the multibody system as applied forces.
The main shaft is elastically coupled with the gear ring at the input of the main gearbox by means of 21
bushing elements in order to reduce loads on the low-speed input shaft (LSS) of the gearbox due to shaft
misalignments caused by elastic deformations. Accordingly a 6 DOF joint with spatial stiffness between
main shaft and gear ring is provided in the model. The gearbox housing is mounted on the nacelle by
40 bushing elements allowing a spatial relative motion. Accordingly a viscoelastic 6 DOF joint is provided
between gearbox housing and nacelle.

The high-speed shaft (HSS) at the exit of the gearbox is connected with the brake by a revolute joint and
a torsional spring in order to describe the shaft torsional stiffness. To compensate shaft misalignments the
brake is connected with the rotor of the generator by bushing elements that are again modelled by a 6 DOF
joint with spatial stiffness. A revolute joint connects the rotor of the generator with the stator that is mounted
on the nacelle by four bushing elements described by a 6 DOF joint with spatial stiffness.
 Figure 4. Topology of the wind turbine models in MSC.Adams and SIMPACK

The controller acts on the multibody system by prescribing the generator torque and the three pitch angular
velocities, see subsection 2.3.

2.2 Multibody model of the gearbox

The gearbox consists of two planetary transmission stages and one spur wheel transmission. The SIMPACK
gearbox model is shown in Fig. 5.

The multibody topology of the gearbox model is shown in Fig. 6. The first planetary stage is built up of four
planetary gears, while the second planetary stage consists of three planetary gears. Both annulus gears are
rigidly fixed at the gearbox casing with a 0 DOF joint. The input torque from the rotor is transferred by the
low-speed shaft (LSS) to the carrier of the first planetary gear stage. Therefore the first carrier represents the
input of the first gear stage. The outputs of each planetary gear stage are the corresponding sun gears. At
the sun gear of the second stage a spur-gear stage is attached. The output gear of the spur-gear stage, which
is connected with the gearbox housing by a revolute joint, leads to the output of the gearbox and hence to
the high-speed shaft (HSS).

To consider torsional stiffness between the planetary stages and between the second planetary gear stage
and the spur gear stage, torsional springs are applied. The contact forces between the engaged gears are
described by wrenches acting separately on the corresponding gears shown in Fig. 6. Within the topology of
Fig. 6 the wrenches result from contact forces between a planetary gear and an annulus gear, forces between
a planetary gear and a sun gear and from the contact force in the spur-gear stage.

2.3 Controller Interface

The controller of the wind turbine comprises a speed control mode, a generator torque control mode, and
a power control mode for the transition between the two other modes. The speed control mode is applied
to run the wind turbine up to its rated speed. The torque control mode is used during full-load operation of
the turbine. Here the aerodynamic rotor torque is controlled in such a way that the rated electrical power
 Figure 5. SIMPACK Model of the gearbox

is generated at the rated speed of the generator. Under partial-load operational conditions the achievable
aerodynamic torque is not sufficient for the generation of the rated power. Then the power control mode is
applied, where at the rated angular speed the maximal achievable power is generated.

The controller is integrated into the operational control and safety concept as prescribed by classification
rules and guidelines. In a failure case an emergency controller mode comes into action. An example of an
emergency mode is considered in section 3.

The MSC.Adams and SIMPACK models run with the original controller of the wind turbine that includes
all nominal and emergency control modes. Controller interfaces are programmed for both software pack-
ages. As shown in Fig. 7 the controller inputs are the current generator speed, the three pitch angles, and
the actually generated power. Additionally time-independent informations like utility frequency and rated
power of the turbine are required. The controller returns the three pitch velocities that are applied to the
pitch actuators. Hence, the wind turbine is pitch-controlled. With the pitch angle the lift of the blades is
adjusted based on the current wind profile and, thus, by the actual main rotor torque.

Actually the pitch angles are rheonomically applied to the multibody system. The controller works with
discrete time steps of 0.01 s. At each time step the multibody model receives new values for the three pitch
velocities. This value is used to define the kinematic informations of the rheonomic revolute joints of the
blades. Using values of pitch velocities from previous time steps, values for the pitch angle and the pitch
acceleration are approximately calculated. As in the multibody simulation a variable-step time integration
is used, the pitch velocity of the last valid time step is kept constant until the controller delivers the value of
the next valid time step.

Considering for example the SIMPACK model, the interface for the generator torque is defined as a user-
defined force element defined by Fortran subroutines programmed in external datafiles. The external
datafiles are compiled and linked to SIMPACK. During simulation at each valid time step the required
informations of the model are gained from the user-defined force element and transferred to the controller.
The controller is a Dynamic Link Library (dll) developed in C-code. After evaluating the input variables
the controller calculates and returns the values of the three pitch velocities and the setpoint of the genera-
tor torque. The setpoint of the generator torque is applied on the multibody system by means of another
user-defined force element. To describe the dynamics of the generator torque controller a time constant for
the build-up of the actual generator torque can be provided that is not yet used, however. The interface for
the rheonomically driven pitch angles is defined as a user-defined kinematical element. Basically the same
implementation is used in the MSC.Adams model.
 Figure 6. Topology of the gearbox models in MSC.Adams and SIMPACK

2.4 Embedding flexible bodies

In both multibody models flexible bodies are embedded by means of a floating frame of reference (FFRF)
description (Fig. 8) [10]. Here the actual position of an arbitrary point P on a flexible body is described by
the position vector ri and the rotation matrix Ti of a floating frame of reference Ki of the body relative to
the inertia frame K0 and a usually small deformation uP ,

rP = ri + Ti ū with ū = u0 + uP . (1)

Generally the number of degrees of freedom (DOF) at a single elastic body based on a finite element
formulation needs to be reduced significantly before embedding the body into a multibody model. Modal

Figure 7. Interfaces of the multibody models in MSC.Adams and SIMPACK

 Figure 8. Kinematics of an elastic body in the FFRF

analysis techniques are used for this purpose. Under the assumption that the elastic body is discretized by
m nodes the small deformations uP ∈ R3m can be expressed by a weighted sum of time-independent mode
shapes,
uP = Φi qi , (2)
where the columns of the matrix Φi ∈ R3m×n represent the modes. The vector qi ∈ Rn comprises gener-
alised coordinates of the elastic motions with n DOF. To define appropriate shape matrices Φi component
mode synthesis (CMS) techniques are used [10, 11].

In MSC.Adams the Craig-Bampton method is applied . Using this technique a sophisticated combination
of eigenmodes calculated by a free vibration analysis and static constraint modes approximates possible
deformation states of the elastic body. The Craig-Bampton method differs between boundary DOF uB
and interior DOF uI . The static constraint modes are obtained by applying at each boundary DOF a unit
displacement while all other boundary DOF are fixed. When fixing the boundary DOF and carrying out
a free-vibration analysis the fixed-boundary normal modes are obtained. According to Eq. (2) the relation
between the physical DOF uP and the modal coordinates qi = [ qC qN ]T of the Craig-Bampton modes are
represented by     
uB I 0 qC
= . (3)
uI ΦIC ΦIN qN
The components of ΦIC are the physical displacements of the interior DOF in the constraint modes, and the
components of ΦIN are the physical displacements of the interior DOF in the fixed boundary normal modes.
The subscripts I, B, N, and C denote the internal DOF, boundary DOF, fixed boundary normal modes, and
constraint modes, respectively.

In SIMPACK a different approach for coordinate reduction in flexible bodies is used. Although it is possible
to use the Craig-Bampton method, alternatively a linear combination of eigenmodes and frequency response
modes (FRM) can be used [14]. Accordingly the elastic deformations uP are divided into deformations uN P
and uFRM
P expressed by the eigenmodes ΦNi and Φi
FRM
with the corresponding coordinates qiN and qiFRM ,
respectively,  
 N FRM
 qiN
uP = Φi Φi = Φi qi . (4)
qiFRM
The shape matrices ΦN i and Φi
FRM
are obtained from the undamped equations of motion of the finite
element body with a harmonic excitation.

2.5 Applied aerodynamic loads

For both simulation packages an interface is available for embedding the AeroDyn subroutine from the Na-
tional Wind Technology Center (NWTC), USA [5]. Using AeroDyn within MSC.Adams and SIMPACK
constant wind profiles can be defined and applied as aerodynamic loads on the multibody system. In com-
bination with TurbSim [7] it is possible to generate arbitrary turbulent wind profiles defined by a small
number of parameters. The generated wind field fulfills the requirement of the IEC 61400 standard.

To include AeroDyn into the multibody simulation of a wind turbine the available interfaces require spe-
cific data. For example a certain number of coordinate systems with predefined orientations and names
are required for the tower, nacelle, hub and blades in order to define the dimensions of the wind turbine.
Moreover an input, a wind, and a tower file are to be prepared as well as files defining the characteristics of
the aerodynamic wind profiles. More informations for the correct handling of the AeroDyn module and the
corresponding interface are available at [6, 3, 12].

3 CALCULATION OF A SPECIFIC DESIGN LOAD CASE (DLC)

Based on the rules and guidelines of Germanischer Lloyd (GL) [1] for classification of wind turbines a
simulation of the Design Load Case DLC 2.2 is considered as an example. Within this DLC a rare hazardous
situation during power production is anticipated, where a failure of the pitch controller causes a runaway of
the pitch angle during power production at nominal power and rated speed of the high speed shaft (HSS).
The goal of the simulations of this DLC is to test the functionality of the integrated controller in combination
with the mechanical dynamics of the wind turbine.

As a result of a MSC.Adams simulation Fig. 9 shows the speed of the HSS and the corresponding pitch
angle for the DLC 2.2. A standard wind profile with a constant wind speed of 14 m s−1 at hub height is
assumed. As initial state for the DLC the wind turbine operates with the rated speed 1150 rpm at the HSS
and a pitch angle of about 12 deg until about t = 30 s. The failure case is then initiated by an erroneous
runaway of the the pitch angle to 0 deg leading to an increased aerodynamic rotor torque and, by this, to
an increased angular speed the HSS. When the admissible maximum value of 1360 rpm is exceeded the
controller identifies a failure and changes into an emergency mode. In this mode the controller sets the
generator torque to zero and adjusts the pitch angles to 90 deg to make the lift vanish and, by this, to
decelerate the rotor.

Figure 9. Pitchangle and Rotational speed at HSS during a simulated pitch runaway

4 CONCLUSIONS

For more detailed modeling of the dynamics of wind turbines elastic multibody models of a prototype
turbine are developed by means of the multi-purpose simulation programs MSC.Adams and SIMPACK.
Aerodynamic wind loads are applied using AeroDyn. Above all, the models are used to simulate special load
and failure cases needed for certification of newly developed wind turbines. For this purpose the interaction
of the mechanical system and the actual controller is relevant. The multibody models are therefore coupled
with the original controller of the real wind turbine. In a next step the models are validated and further
refined by means of measurements at the prototype wind turbine. In addition the models will be used
for optimization of the drivetrain components and for the development of enhanced controllers using, for
example, individual control of the pitch angles.
 5 ACKNOWLEDGEMENT

The author would like to thank the Federal Ministry for the Environment, Nature Conservation and Nuclear
Safety for supporting this research work.

REFERENCES

[1] Germanischer Lloyd: Guideline for the Certification of Wind Turbines. Germanischer Lloyd, Ham-
burg: 2010.
[2] Hau, E.: Wind Turbines, Fundamentals, Technologies, Application, Economics. 2nd Ed., Springer
Verlag, Berlin Heidelberg: 2006.

[3] Hauptmann, S.: AeroDyn Interface for SIMPACK, User’s Guide. Universität Stuttgart, Institut für
Flugzeugbau: 2008.
[4] Hauptmann, S.; Mulski, S.; Kühn M.; Mauer, L.: Advanced drive train modeling in a virtual wind
turbine using the multibody simulation code simpack. Proceedings of the European wind energy
conference (EWEC), Milan: 2007.

[5] Laino, D.: NWTC Design Codes http://wind.nrel.gov/designcodes/simulators/aerodyn/: Last modi-

fied 21-February-2012, accessed 21-February-2012.
[6] Jonkman, J.; Laino, D.: NWTC Design Codes (ADAMS2AD).
http://wind.nrel.gov/designcodes/simulators/adams2ad/: Last modified 21-February-2012, accessed
22-February-2012.
[7] Kelly, N.; Jonkman, B.: MWTC Design Codes (TurbSim).
http://wind.nrel.gov/designcodes/preprocessors/turbsim/: Last modified 03-February-2011; ac-
cessed 03-February-2011.
[8] Kochmann M.; Ristow M.: Simulation of Drivetrains on Wind turbines within the Framework of
Certification - with SIMPACK. SIMPACK News,pp 20-21: 2010.
[9] Munteanu, I.: Optimal control of wind energy systems: towards a global approach. Springer Verlag,
London: 2008.
[10] Shabana, A.: Dynamics of Multibody Systems. 3rd Ed., Cambridge University Press, New York:
2011
[11] Schlecht, B.; Rosenlöcher, T.: Efficient Strategies To Determine Reliable Load Assumptions For
Wind Turbine Components. Proceedings of the European wind energy conference (EWEC), Warsaw:
2010.
[12] SIMPACK: How-to-SIM2AD. SIMPACK: 2010.

[13] Woernle, C.: Mehrkörpersysteme, Eine Einführung in die Kinematik und Dynamik von Systemen
starrer Körper. Springer Verlag, Berlin Heidelberg: 2011.
[14] Zierath, J.; Woernle, C.: Modeling Smooth Contacts in Elastic Multibody Systems. Multibody
Dynamics, ECCOMAS: 2009.

View publication stats