simpack ag, friedrichshafener strasse 1, 82205 gilching ...€¦ · simpack ag, friedrichshafener...

SIMPACK NewsSIMPACK NewsSIMPACK AG, Friedrichshafener Strasse 1, 82205 Gilching, Germany

27 SOftwareFatigue Analyses with FAT4FEM in SIMPACK's PostProcessor

25 SOftwareGear Pair Enhancements with SIMPACK Version 8904

16 CuStOmer appliCatiOnDetermining Reliable Load Assumptions in Wind Turbines using SIMPACK

30 SOftware200 New Features and Improvements introduced with SIMPACK Version 8904

06 CuStOmer appliCatiOnElectronic Stability Program (ESP) for Trucks on the Daimler Driving Simulator

02 CuStOmer appliCatiOnSimulation of the Dynamic Behavior of Aircraft Landing Gear Systems

10 CuStOmer appliCatiOnSimulating Tank Vehicles with Sloshing Liquid Load

13 CuStOmer appliCatiOnDevelopment of a SIMPACK User Routine for Dynamic Light Rail Vehicle Gauging Simulations

22 CuStOmer appliCatiOnCoupling of MBS and CFD: an Oscillating Aeroelastic Wing Model

21 newSSIMPACK Academy Events in 2010

20 CuStOmer appliCatiOn Simulation of Drivetrains on Wind Turbines within the Framework of Certification — with SIMPACK

Simulation of the Dynamic Behavior of Aircraft Landing Gear Systems

The intensified efforts to provide alterna-tive and renewable sources of energy led to a substantial worldwide growth in

the wind turbine industry over the last few years. The advantages of this clean energy source and the recent successes in increas-ing power output rates for on- and offshore wind turbines is extremely... See page 16

Electronic Stability Program (ESP) for Trucks on the Daimler Driving Simulator

Determining Reliable Load Assumptions in Wind Turbines using SIMPACK

The Electronic Sta-bility Program (ESP) safety system is a common feature in passenger cars nowadays and has proven to be an es-

sential component in the reduction of traffic accidents. Electronic "stability control" and "rollover control" have been available since 2001 for commercial... See page 6

SEPTEMBER 2010

Simulation and the resulting prediction of the dynamic behavior of an aircraft and its landing gear system during ground maneuvers is an essential part in the design process. A realistic estimation of unwanted oscillations, such as gear walk and shimmy, for the landing gear and the whole aircraft can be readily obtained with an MBS-model. It is then possible to adjust the model for changes in the structural design of the airframe and the landing gear so as to optimize the... See page 2

forward forward

roll- oscillation

longitudinal-oscillation gear walk

lateral-oscillation yaw-oscillation

shimmy

2 | SIMPACK News | September 2010

CuSToMER APPLICATIon | Reinhard Lernbeiss, TU Wien, Institute of Mechanics and Mechatronics

Simulation of the Dynamic Behavior of Aircraft Landing Gear Systems

Simulation and the resulting prediction of the dynamic behavior of an aircraft and its landing gear system during ground maneuvers is an essential part in the design process. A realistic estimation of unwanted oscillations, such as gear walk and shimmy, for the landing gear and whole aircraft can be readily obtained with an MBS-model. It is then possible to adjust the model for changes in the structural design of the airframe and the landing gear so as to optimize the aircraft stability.

InTRoDuCTIonVibrations resulting from the elastic behavior or from dynamic loads may result in material fatigue and failure. Loads acting on the land-ing gear at touch-down are of major inter-est. Addressing these issues during testing is crucial. Elastic properties of the airframe and the landing gear have an essential influence on their dynamic behavior. Emphasis was placed on developing an “elastic” model of the whole aircraft so as to realistically simu-late a complete landing and rollout with its braking, and in particular, the influence of the landing gear. To demonstrate the capabilities of this approach an existing aircraft (Airbus A320-200) was selected. To control the aircraft during flare, touch-down and roll-out, a control system was introduced, capable of achieving any desired angle of the aircraft in relation to the runway at the exact moment of touch-down. This enabled the simulation of possible crosswind condi-tions as well as different landing techniques applied by the pilot. It was possible to

Fig. 1: Dynamic phenomena

pre-select certain values of residual vertical speeds upon touch-down. Additionally, the application of the brakes was performed by an automatic braking system combined with an anti-skid system.

DYnAMIC PHEnoMEnAUndesired oscillations can occur in the longitudinal, lateral and yaw directions. Longitudinal vibrations are normally induced by changes of vertical and longitudinal loads acting on the wheels. They can be generated

by landing impact or during braking and are

commonly called gear-walk. The lateral and yaw oscillations are called shimmy oscilla-tions when generated by self excitation forces. However, such vibrations may also be in-duced by asymmetric conditions occurring at landings with prevailing crosswind. Even

the asymmetric struc-ture of the landing gear itself, as occurs on most main landing gear

systems, may be a source of unstable con-ditions, see Fig. 1. Rolling motions of the wheels about their longitudinal axis also exist.

SIMuLATIon MoDELSTo generate a good approximation of the mass distribution of an aircraft like the Airbus A320 and the structural properties, existing and accessible data were used together with the statistical mass approximation method published by Raymer. These data were the main sources used to establish a CAD-model of all of the structures. To comply with the goal of simulating different loading cases of passengers, cargo and fuel, a balance calculation similar to those applied before each flight of the real aircraft was used. The data obtained for the air-frame structure was pre-processed to facilitate the generation of an elastic model in SIMBEAM. The elastic structure was assembled with the use of beam elements accounting for

“Emphasis was placed on developing an “elastic” model

of the whole aircraft...”

SIMPACK News | September 2010 | 3

Reinhard Lernbeiss, TU Wien, Institute of Mechanics and Mechatronics | CuSToMER APPLICATIon

Fig. 3: Arrangement of flexible beams of fuselage and wings

Fig. 2: Airbus A320; structure of aircraft

the main parts of the fuselage and the whole wings. Other elements, for example the foremost part of the fuselage in front of the nose landing gear, as well as the empennage, are considered rigid, but have the correct mass-properties and were attached to the corresponding elastic beam element. Implementing beam elements to generate elastic structures reduced the required time for simulation. Different beam elements were defined between certain markers and reflected a change in structural properties. In addition, those markers were set at positions to easily accommodate major mass concentrations and structural attachments, especially for the wings and the landing gear, see Fig. 3. To compensate for possible elements which have no structural influence, for example aircraft systems and fairings, the masses were equally distributed along the elastic beams between certain positions, and the mass was adjusted accordingly.

data aquisition SIMPACK output

data aquisition SIMPACK

intput

output

reverse thrust

autobrake selection

yaw & roll pre-selection

aerodynamic wings

flare trajectory

aerodynamic hor. stabilizer

engine thrust

brakes & directional

control

SIMPACK model

1st Eigenmode 2nd Eigenmode

4th Eigenmode


CuSToMER APPLICATIon | Reinhard Lernbeis, TU Wien, Institute of Mechanics and Mechatronics

All other elements were added as rigid mass elements. This facilitated changing the loading of the aircraft and the amount of fuel for simulation of different landing mass and centers of gravity. Some of the resulting eigenmodes of the elastic aircraft model are shown in Fig. 4.

LAnDInG GEARA similar procedure was applied to establish elastic models of the landing gear system. The elastic structure of the landing gear model was comprised of a system of beam elements. The method of using distributed masses and mass concentrations, where needed, was also applied. To generate appropriate forces acting on the wheels and tires, a modified HSRI-tyre model was used. An automatic braking system in conjunction with an anti-skid system was used to achieve braking action during the rollout phase. The functions of these systems were basically reproduced by models programmed with MATLAB® and Simulink® using co-simulation.

ConTRoLLInG oF THE LAnDInG MAnEuVERTo accomplish a realistic landing-flare, it is imperative to incorporate aerodynamic forces. However, one of the goals is to keep total simulation time low. Therefore, the aerodynamic model was kept as simple as possible and only those aerodynamic loads were applied which are essential to generate a nearly realistic landing maneuver and which have possible influence on the elastic structure with respect to the dynamic behavior of the landing gear. On the wings Fig. 5 and 6: Landing simulation

Fig. 4: Eigenmodes

lift and drag forces were calculated using the basic aerodynamic equations acting on the corresponding surfaces belonging to each marker. The angle of attack is measured for each section of the wing considered, taking flexibility of the wings into account. With this aerodynamic model, neither static nor dynamic stability of the aircraft model is possible. This is accomplished by using artificial stability generated by a flight control model which simply generates forces on the horizontal stabilizer and elevator to produce flight stability and control. In the case of a real aircraft, the vertical speed upon touch-down is minimized. For the simulation, however, pre-determined vertical speeds are required for the intended parameter variations. Therefore, a schedule for the target vertical speed, dependent


Reinhard Lernbeiss, TU Wien, Institute of Mechanics and Mechatronics | CuSToMER APPLICATIon

Fig 7: Nose landing gear; Airbus A320

upon height, was introduced. To simulate crosswind conditions upon landing and corresponding landing techniques used by the pilot or the automatic landing system, certain angles of roll and bank were selected. These angles were kept constant during flare until touch-down of the main wheels by a separate controlling system.

SIMuLATIon SET-uPThe model of the elas-tic aircraft structure together with the land-ing gear system in the MBS-software SIMPACK was simulated with the controlling system of the aircraft, programmed in Simulink using a co-simu-lation. Simulink was used for aerodynamic forces, anti-skid, autobrake sys-tem and steering on the runway during roll out. Thrust control during flare and reverse thrust were also provided, see Fig. 5 and 6.

DRoP TEST SIMuLATIonDuring the design and development process, a so-called drop test con-ducted in a laboratory is used. Here a single real landing gear unit, loaded with the appropriate mass, falls onto a rotat-ing drum which repre-sents the moving runway surface. This test is a vital source of information and the data generated are used to optimize both

the design of the landing gear and for the flight test later in the development process. Therefore, it is imperative to conduct suf-ficient simulations of that test in advance to save resources and time. In addition, the simulation of drop tests of the landing gear with flexible structure models enables a touch-down with different side slips (yaw angles) and the motion of the de-rotation of the aircraft, which is the lowering of the nose after touch-down of the main landing gear.

SIMuLATIon oF LAnDInG An AIRCRAFTIt is crucial to have sufficient model detail to gain insight into the dynamic behavior of landing gear. Fig. 8 presents a comparative study of the relative displacement and twisting of the wheel axis of the main landing gear using different modeling techniques of the elastic properties of the aircraft structure. It can be seen that the results may differ quite significantly.

ConCLuSIonTo facilitate the design process, it is advan-tageous to implement and use simplified models to simulate a number of operational aspects to prevent undesired and costly but necessary improvements during flight tests. It is of utmost importance to have an easy to use and changeable model at any stage of development to predict the behavior of the landing gear. It is then possible to modify the design or to make appropriate adjustments to a shimmy damper or similar device at an early stage of design. Implementing a flexible structure in the simulation model is essential. In addition, it is possible to produce a realistic landing maneuver using only limited applica-tion of aerodynamic calculations and save

computational time to fa-cilitate the design process by testing various structural configurations early in the design process. The model presented enables the use of real-time simulations

used in flight simulators, and is necessary for aircraft that use an extensive amount of flex-ible composite components.

“It is crucial to have sufficient model detail to

gain insight into the dynamic behavior of

the landing gear.”

Fig 8: Relative displacement of the wheel axis; different levels in modeling the aircraft structure

oversteer understeer


CuSToMER APPLICATIon | Klaus Wüst, Daimler AG, CAE Commercial Vehicles

Electronic Stability Program (ESP) for Trucks on the Daimler Driving Simulator

ESP FoR CoMMERCIAL VEHICLESThe Electronic Stability Program

(ESP) safety system is a common feature in passenger cars nowadays and has proven to be an essential component in the reduction of traffic accidents. Electronic "stability control" and "rollover control" have been available since 2001 for commercial vehicles. The system used for trucks and buses has two basic functionalities:The "stability control" functionality works similar to passenger car ESP systems. Sensors measure the steering wheel angle, the yaw velocity and the lateral acceleration of the vehicle and feed the values into a simple vehicle model. As soon as this model detects a big

difference between the desired and measured vehicle path, either in an oversteering or an understeering direction, individual wheels are braked to generate a moment around the vehicle’s vertical axis, which stabilizes the vehicle movement (Fig. 2). For commercial vehicles, especially when loaded, these limits of adhesion are usually only reached on roads with a low coefficient of friction. For trucks and buses travelling on dry road surfaces, it is more important to prevent them from reaching the rollover limit (Fig. 3).The “rollover control” functionality limits the lateral acceleration of the vehicle to a pre-defined value which is dependent on the vehicle mass. When the vehicle reaches this maximum value, all the wheels of the towing vehicle and the trailer are braked to reduce speed.For Daimler trucks, the system is integrated into the electronic braking system and is sold as “Telligent Stability System” in a safety package together with other systems like lane keeping assistant and active cruise control. However, until now the system was only Fig. 2: ESP stability control

Fig. 3: Truck rollover

Fig. 1: Mercedes-Benz Actros “Safety Truck”


Klaus Wüst, Daimler AG, CAE Commercial Vehicles | CuSToMER APPLICATIon

Fig. 5: Simulation-assisted ESP development

available for tractor/semitrailer combinations, and the percentage of vehicles equipped with the system is still low.This is why the legislative committee of the European community has decided to make ESP systems mandatory for trucks and buses. The newly defined legislation ECE-R13 will go into effect this year starting with mandatory ESP systems for coaches and for tractor/semitrailer combinations, where the system is already available. All other trucks and buses (apart from vehicles used for construction purposes), will have to follow by 2014.

Fig. 5: ESP field testing in reality and the respective SIMPACK model simulation

This makes it necessary for truck and bus manufacturers and braking system suppliers to develop the system for a wide variety of commercial vehicles. As the number of vehicles which can be used for proving the system functionalities in field testing is limited by material expense and time, it is necessary to use vehicle dynamics simulation to support the development of the system. A myriad of commercial vehicle parameters, such as different axle configurations, different wheelbases, tire variations, varied loading conditions, etc. can be incorporated into the simulation.

ESP S-functionSIMPACK Code Export resultsmaneuver


CuSToMER APPLICATIon | Klaus Wüst, Daimler AG, CAE Commercial Vehicles

Fig. 6: SIMPACK vehicle model

ESP SoFTWARE-In-THE-LooP CoDES FoR SIMPACK VEHICLE MoDELSTo study vehicle behavior with ESP in simula-tion, a code of the system has to be inte-grated into the vehicle dynamics simulation tool. The CAE analysis division of Daimler Trucks uses SIMPACK as a standard multi-body simulation tool for different purposes. SIMPACK offers various possibilities for the integration of system codes into the simula-tion, e.g. the MatSIM interface for the inte-gration of Simulink® codes.The SIMPACK model consists of detailed component based models of axles and cab

mounting. A detailed model was also used for the steering system. The frame includes torsional bending behavior. For the integration of truck ESP into SIMPACK, Daimler and the system supplier WABCO have chosen MATLAB® and Simulink as an exchange and integration platform. The Daimler CAE analysis division uses the SIMPACK code export feature to generate a Simulink S-Function, which is combined with an S-Function of the ESP code, delivered by WABCO. The driving maneuvers and the evaluation of the results are generated within MATLAB and Simulink.

TRAnSFER oF SIMPACK MoDELS To THE DAIMLER DRIVInG SIMuLAToRThe Daimler driving simulator in Berlin has been in existence since 1995 and will be moved to Sindelfingen in 2010. A complete car or a truck cabin is installed on a hexapod which can additionally be moved in the horizontal direction. The movements of

the simulator are generated by a vehicle simulation model. This simulator serves for investigations of the interactions between the driver and the vehicle. On one hand, it is used for subjective evaluations of driver assistance systems. For example, the “brake assist” that guarantees full braking application during emergency situations was

developed based on simulator investigations. On the other hand, the simulator is also used for the subjective evaluation of parameter changes within the chassis layout during the development phase of new vehicles.The Daimler Trucks CAE analysis division has used the driving simulator since 2006, when the real-time capabilities of SIMPACK

Fig. 7: Simulink simulation with exported SIMPACK model and ESP code

Fig. 8: Daimler driving simulator


Klaus Wüst, Daimler AG, CAE Commercial Vehicles | CuSToMER APPLICATIon

allowed the transfer of simulation models used for the chassis layout directly to the driving simulator (see SIMPACK News 2/08). Since then, several simulated driving tests have been conducted with SIMPACK models. The results were used to define target values for the vehicle dynamics of new truck or van generations and for the subjective evaluation of many chassis variants before field testing.

ESP InVESTIGATIonS on THE DRIVInG SIMuLAToRA first test with ESP for trucks on the driving simulator was conducted in November 2009. The test served as a basis for investigating if the simulator could give additional value to the simulation-supported development of ESP for trucks. The basic questions were if the simulator would be able to realistically reproduce the interventions of an ESP system for trucks, and if the simulator could be used for optimizing the system, e.g. for defining the intervention threshold values for different vehicles and various vehicle parameters.As the Windows-based MATLAB and Simulink Software-in-the-Loop (SiL) environment was not suited for the UNIX operating system of the driving simulator, a decision was made to use a different approach for this first test. A Fortran code, which simulates the basic ESP functionalities of stability control and rollover protection, was integrated by SIMPACK into a real-time vehicle model as a user routine. The model was then exported into the simulator environment together with the user routine,

which then could be parameterized after the code export via the SIMPACK “subvar” file. This has the advantage of a flexible parameterization during the simulator tests. In addition to the parameters of the ESP system, vehicle parameters were also varied during the simulator test, i.e. loading conditions, steering parameters and tire

characteristics.To evaluate the ESP interventions on the driving simulator, different driving situations were chosen, all based on straight-

ahead driving on a highway track. Pylons were used for various lane-change and slalom maneuvers. With the aid of different sets of MF-Tyre tire parameters either a dry road surface or a low friction road could be simulated.As a main result, it was proven that the interventions of the ESP system on the driving simulator produced a realistic feel of the system. In addition, the influence

of varied vehicle parameters and loading conditions on the interventions of the systems were shown to be realistic. Yet, for a complete evaluation of the ESP system on the simulator, the simple code used for the first test is not sufficient. For this, it will be necessary to integrate the complete system functionality, i.e. the interaction of the ESP system with the vehicle drivetrain.

RESuLTS AnD FuTuRE WoRKDuring a first test of a truck ESP system on the Daimler driving simulator, it could be shown that the combination of SIMPACK vehicle models with a Software-in-the-Loop code of the ESP system is able to realistically reproduce the stability interventions of the system and thus can be used for optimizations of the system for a wide variety of vehicles and vehicle conditions. In the next step, a complete WABCO code of the truck ESP will be compiled on a Linux system and integrated into SIMPACK as a user routine to obtain the complete functionality of the system on the driving simulator.

Fig. 9: Transfer of SIMPACK models to the driving simulator

Fig. 10: Lane change and slalom maneuver on the driving simulator

“...the combination of SIMPACK vehicle models with a

Software-in-the-Loop code of the ESP system is able to

realistically reproduce the stability interventions of the system...”


CuSToMER APPLICATIon | Alexandra Lehnart, Florian Fleißner, Peter Eberhard, Institute of Engineering and Computational Mechanics, University of Stuttgart

InTRoDuCTIonSIMPACK is a very powerful tool in its application to coupled multi-body system (MBS) simulations of various vehicles. But when considering tank vehicles, the motion of the cargo and the tank design (including the number of compartments or the general shape) need to be accounted for, too. One way to approach this challenge is to use simple non-physical pendulum models for the cargo. However, this is just a crude approximation with little regard to the reality of the tank’s shape and cargo dynamics. Much better results can be obtained by coupling the MBS simulation with another simulation method specifically designed to simulate the liquid cargo material interacting with the tank. In our approach, we couple SIMPACK with PASIMODO (particle Simulation and mOlecular Dynamics in an Object oriented fashion), a Lagrangian simulation framework for the 3D simulation of granular materials and fluids, developed at the Institute of Engineering and Computational Mechanics at the University of Stuttgart.

Simulating Tank Vehicles with Sloshing Liquid Load

The cargo, particularly a liquid cargo, of a transport vehicle can have a significant influence on the driving characteristics of a vehicle. The design of the tank is of upmost importance as it greatly affects the dynamics of the cargo and, therefore, the dynamics and stability of the vehicle. These issues have to be taken into account when performing a simulation. A co-

simulation approach is proposed coupling a multi-body system simulation using SIMPACK for the vehicle with a particle-based fluid simulation using the software PASIMoDo for the cargo.

Fig. 2: Simulink® model of a PASIMODO-SIMPACK co-simulation loop with fixed synchronization time interval

Fig. 1: Tank vehicles with sloshing liquid load

kernel function


Alexandra Lehnart, Florian Fleißner, Peter Eberhard, | CuSToMER APPLICATIon Institute of Engineering and Computational Mechanics, University of Stuttgart

Fig. 4: Exemplary particles with their corresponding moving kernel functions

Fig. 3: Model of the truck with 17 degrees of freedom

DYnAMIC Co-SIMuLATIonThe co-simulation delegates the simulation of the tank and silo vehicles to the two specialized programs. PASIMODO calculates the forces on the tank geometry resulting from the sloshing liquid or moving solid cargo, while SIMPACK calculates the behavior of the vehicle. For this purpose, SIMPACK needs to know the particle forces and torques on the tank to account for their influence on the vehicle’s dynamics. On the other hand, PASIMODO needs the recent tank states to be able to calculate the resulting particle forces. This exchange is established via MATLAB® and Simulink® as SIMPACK already provides a co-simulation interface that allows for data exchange. To couple PASIMODO with MATLAB and Simulink, we took advantage of PASIMODO’s plug-in interface, which enables users to implement their own custom subroutines in C++. These can be used to reposition geometries during a simulation, e.g. the tank’s internal surface geometry according to the recent tank states received from SIMPACK. Then, the plug-in sends the calculated particle forces and torques with respect to the tank’s center of gravity back to SIMPACK. The data exchange is carried out via a TCP/IP interface both between PASIMODO and Simulink as well as between Simulink and SIMPACK. The exchange always takes place after a fixed time interval, but in between, each simulation program can use its own variable time stepping. This is especially relevant for the particle simulation as the particle dynamics solver usually has a much smaller time step than the dynamics solver of the truck. For these time steps, the position of the tank geometry is updated via extrapolation from the already received tank states. Fig. 2 depicts the co-simulation procedure as implemented in the Simulink model.

VEHICLE MoDELFor the simulation of the vehicle, a classical MBS approach is used. The model of the truck has 17 degrees of freedom, schematically shown in Fig. 3. The computation of the tire forces follows the Tire Similarity Model from Pacejka. To account for a driver, two additional degrees of freedom are added which are influenced by a feedback controlled driver model. The truck has a total dead weight of 10310 kg, a wheelbase of 3.1 m and a track of 1.66 m.

PARTICLE MoDELConsidering the MBS model and the particle model, different aspects of the vehicle are of interest. In SIMPACK, the forces and torques on the tank resulting from the moving cargo are supplied by the particle simulation. They can be treated as applied forces on the tank’s center of gravity and no further information on the tank’s geometry is needed. The particle simulation, on the other hand, needs a detailed internal tank geometry and the states of the tank to place it appropriately, but doesn’t need any information about the remaining MBS.The program PASIMODO follows the Discrete Element Method approach for the simulation of granular systems and the Smoothed Particle Hydrodynamics (SPH) method for fluids. SPH is a discretization method in space for partial differential equations, where the discretization points, called particles, carry the important information like velocity, density and pressure as mean values over the volume surrounding them and move along the velocity field of the material they represent. The discrete values at the particle positions are then smoothed by means of a differentiable kernel function, often similar to a Gaussian function, and summed up to yield an approximation of the corresponding function that is also differentiable. Using these approximations in the partial differential equations gives ordinary differential equations in time that can be

solved with a given time stepping scheme. Fig. 4 shows some moving particles with their corresponding kernel functions.Boundaries and non-spherical objects are defined by triangular surface meshes. To prevent particles from passing through the mesh, a penalty force is evaluated and

applied for each pair of particle and triangle. If the mesh is not fixed,

“PASIMODO calculates the forces on the tank geometry, while SIMPACK calculates the

behavior of the vehicle.”


CuSToMER APPLICATIon | Alexandra Lehnart, Florian Fleißner, Peter Eberhard, Institute of Engineering and Computational Mechanics, University of Stuttgart

it is defined relative to a moving coordinate system that lies in the center of gravity and accumulates all the forces and torques on the triangles. The position and orientation of the coordinate system and with it the mesh geometry can then be either integrated in time or, as in our case, taken from the tank states supplied by SIMPACK. The accumulated forces and torques are sent to the MBS simulation.

RESuLTSTo investigate the influence of moving cargo on driving dynamics, classical benchmark driving maneuvers are performed, among Fig. 7: Relative angular velocity at the rear tire

Fig. 6: Normalized displacement of the cargo’s center of gravity in longitudinal direction

Fig. 5: Snapshots during a full braking maneuver with a tank with an elliptical cross section and one and three compartments, respectively

“Regarding the evaluation of the simulation results, the

co-simulation approach offers all the possibilities that SIMPACK

offers, together with those from PASIMODO.”

them a full braking with different tank designs. As cargo, water is used with a total mass of 4700 kg.Regarding the evaluation of the simulation results, the co-simulation approach offers all the possibilities that SIMPACK offers, together with those from PASIMODO. The exchanged data can be stored by MATLAB and loaded later to compare results from different simulations. This includes the states and forces for every body in the simulation. PASIMODO offers the additional ability to track the cargo’s center of gravity.Fig. 5 depicts some snapshots taken from a full braking with an undivided and tripartite tank, respectively. Please note that in the visualization particles moving within the liquid are displayed. The corresponding displacement of the cargo’s center of gravity is shown in Fig. 6. In Fig. 7 the tire locking that occurs during full braking with the undivided tank can be seen, while the compartments in the other tank prevent it.

ConCLuSIonThe results obtained show that the SIMPACK-PASIMODO co-simulation is suitable for predicting the stability of driving maneuvers of transport vehicles. It can be used to investigate the impact of different

tank designs on the behavior of the vehicle system, easily taking into account compartments or curved walls. Comparisons between different tank designs show

the significant influence that the movement of the cargo and the tank design have on the driving characteristics of the vehicle. A positive effect of subdivisions in the tank in longitudinal directions in terms of braking stability can be seen, as they reduce the sloshing motion in that direction.For details about PASIMODO please see www.itm.uni-stuttgart.de/research/pasimodo/pasimodo_en.php. More information about the methods and simulations is provided in the authors’ publications.


Gero Zechel, Michael Beitelschmidt, Dresden University of Technology; | CuSToMER APPLICATIon Helmut Netter, Bombardier Transportation

Development of a SIMPACK user Routine for Dynamic Light Rail Vehicle Gauging

Simulations

More and more cities are seeking public transportation systems that offer safe, comfortable and effective mobility integrating seamlessly into the urban landscape. To meet this demand, made-to-measure vehicles have to be developed that allow high capacity cars on narrow spaced infrastructure. In a joint venture between the Chair of Vehicle Modeling and Simula-tion at the Tu Dresden and LRV (Light Rail Vehicle) Vehicle Engineering at the Bautzen site of Bombardier Transportation, a SIMPACK user Routine has been developed that obtains the dynamic vehicle envelope for arbitrary train/track combinations with the push of a button.

LIGHT RAIL VEHICLE GAuGInGThe complex kinematic and highly dynamic behavior of modern articulated light rail vehicles require elaborate research of the resulting dynamic vehicle envelope and the structure gauge needed to rule out collisions of vehicles with other vehicles and trackside objects.The variety of vehicle configurations, the uniqueness of infrastructures, and the generally low vehicle quantities common in light rail demand flexible and efficient gauging methods. While kinematic studies are crucial for the early conceptual design, all dynamic factors have to be included for design verification at the end of the development phase.Conducting early gauging with simplified kinematic models and integrating gaug-ing simulations into full-scope multi-body simulations, like the one shown in Fig. 1, throughout development of a vehicle is highly beneficial.

InPuT ConCEPTEven for sophisticated MBS models, the outer shape of the vehicle can be completely ignored. For gauging, however, the precise vehicle contour is vitally important. To make

it available in the models without relying on CAD data, SIMPACK Input Functions have been chosen. They can be used to describe the contour as the car width over the car center plane. To define the origin of these contours on the car, body fixed markers can be created separately. This way, the contours can be created in

Fig. 1: MBS model of a five car 100% low floor tram

Fig. 2: The vertical grid (blue) on top of the track centerline (blue, dotted) used as the reference system for all track related data collected during a run

s_Track / m

30 35 40 45 50 55 60 65 70 75 80 85 90 95 100 105 110 115

Gauge +

Y / m

1.100

1.140

1.180

1.220

1.260

1.300

1.340

1.380

1.420

1.460

1.500

1.540

1.580

1.5418

1.2000

result.KinTest_Wgt1.Gauge +Y over Track



s_Track / m

30 35 40 45 50 55 60 65 70 75 80 85 90 95 100 105 110 115

Gauge +

Y / m

1.100

1.140

1.180

1.220

1.260

1.300

1.340

1.380

1.420

1.460

1.500

1.540

1.580

x_C

onto

ur

/ m

-2.5

-2.0

-1.5

-1.0

-0.5

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

5.5

3.756


result.KinTest_Wgt1.Causative Contour Point for +Y

x_Contour / m

-2.5 -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5

y_C

onto

ur

/ m

| G

auge +

Y / m

0.200

0.300

0.400

0.500

0.600

0.700

0.800

0.900

1.000

1.100

1.200

1.300

1.400

1.500

1.600

1.5418

3.756

result.KinTest_Wgt1.Contour +Y over Contour X

result.KinTest_Wgt1.Gauge +Y over Contour X


CuSToMER APPLICATIon | Gero Zechel, Michael Beitelschmidt, Dresden University of Technology; Helmut Netter, Bombardier Transportation

substructures, varied using substitution values, stored in databases, and reused over several cars or models. For gauging simulations using these contours a SIMPACK User Routine has been developed.

IMPLEMEnTATIonTo achieve a seamless integration into the SIMPACK user interface, the concept was realized by means of a SIMPACK User Result Element written in Fortran using the built-in editing and compiling tools with a third-party compiler.This User Result Element can be integrated in any vehicle model by using the Model Setup GUI. In the Result Element window, the Input Function and the Marker can be picked from the list of available MBS elements and output options can be set. Although one Result Element represents only a single contour, it can be added as many times as needed to cover the whole vehicle. In general, Result Elements are evaluated offline when Measure-ments are performed. They are called by SIMPACK beforehand for initialization, then at every communication point, and afterwards to write the output vectors.While the measurements are running, four data collections are built up and stored in memory allocated by SIMPACK using the available Access Functions. The first data collection keeps state information

Fig. 5: Grey: The left side contour of the first car (view from above, car pointing to the right). Red: The dynamic envelope of the car during the whole run (relative to the car reference system)

Fig. 3: The maximum lateral distance of the first three cars to the track centerline, plotted over the track length

Fig. 4: The first car’s maximum distance along the track (black), along with a curve indicating the contour point responsible for the maximum value (blue)

“The concept was realized by means of a

SIMPACK User Result Element.”

that is needed for every time step, like the number of contour points, the track length, and a time step counter. The second data collection is more complex — it stores all values related to the infrastructure. To do so, a grid is spanned over a vertical surface positioned on top of the track centerline, as shown in Fig. 2 in blue. Every point in the area around the track can be assigned to a unique grid point as long as the distance from the track is less than the smallest track radius — this covers the relevant area at all times. The maximum centerline distance any contour point reaches within each of the resulting volume segments (Fig. 2, shown in red) is gathered during the measurement run.To get this information, an algorithm is run for each time step that circles through all contour points one by one and calculates their distance to the grid using an iterative method. If a new maximum distance is found along the track, this value, as well as the index of the contour point responsible for the new maximum and the absolute position of the contour point, is stored in the memory.The third data collection is filled in a similar way — but stores values in reference to the contour itself. At the end of the run, it therefore contains the maximum distance from the track for each contour point along with the information at which grid point this maximum occurred. Finally, the fourth data collection gathers time-domain data like the absolute position of each contour point to certain time steps.While memory usage stays reasonably low, these four data

collections are not suitable for plotting as is, so they are evaluated at the end of the run to gain comprehensive one-dimensional output vectors. This data is stored in the resulting standard result-file and can be easily plotted

using the SIMPACK PostProcessor.

DATA EVALuATIon AnD PLoTTInGFirst and foremost, the track related data collection is evaluated to get the maximum distance of the vehicle to the track centerline over the track length. Two output vectors are created for the left and for the right hand side of the track. Fig. 3 shows the results for a kinematic vehicle model passing through a right turn. Three Result Elements where defined for the first three cars of the vehicle. In the plot, the results for the left-hand side of the track for each car are overlaid by drag-and-drop in the SIMPACK PostProcessor. The plot shows that the maximum distance from the track at curve entry and

s of Track

-1.4 -1.2 -1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4

Gauge -

Y o

ver

Tra

ck

0.00

0.06

0.12

0.18

0.24

0.30

0.36

0.42

0.48

0.54

0.60

0.66

0.72

0.78

0.84

0.90

0.96

1.02

1.08

1.14

1.20

1.26

1.32

1.38

1.44

1.50

1.56

1.62

1.68

1.74

1.80

1.86

1.92

1.98

2.04

2.10

2.16

2.22

2.28

2.34

2.40

2.46

2.52

2.58

2.64

2.70

2.76

2.82

2.88

2.94

3.00

3.06

3.12

3.18

3.24

3.30

3.36

3.42

3.48

windloadonstraight.result.GaugingExampleQ.TrackSlice s= 20.00 .. 22.00 H(Y)

nowindloadonstraight.result.GaugingExampleQ.TrackSlice s= 20.00 .. 22.00 H(Y)

TopDownView Indep Y

-4 -2 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40

TopD

ow

nV

iew

X(Y

)

30

32

34

36

38

40

42

44

46

48

50

52

54

56

58

60

62

64

66

68

70

72

74

76

78

80

result.Gauge_Wgt1.TopDownView Track Centerline X(Y)

result.Gauge_Wgt3.TopDownView Movement Visualiz. X(Y)



result.Gauge_Wgt1.TopDownView Envelope -Y X(Y)

result.Gauge_Wgt1.TopDownView Envelope +Y X(Y)






Gero Zechel, Michael Beitelschmidt, Dresden University of Technology; | CuSToMER APPLICATIon Helmut Netter, Bombardier Transportation

part of the curve exit is caused by the second car (red), while the first car (black) leads to the maximum distance along the track between curve entry and exit (1.54 m).To provide information about what part of a contour is responsible for maximum displacement, the track related data collection is also filtered to produce the output vector shown in Fig. 4. The plot indicates that the contour point located 3.76 m in front of the bogie (where the body fixed marker is placed) results in the maximum distance for the given example. Finally, the evaluation of the contour-related data collection provides the outputs shown on Fig. 5. The left side contour of the first car is depicted in grey, the dynamic envelope of the car during the whole run is shown in red. This plot confirms the values found above, and shows that the head contour needs to be changed to lower the maximum distance to 1.4 m from the track centerline (everything in the range of 3.5 m to 4.3 m in front of the bogie).By applying more advanced evaluation techniques to the track related data collection, plots are made possible that show the structure gauge needed by the vehicle along the track. This can be helpful when analyzing the rolling behavior in curves, the clearance at railway platforms and the influence of crosswind loads, Fig. 6, for example. If gauging data is needed in the context of the simulated scenario, absolute positions stored in the track-related and the time-domain data collection can be combined to get a general idea of the vehicle’s kinematic and dynamic gauging behavior. Fig. 7 shows an example plot of three cars passing an S-bend. In addition to the track centerline and the car contours at two different time steps, the kinematic envelope of each car projected to the ground can be seen. Fig. 8: Dynamic model of a 5-car tram passing a gate after a curve (top view)

Fig. 7: Top-down view of an S-bend track showing the track center-line, the position of three car contours at two different time steps, and the kinematic envelopes of each car projected to the ground

Fig. 6: Structure gauge for a vehicle on a straight track with and with-out crosswind load

ConCLuSIonWhen combined in the right way, SIMPACK’s built-in tools and functionalities can be used for gauging simulations with high accuracy, even for highly dynamic situations like the one shown in Fig. 8. The complexity of the Gauging User Routine can be reduced to consider just one single contour, because it can be utilized many

times in a single model at the same time and the results can be overlaid in the PostProcessor. Its use, however, is not restricted to the car body — it can also be applied to pantographs, bogie parts, or even markers fixed to the inertial system to include objects of the

infrastructure like railway platforms or masts of the catenary wire.This extreme versatility is possibly the biggest advantage compared with many other gauging software tools as it allows for the analysis of a huge variety of vehicle designs and infrastructure systems which LRV-engineering has to cope with every day.

“...SIMPACK’s built-in tools and functionality can be used for gauging simulations with

high accuracy...”


CuSToMER APPLICATIon | Berthold Schlecht, Thomas Rosenlöcher, Institute of Machine Elements and Machine Design, Chair of Machine Elements, Dresden University of Technology

Determining Reliable Load Assumptions in Wind Turbines using SIMPACK

The intensified efforts to provide alternative and renewable sources of energy led to a substantial worldwide growth in the wind turbine industry over the last few years. The advantages of this clean energy source and the recent successes in increasing power output rates for on- and offshore wind turbines is extremely encouraging. However the need for more durable wind turbines must still be addressed. The operation of anchored flexible light weight constructions under high dynamic stochastic loads was a relatively new challenge that arose with the wind turbine. These challenges are being met with the aid of the multi-body system (MBS) software SIMPACK.

InTRoDuCTIonThe Chair of Machine Elements at Dresden University of Technology does research which, for the past several decades, has focused on machine elements like shafts, gearings and bearings. Since 2001, MBS software has played an integral role in the development of guidelines, standards and verification of the dynamic behavior of drivetrains. MBS software has been used to develop modeling strat-egies that realistically represent the dynamic behavior of machine elements and have a high correlation with measured data. The main focus of research has been the investigation of the dynamic behavior of large drivetrains which can be found in roller mills, compressors, ships, fans, shearers, cranes and wind turbines.

MoTIVATIonIn comparison to the wind turbine design software used by wind turbine manufacturers, multi-body system simulation software allows for a more precise modeling of the drivetrain components. Instead of a detailed rotor model and a simple 3-mass torsional vibration model for the gear box and the generator, all components can be represented with up to six degrees of freedom and with coupling stiffnesses. The resulting simulation model of the wind turbine consists of the substructure's rotor, drivetrain, coupling, generator, supporting structure, tower and foundation (Fig. 1). It allows the determination of the natural frequencies of the complete structure, and additionally, shows the mode shapes of all drivetrain components. To detect possible ranges of resonances, the effects of the rotor, the gear meshing frequencies, and rotation speeds of the components can be compared to the calculated natural frequen-cies. Simulation in the time domain enables the determination of torques, forces, displacements, velocities and accelerations for the modeled components and degrees of freedom. The resulting values at the different load conditions can be used for the design of the components, bearings and gearings, as well as the gear box housing and supporting structure.

BASICS oF DRIVETRAIn SIMuLATIonThe analysis of drivetrains operating under high dynamic loads pre-supposes the assembly of a detailed simulation model which is able Fig. 1: Flexible multi-body system model of a wind turbine drivetrain

Fig. 2: Gear box substructures


Berthold Schlecht, Thomas Rosenlöcher, Institute of Machine Elements and Machine Design, | CuSToMER APPLICATIon Chair of Machine Elements, Dresden University of Technology

to represent the dynamic behavior of the drivetrain in the frequency and time domain. Even if high performance computers are available the level of detail of the simulation model has to correspond to the formulated question to ensure a feasible calculation effort. Based on the available data and the experience of the engineer, a discrete simulation model can be assembled. A suc-cessive and modular assembly of fully param-eterized simulation models allows a clear and reproducible modeling process. The modular concept requires decomposition of the drivetrain into its substructures. Using this approach, a simulation model of a com-

mon wind turbine consists of the following substructures: rotor, main shaft, coupling, generator and an additional subdivided gear box. For the gear box, a combination of helical/spur and planetary gear stage substructures is necessary (Fig. 2). Each substructure consists of model components which can be subdivided into shafts, gear stages, bearings and supporting structures. This combination of single substructures leads to the complete simulation model of the wind turbine.

Fig. 3: Modeling of shafts by discretisation, SIMBEAM model and FEM-method

Fig. 5: Variable gearing stiffness over the contact path using Fourier coefficients

Fig. 4: Gear box of a 3 MW wind turbine

“The degrees of freedom of each substructure can be adjusted to

accommodate varying degrees of detail of the overall model.”

The degrees of freedom of each substructure can be adjusted to accommodate varying degrees of detail for the overall model. Sub-modeling enables easy verification of the appropriate level of model detail.

MoDELInG oF DRIVETRAIn CoMPonEnTSThe dynamic behavior of a drivetrain results from the gear ratio and the distribution of the mass, mass of inertia and stiffness. The de-termination of mass parameters is possible with three-dimensional CAD models or by using simple analytical approaches. Great effort is required to accurately calculate the various stiffness parameters for all of the drivetrain components.The torsional stiffness of the drivetrain is mainly characterized by the stiffness of the shafts. Special consideration must be given to slender shafts whose elastic properties need to be accounted for. Additionally, the bending stiffness of such shafts may have consid-erable influence on the dynamic behavior as well as the resulting displacements. The required simulation model can be assembled in three ways: a) by using the method of discretization,

b) via the beam approach or c) by implementing modally reduced finite-element models (Fig. 3). For models that include the axial and ra-dial motion of the shafts, the properties of bearings must be accounted for. Essentially, the modeling of the bearings is realized by

a force element which introduces the reaction forces in the axial and radial directions as well as the reaction moments, if necessary. The bearing properties can be described by average bearing stiffness, characteristic curves or complex models imported as DLLs.The transmission ratio between the rotor and the high speed gen-erator can be realized by a gear box consisting of a set of planetary and helical gear stages (Fig. 4). The changing speeds and torques in a gear box as well as the varying gearing stiffness resulting from the total overlap ratio have an important influence on the dynamic behavior of the drivetrain and must be considered in the simulation model. SIMPACK offers the special force element Gear Pair (FE 225)


CuSToMER APPLICATIon | Berthold Schlecht, Thomas Rosenlöcher, Institute of Machine Elements and Machine Design, Chair of Machine Elements, Dresden University of Technology

which enables a comfortable modeling of gearing. An alternative modeling approach offers the mathematical description of the re-sulting forces in the gearing by user routines. Based on the calcula-tion of the tooth normal force in the ideal pitch point, the complete tooth contact is simplified and described in one point. The tooth normal force consists of stiffness and damping dependent parts. Information about the displacements and velocities in tangential, radial and axial directions resulting from the relative position of the gears can be determined from the joint states and the corresponding trigonometric relationships. The gearing stiffness can be considered as average contact stiffness according to DIN 3990 and variable gearing stiffness over the path of contact using Fourier coefficients (Fig. 5). In order to improve the model even further, modeling of the cou-pling and generator is necessary. The most important influence on the dynamic behavior results from the rotorblades. A simplified approach for modeling of the rotorblade stiffness can be done by splitting the blades into mass segments coupled by spring damper elements to represent the bending stiffness (Fig. 6). Therefore, the information of the mass and stiffness distribution as well as the natural frequencies in edge- and flapwise direction are sufficient. In addition, if the information of every profile section is available, the rotorblade generation in SIMPACK offers an automated procedure. A detailed modeling of the shafts, bearings, gearings, coupling, generator and rotor allows the representation of the dynamic interaction between the components and the determination of displacements, velocities, accelerations, forces and torques. Even if spring-damper elements are used to sup-port the components, the surrounding structure, e.g. the gear box housing or the main frame, is assumed to have a stiff coupling to the global reference system. Thereby the influences resulting from the flexible structure of a wind turbine are neglected. The enhance-ment of the stiff multi-body system model using modally reduced finite-element structures allows the consideration of these effects. The implementation of a flexible structure in SIMPACK is based on a meshed finite-element model of the component geometry and the definition of the material properties. The connection points to the

Fig. 6: Discretisation of a rotorblade

supporting spring-damper elements can be modeled by means of multi-point constraints (MPC). Due to the resulting complex-ity and degrees of freedom of the FEM-models a reduc-tion of the structure is required. The application of the approach according to Craig-Bampton requires the definition of the connec-tion points between the flexible structure and the rigid bodies. The mode shapes of the reduced model are used to determine the deformation under load. The number of natural frequencies chosen for the modal reduction defines the valid frequency range

and the accuracy of the model, which is also influenced by the choice of frequency response modes in the SIMPACK Add-On Module FEMBS.

The implementation of flexible structures allows representation of the flexibility of the supporting structure as well as the consider-ation of the stiffness of the drivetrain com-ponents, e.g. shafts and planetary carriers, with a higher degree of accuracy.

AnALYSIS oF nATuRAL FREQuEnCIES AnD EXCITATIonS The resulting flexible multi-body system model allows the determination of the natu-ral frequencies and can take into account various degrees of freedom. The resulting frequencies can be compared to the excita-tion frequencies to determine possible reso-nances. Relevant excitations are the first, second, third, sixth (ninth, twelfth) order of the rotor rotation frequency. Additionally, the first and second orders of the rotation frequencies of all drivetrain components, as well as the meshing frequencies of the gear stages, have to be considered. The compari-son of natural frequencies and excitations by means of a Campbell diagram reveals possible resonances. The analysis of the corresponding mode shapes allows further statements as to whether the excitation of a natural frequency can cause critical opera-tion points or not.

AnALYSES In THE TIME DoMAInThe detailed flexible multi-body system model also offers the possibility of determin-

Fig. 7: Wind loads on flexible rotorblades

“SIMPACK offers the special force element Gear Pair (FE 225)

which enables a comfortable modeling of gearing.”


Berthold Schlecht, Thomas Rosenlöcher, Institute of Machine Elements and Machine Design, | CuSToMER APPLICATIon Chair of Machine Elements, Dresden University of Technology

ing resulting displacements,

deformations, velocities, accelera-

tions, forces and torques under the dynamic loads

resulting from the stochastic wind field. To model a realistic

wind field, different approaches are available. The common wind

turbine design software tools like Bladed, Flex5 and AeroDyn are mainly

used to define the design loads. An in-terface between AeroDyn and SIMPACK is

available (FE 237: Wind AeroDyn*)which offers the possibility of determining wind loads based on the rotorblades, tower and wind turbine parameters. If all the required information is not given, different simplified approaches based on measurements can be used. A wind model based on the information of the wind speed, power coefficients and assumed load distribu-tions over the blade length and the height can also be used. Using nine segments for each blade and by superimposing a stochastic wind field, the tower shadow can be calculated. An enhancement of this approach replaces the as-sumptions for the wind field and load distribu-tion by measurement results captured as torque and bending moments at the main shaft. An algorithm adjusts the single blade forces to achieve the measured results on the main shaft in the simulation model so that the measured states of the wind turbine can be represented by the model. The changing forces at the flex-ible modeled rotorblades are shown in Fig. 7 as arrows and scaled deformations. The operation of the wind turbine under dif-ferent load conditions and extreme load cases like emergency stops can also be calculated. The resulting speeds, torques and forces in the gearing of the first planetary gear stage are shown in Fig. 8. In addition, the information of displacements of the main shaft, gear box and gear box components like the sun shaft (Fig. 9) can be obtained. This data can lead to a deeper understanding of the dynamic behavior of the system and the resultant loads under different load conditions.

Fig. 8: Emergency stop simulation, speed, pitch angle and force

Fig. 9: Displacement of the sun

ConCLuSIonWind turbines are anchored flexible complex drivetrains which oper-ate under highly dynamic stochastic loads. For onshore wind tur-bines, and especially for offshore wind turbines, the request for high reliability requires comprehensive knowledge of dynamic behavior already in the design phase. This includes information about possible excitations and natural frequencies which can cause resonances in the operational speed range. Additionally, the displacements, defor-mations and resulting forces in the drivetrain, as well as the influ-ence of wind turbine control under maximum loads during normal operation and emergency cases, has to be analyzed. The multi-body method offers the ability to realistically model a wind turbine while considering all relevant components and degrees of freedom. This approach enables the required knowledge to be obtained in order to fully understand the dynamics of wind turbines.


CuSToMER APPLICATIon | Markus Kochmann, Milan Ristow, Germanischer Lloyd Industrial Services GmbH

Simulation of Drivetrains on Wind Turbines within the Framework of Certification — with SIMPACK

Programs such as Bladed or Flex5 are well suited for calculating loads in order to verify the general stability of a wind power plant. These programs have their strengths in the investigation of the dynamic behavior of the overall system. However, the analysis is usually limited to a small range of frequen-cies. The investigation of phenomena in a larger range of frequencies requires a more detailed representation of the system. SIMPACK simulation software provides an excellent platform for this detailed model and analysis.The drivetrain of a wind turbine is an inte-gral and expensive assembly component of the plant. Manufacturers of transmissions, manufacturers of wind turbines and certifi-cation bodies (such as Germanischer Lloyd Industrial Services GmbH, Renewables Cer-tification (GL)) engage in expansive efforts to put a reliable design into practice. One critical aspect of the design is the dynamic behavior of the drivetrain, which plays an important role in the determination of local loads and the physical integrity of the me-chanical components.

MuLTI-BoDY SIMuLATIon MoDELInG Multi-body simulation models are frequently used for the determination of natural frequencies and the investigation of the dynamic behavior of the drivetrain. The possible degrees of detail for such models range from simple mass shock absorber systems, with rotatory degrees of freedom, up to very complex systems with flexible bodies and super-elements for the consideration of the flexible housings and support structures.Often a simple model provides sufficient information regarding the dynamic behavior of a complex dynamic system and helps with the investigation and understanding of dynamic phenomena. However, for special problems, more detailed and more complex models are necessary.

TYPE CERTIFICATIon For several years, the investigation of the dynamic behavior of the drivetrain has been part of the Type Certification of wind turbines. A type certification is the confir-mation of conformity for adherence to fixed re-quirements (e. g., guide-lines and standards) for certain types of wind tur-bine. An important com-ponent of a type certification is a thorough design assessment. The certification proce-dure is based on the international standard IEC 61400-1 [1], [2] and/or the GL guidelines for the certification of wind turbines [3], [4] and [5].

CoMPonEnT oF THE DESIGn ASSESSMEnT As a component of the design assessment, an independent evaluation of the dynamic behavior of the drivetrain is conducted by

GL. The necessary model data are derived from technical drawings, CAD and FEM models. On the basis of such information, a realistic representative multi-body simulation model can be prepared. SIMPACK is used by GL in order to investigate the dynamic behavior of the drivetrain. In July 2010, the revision to guideline [4] was published as "GL 2010" [5]. Experi-ences from various certifications, research projects, discussions between GL and

external experts and, above all, the technical specialized committee (with many specialists from the wind energy industry) have led to a

new issuance which considers state-of-the-art knowledge regarding the development of wind turbines. In the "GL 2010" guidelines [5], an entire application-oriented appendix is dedicated to the investigation of dynamic behavior of the drivetrain. The recommendations assume investigations with the use of multi-body simulation systems. The study of the dynamic behavior of the drivetrain on the basis of "GL 2010" will lead to the following changes:

Simulation with multi-body systems is relatively new in the emerging wind energy industry. Representative tests on a constructed model are practical only on a limited basis be-cause of the long draft lifespan of 20 years. Realistic simula-

tion is of paramount importance in product development.

Fig. 1: 3D MBS model of wind turbine (including flexible blades, detailed drivetrain and generator)

“SIMPACK is used by GL in order to investigate the dynamic behavour

of the drivetrain.”


Markus Kochmann, Milan Ristow, Germanischer Lloyd Industrial Services GmbH | CuSToMER APPLICATIon

SIMPACK AG | SIMPACK nEWS

• Torsion, bending and axial degrees of freedom must be considered.

• Simulation models with only rotational degrees of freedom can be used (if the calculation results are verified with measurements).

• For certain analyses, the simulation of a run-up must be conducted in the time domain.

In part, such new requirements are arising from current EU co-funded research projects that are concentrating on the validation of simulation models through measurement. Together with manufac-turers of wind power plants and transmissions, research institutions and universities, GL is involved with the PROTEST research project.

One emphasis is the measurement of a drivetrain of a wind power plant of the megawatt class. At the same time, results arising from simulations with SIMPACK are compared with the measurement results. Further information is available at the Internet site for the project (see www.protest-fp7.eu).

ConCLuSIonIn order to obtain a more in depth view of the dynamic behavior of complex systems (e.g., the drivetrain of a wind power plant), a clear trend to more complex and validated simulation models is imperative.SIMPACK is high-performance and reliable simulation software is an important tool for achieving this goal.

InFoRMATIonFor more information about GL and renewables please see: www.gl-group.com/GLRenewables

REFEREnCES[1] IEC IEC 61400-1 “Wind turbine generator systems – Part 1: Safety require-ments”, 1999. 2nd edition, February 1999.[2] IEC 61400-1 “Wind turbines – Part 1: Design requirements”, 3rd edition, August 2005.[3] Germanischer Lloyd, Hamburg, Germany: “Guideline for the Certification of Offshore Wind Turbines”, 2005 Edition.[4] Germanischer Lloyd, Hamburg: “Guideline for the Certification of Wind Turbines”, Edition 2003 with Supplement 2004.[5] Germanischer Lloyd, Hamburg: “Guideline for the Certification of Wind Turbines”, 2010 Edition.

Fig. 2: Detailed SIMPACK model

SIMPACK Academy Events in 2010

MBS nuMERICS ACADEMY, 14.–16.09.2010, in Andechs, Southern Germany

Prof. Dr. Martin Arnold, Institute of Mathematics, Martin Luther university Halle-Wittenberg and Dr. Gerhard Hippmann, Solver Technology, SIMPACK AG

“Multi-Body System Numerics“ aims to enable the calculation engineer to understand the capabilities and limits of numerical methods in multi-body dynamics. Elaborate presentation of the theoretical background is combined with software issues as well as practical examples and tips.

WInD TuRBInE AnD DRIVETRAIn ACADEMY, 05.–07.10.2010, in Gross Schwansee, northern Germany

•Part1:DynamicsandSystemDesignofWindTurbines,05.–06.10.2010 Prof. Martin Kühn, ForWind — university of oldenburg and Stefan Hauptmann, university of Stuttgart

This course combines theoretical background with recent industrial experience to provide a comprehensive introduction to state-of-the-art wind turbine dynamics and design from a system´s view-point.

•Part2:DesignandAnalysisofDrivetrainsinWindTurbinesandOtherLargeIndustrialApplications,06.–07.10.2010

Prof. Berthold Schlecht and Thomas Rosenlöcher, Technical university of Dresden

This course covers the dimensioning and design of drivetrain systems and their components. In addition, the advantages and accuracy of multi-body simulation (MBS) will be shown.

For more information and registration please visit: www.SIMPACK.com/SIMPACK_academy.html

76 elastic markers wing,

MBS joint fixed support / adjustment of angle of attack / heave- and pitch exitation,

MBS joint for the definition of wind tunnel co-ordinate system (Co-ordinate system for the exchange of loads / deformation)


CuSToMER APPLICATIon | Jürgen Arnold, Wolf Krüger, Gunnar Einarsson, Deutsches Zentrum für Luft- und Raumfahrt, Göttingen

CouplingofMBSandCFD: an oscillating Aeroelastic Wing Model

Fig. 1: CFD model of the AMP wing

Fig. 2: MBS model of the AMP wing

AERoELASTIC SIMuLATIon uSInG SIMPACK AnD CFDAeroelastic simulations in terms of pure fluid-structure interaction have reached a satisfactory level of maturity for both steady and unsteady problems. A step beyond this classical scope is the additional consideration of large motions superimposed by flight maneuvers. In DLR, such a coupling has been developed based on elastic multi-body systems coupled with CFD calculation. The intention of this article is to describe the model set-up used for the coupled calculations, as well as to describe the options to introduce MBS-generated motions into the CFD calculations. Results are given for the so-called AMP wind tunnel model for heave and pitch oscillations at a Mach number of 0.6. A more detailed illustration of the work as well as a list of references is given in [1].

GEnERAL SET-uPFor the simulation of a complete elastic aircraft using MBS, the flight mechanics (FM)are represented as non-linear MBS joints. This approach is possible for transport aircraft, where the flight mechanics can be a 6-degree-of-freedom joint, but also for wind tunnel models with a reduced number of degrees of freedom or for helicopters, where the complex kinematics of the system (e.g. the rotor hub) can be introduced. The elastic members are included in modal form

Multi-body simulation has been shown to be a valuable software tool for virtual aircraft design. It is a standard approach for the analysis of landing gears, of air-craft on the ground, and for the design of high-lift systems. The medium level of complexity of typical multi-body mod-

els also makes it a suitable tool for the application of flight mechanics in combination with elastic deformations. The de-velopment of reliable aerodynamic models, in addition to the existing interface for complex elastic structures, has been a major activity in the DLR Institute of Aeroelasticity during the past several years. The coupling procedure of SIMPACK to CFD is shown in this article for the application of the Aeroelastic Model Programme (AMP) wind tunnel wing simulated with SIMPACK and the DLR TAu code.

CFD

(CSS)

MBS

No. of Iterations [-]

dz [m

]


Jürgen Arnold, Wolf Krüger, Gunnar Einarsson, Deutsches Zentrum für Luft- und Raumfahrt, Göttingen | CuSToMER APPLICATIon

Fig. 3: Temporal coupling scheme

via the FEMBS interface, the equations of motion solved by the MBS tool. The CFD-based aerodynamics are calculated by a dedicated CFD solver and coupled to the MBS system via co-simulation.

CFDThe behavior of the flow around the wing is simulated with the TAU Code, a CFD tool developed by the DLR Institute of Aero-dynamics and Flow Technology. The TAU Code solves the com-pressible, three-dimensional, time-accurate Reynolds-averaged Navier-Stokes (RANS) equations using a finite volume formulation. The TAU Code is based on an unstructured grid approach, capable of using hybrid grids.The TAU Code functionality is organized into modules. The following modules have been used for the process described in this paper: the Preprocessor module, which uses the information from the initial grid to create a dual-mesh; the Solver module, which performs the flow calculations on the dual-mesh and applies guided rigid body motions when specified; the Deformation module, which propagates the deformation of surface coordinates to the surrounding grid; and the Postprocessing module, which is used to convert TAU Code result files to formats readable by popular visualization tools.The Solver module can be executed in Euler mode, or using Navier-Stokes (RANS) equations with 1-Equation or with 2-Equation turbulence modeling. The results shown in this paper are all based on the Euler mode. This is mainly done for reasons of computation time. The coupling procedure as such is identical for RANS calculations. For steady calculations, an explicit multistage

Runge-Kutta time stepping scheme is used. For time-accurate computations, an implicit dual-time stepping approach is used. Fig. 1 shows the aerodynamic model of the AMP wing used in the work.The TAU Code modules have been wrapped with Python interfaces, and can thus be used as library functions from within a Python

script. To couple TAU to SIMPACK, the TAU Solver is called from a coupling script. The flight mechanic data calculated by SIMPACK is sent to

the Motion module of the TAU Code, which builds the required transformation matrices used by the Solver module. The data transfer

“The data transfer is performed via the Python

program and a socket connection using the SIMPACK IPC

co-simulation interface.”

is performed via the Python program and a socket connection using the SIMPACK IPC co-simulation interface.

STRuCTuRAL MoDELThe structural model of the wing has been set up in the FE code NASTRAN and has been subject to a modal analysis. The results have been exported to SIMPACK using the FEMBS Interface. The model used in SIMPACK consists of a wing model and model support represented by 76 markers on the elastic structure and 20 elastic modes. Fig. 2 shows the MBS representation of the structural model including the used reference frames and joints defined for prescribed motion.

SPATIAL AnD TEMPoRAL CouPLInGDue to the different discretizations of the CFD and the elastic MBS model, dedicated routines for spatial coupling have to be used. For time-marching simulation, a temporal coupling scheme has to be employed. Spatial coupling of SIMPACK to TAU is realized via the DLR inhouse development PyCSM. The approach makes use of node-based, conservative interpolation methods to map aerodynamic forces between structural and aerodynamic grids, and a non-conservative interpolation to map deformations. CFD and MBS codes exchange their results (forces/deformation) at each simulated time step in co-simulation through a TCP/IP socket. The communication scheme is

Fig. 4: Results of quasi-steady coupling

the same for both approaches with steady and unsteady aerodynamic forces. Values of 0.0353 m and 1853 N are obtained in the wind tunnel coordinate system. To validate the result, pressure distributions of the experiment have been compared to the results obtained with CFD/MBS. Numerical and experimental results correspond very well; see Fig. 5 for data at a location of 69.2 % wing span.

HEAVE AnD PITCH oSCILLATIonS Two different approaches have been investigated to find the response of the elastic AMP-Wing to the forced heave and pitch excitations at the model support. The first approach represents the rigid body motion due to heave excitation and the elastic wing deformation from aerodynamic loads together in the TAU Deformation module. The second approach uses the TAU Motion module to represent the heave or pitch, and the Deformation module to represent the elastic deformation only. In the latter approach, the Motion module is supplied with twelve flight mechanics (FM) parameters. The FM-parameters are comprised of three angles and angular rates, each in the body coordinate system and three translations and translational velocities each defined in the geodesic coordinate system. They are measured as MBS sensor data and communicated through a Python-shell to the TAU Motion module. The oscillating deformation at the wing tip in the z-direction and the constrained force

the so-called 'Conventional Serial Staggered CSS', a first-order scheme, see Fig. 3. This approach allows TAU to run on a high performance computing cluster using highly parallel computation, if required.

SIMuLATIon RESuLTS AnD CoMPARISonThe test cases used for comparison are simu-lations for a Mach number of 0.6, a pressure of 0.9 bar, and an angle of attack of 2.03°. Three different configurations are regarded, first a steady state solution, second a sinusoi-dal heave oscillation at the model support of f = 12.27 Hz and an amplitude of ∆z = ±0.0462 m, and third a pitch oscilla-tion at the same frequency with a pitch of ∆α = ±0.5° around the same point. For the pitch and the heave case, the resulting motion of the wing tip will be a combina-tion of rigid body motion of the excitation plus an elastic structural deformation.

STATE oF EQuILIBRIuMThe state of equilibrium for the deformed AMP-Wing is computed with two different approaches, both starting from the undeformed wing shape. First, a quasi-steady coupling procedure, taking five coupling steps into account (see Fig. 4), second, a transient simulation for the physical time t of 1.0 s using time-accurate unsteady aerodynamics and 500 co-simulation steps. The resulting static deformation at the wing tip in the z-direction and the constrained force in the z-direction at the support are

in the z-direction at the model support are the same for both approaches. This is true for the heave as well as for the pitch case. Fig. 6 shows the corresponding time histories of the wing tip deflection for the heave case. The red line represents the total deflection; the blue line the purely elastic part of deflection. Pitch oscillation data look very similar and are given in [1]. Unfortunately, no experimental results for direct comparison are currently available for these cases.

ouTLooKThe work described has been a test case for the interface of SIMPACK to CFD aerodynamics, i.e. the TAU Code. This coupling is of great interest for aeronautical applications both in the area of fixed wing aircraft and for helicopters. The implemented approach forms the basis of an extensive application of MBS/CFD coupling pursued by the DLR Institute of Aeroelasticity. REFEREnCES[1] Arnold J., Einarsson G., Krüger W. R. (2009): "Multibody simulation of an oscillating aeroelastic wing model." NAFEMS International Journal of CFD Case Studies, Volume 8, pp 5-18.


CuSToMER APPLICATIon | Jürgen Arnold, Wolf Krüger, Gunnar Einarsson, Deutsches Zentrum für Luft- und Raumfahrt, Göttingen

Fig. 6: Wing tip deflection for wing heave motionFig. 5: Comparison of pressure distribution


Steven Mulski, SIMPACK AG | SoFTWARE

Gear Pair Enhancements with SIMPACK Version 8904

HISToRYInitially developed for Formula 1 high performance engines back in 2003 (by Lutz Mauer, an executive board member of SIMPACK AG), the SIMPACK Gear Pair functionality has since been used in a large variety of industrial sectors, e.g. automotive, wind, rail, shipping, aerospace, concrete mills, material handling, etc.

GEnERALIn SIMPACK, a large variety of elements are available for the simulation of torque converters. Depending upon the task at hand, elements of various level of detail may be used for achieving the optimum balance between solver speed and accuracy. For example, simple one-dimensional elements may be used for torsional analyses whereas gearbox elements (e.g. planetary gear stage) may be used for more detailed analyses when reaction moments on the housing are required. For simulations where individual tooth contact forces are required, the SIMPACK Gear Pair force element, FE 225, may be used. This element enables the additional analyses of the meshing forces and moments, shaft bending, bearing

forces, and a host of other pertinent analyses (Fig. 2).The gear pair FE 225 is an analytical element, and therefore, extremely fast simulation times can be achieved. Graphical primitives are defined for the gear wheels which are subsequently used for the force calculations. This results in accurate animation of the gear tooth contacts and play.

The element includes the following functionality [1, 2]:• Involute spur, helical and bevel gears• Internal and external gears• Profile Shift• Backlash and friction• Single and multiple tooth contact (internal

excitations due to tooth meshing)• Dynamically changing gear pair center

distance and backlash (particularly important for floating suns (Fig. 3) and elastically mounted shafts)

The major gear pair enhancements with version 8904 are:• Rack and pinion gearing• Bevel gear primitive• Tooth modification• Flank modification• Shuttling forces• Easy slicing for non-parallel gear wheels

and gear wheels with flank modification• Easy handling of output values and

animation of contact forces

GEAR PAIR PRIMITIVES MAJoR EnHAnCEMEnTSWith bevel gears, a new parameter, the “Rim thickness”, has been added for a more realistic graphical representation (Fig. 4). For all gear pair types, tooth and flank modification has been added. The modifica-tions are primarily used for smoothing the non-linear internal excitations due to the continually changing number of teeth in contact. The following modification types have been added:• Tip (Fig. 5)• Root• Circular• Left and Right Side• Lead Crowning (Fig. 6)• Lead Angular• Bias (Twist)• Input Function ArrayAll modification types can be input for the right and left flanks or for both together.

Fig. 1: Bevel gear with crowning

Fig. 2: Gear box with Gear Pair forces and other resultant forces

Fig. 3: Motion of floating sun within a planetary stage (© IMM, TU Dresden)

Several new functionalities are available with the SIMPACK Gear Pair module in SIMPACK version 8904. not only has the visualization and handling of bevel gears and force arrows been vastly improved but major new functionalities (e.g. tooth profile and flank modifications, easy modelling of non-parallel axes, etc.) have been added.


SoFTWARE | Steven Mulski, SIMPACK AG

GEAR PAIR FoRCE ELEMEnT MAJoR EnHAnCEMEnTSFor simulating gear pairs with non-parallel axes, “slicing” of the gear wheels is necessary [3]. Previous to version 8904, extra gear wheel primitives and force elements had to be used for this purpose. This functionality is now achieved by setting single parameter (i.e. “Number of slices”) within the gear pair force element. The handling of the offset angles for helical gears is now fully automatic. Slicing is also necessary if flank modification is used. Shuttling forces, i.e. the axial displacement of the contact forces, has now been implemented. In the case of helical gears, this will result in an additional tilt moment.The graphical representation of rack gears has been available for a long time now. With

Fig. 4: Bevel gear primitives

Fig. 5: Tip profile modification

Fig. 8: User choice for advanced output values

Fig. 7: Rack and pinion gear

Fig. 6: Crowning, left and right flank

Fig. 9: Animation arrows of normal loads “Indiv. load (fl_n) i,k”

version 8904, the calculation of the rack and pinion forces has also been implemented (Fig. 7). With version 8904, the user can now easily switch on and off, and choose between, various output value types. This enables easier handling and a more efficient use of data storage space. The different types of output are described below.

GEAR PAIR DATA CHECKIn order to check the input parameters and initial conditions of the gear pairs within a model, a user can perform a “Test Call”. This will result in a list being generated for

each gear pair consisting of important input parameters and calculated data. Information such as the theoretical center distance, radial offset, axial offset, transverse contact ratio, overlap ratio, and total contact ratio will now be readily available.

GEAR PAIR ouTPuT VALuESBy way of parameterization, a user can choose for which gear pairs the “Basic Output Values” will be generated. These values include such data as the relative

angles and angular velocities, “total normal contact stiffness” and the “dynamic transmission error”.Similarly, a user can also choose which “Advanced Output Values” are to be saved (Fig. 8). These values are primarily used for analyzing the coupling forces of the gear pairs, either for the sum of all teeth in contact or the individual tooth-pair contacts. In addition the “Advanced Output Values” enable easy animation of the force arrows in the PostProcessor (Fig. 9).After an integration run is complete a user can subsequently choose which output values to generate. Re-running the time integration is not necessary. Only re-performing “measurements” is required.

ConCLuSIonSIMPACK version 8904 represents a major milestone in the development of the gear

pair module. New functionalities such as tooth and flank modification and automatic slicing and force arrow visualization, enable not only easier and faster model generation but also improved accuracy and quicker analyses. Although a significant development step has been achieved, the demanding and varied applications of the gear pair element will continue to result in further, more advanced requirements. The development of the gear pair element does not have one static functionality goal, after which the development can be seen as being completed, but rather a continually

advancing goal to which subsequent further SIMPACK development will ensure that the gear pair element can accompany industrial users long into the future.

REFEREnCES[1] L. Mauer, ‘GearWheels in SIMPACK’, SIMPACK News, July 2004[2] L. Mauer, ‘Modeling and Simulation ofDrive Line Gears’, SIMPACK News, July 2005[3] E. Pfleger, ‘Simulation of the Dynamic Behaviour of Nose Suspension Drives for Rail Vehicles Using SIMPACK-GearWheel’, SIMPACK User Meeting 2006All articles and presentations available at www.SIMPACK.com


Stefan Dietz, SIMPACK AG; Wolfgang Erhardt, SAFE-FEM GmbH | SoFTWARE

Fatigue Analyses with FAT4FEM in SIMPACK's PostProcessor

Fig. 1: Normal stress in the x-direction (axial) of the crank shaft.

Fig. 2: FAT4FEM — GUI for defining material properties to be used in fatigue analysis

Since the release of SIMPACK version 8900 in 2007, the computation and graphical representation of compo-nent stresses has been available in the SIMPACK PostProcessor. With the forthcoming SIMPACK version 8904 this feature is expanded with the ability to perform fatigue analyses. These may be configured and started in the PostProc-essor. After the computation, the results and safety factors may be displayed in the PostProcessor as a contour plot. The fatigue analyses are performed by FAT4FEM which is an easy-to-use and easy-to-learn fatigue tool that is based on the critical plane approach.

THE EXAMPLE MoDELThe FEM model of the crankshaft, which was generated for the stress calculation, was created in NX NASTRAN. The model has approximately 3 million degrees-of-freedom that, in the interest of computational time, must be reduced to a lower number of degrees-of-freedom for a multi-body simu-


SoFTWARE | Stefan Dietz, SIMPACK AG; Wolfgang Erhardt, SAFE-FEM GmbH

lation. The crankshaft in SIMPACK is defined by approx. 50 modal degrees of freedom and does not deviate by more than 1 per-cent from a static solution in NX NASTRAN. The reduced flexible crankshaft is coupled into the multi-body system at the main and large end bearings with the assistance of force elements in order to incorporate the stiffness and attenua-tion properties of the bearings. The connec-tions of the crankshaft with the flywheel and the torsion vibration damper are modeled with the help of joints and/or constraints (i.e. rigidly connected). In the multi-body model, the dynamics of the conrods, pistons and the engine with the engine mounts are all taken into consid-eration. Gas forces are applied as external loads. Consideration of the hydrodynamics in the main bearings and large end bearings, as well as the flexibility of the engine block, is also possible; however, the corresponding influences are not illustrated in this model.

STRESS CALCuLATIonFor modal superposition of stress, unit stress vectors are needed in SIMPACK. These are the stresses of the eigen vectors as well as stresses of so-called "inertia relief modes". The eigen vectors are used to show stress as a result of free vibrations. In this model, only 40 eigen vectors are needed to illustrate vibration behavior in the relevant frequency range with sufficient accuracy. The stresses, caused by coupling forces in the bearings, are represented by "inertia relief modes". An inertia relief mode is the result of a static

Fig. 3: FAT4FEM — GUI for defining failure criteria for fatigue analysis

calculation in the FEM program. Here, the induced forces are compensated by inertia forces from the rigid body motion, i.e. the external load and the inertia forces form an equilibrium of forces and moments. Consis-tent with the mathematical formulation of linear elastic bodies, deformations are not considered in this equilibrium of forces. A

unit load (force and/or moment) is a load in the direction of the coordinate axes that is scaled to unity. Only the relevant unit forces

were considered for the bearings. These are the forces in x- (longitudinal direction) and y- and z-directions (radial) at the axial main bearing and at the interface points to the connecting rods. Forces in the y- and z-directions were considered at the main bearings, which do not transmit axial forces, and the forces and moments in all directions at the joints to the flywheel and the torsion vibration damper. In total, 35 inertia relief modes have been used in this case. This cor-responds to a total of 75 unit stress vectors. The modal coordinates of the eigen vectors considered in the SIMPACK model, as well as the forces in the coupling elements of the flexible crankshaft, are the result of a multi-body simulation that is required for superimposing the unit stress vectors. The stresses of the eigen vectors are scaled with their respective modal coordinates and the stresses of inertia relief modes with the corresponding forces. After the scaling of the unit stress vectors, their superposition and representation in the PostProcessor fol-lows; see Fig. 1. The combination of inertia

relief modes and eigenmodes can accurately represent the stresses due to static and dy-namic loading conditions. When simulating an engine running at constant speed (rpm), forces, modal coordinates, etc. are issued for approx. 720 communication points in time. With this modal superposition, results of static and time-dependent simulations can be efficiently and accurately reproduced in SIMPACK.

LIFESPAn ESTIMATIon WITH FAT4FEMLifespan estimation is carried out with FAT4FEM ("Fatigue for FEM") available from Safe-FEM GmbH, www.safe-fem.de. This software has interfaces to Abaqus/CAE, ANSYS and FEMAP. FAT4FEM is integrated in to SIMPACK and can started from the PostProcessor. FAT4FEM is based on the critical plane approach. The critical plane approach looks for every comparison of two load cases and for the level at a node which yields the most damaging equivalent stress. This technology is suited particularly (but not solely) to problems concerning rotating principle stress directions (which is the case here). The critical plane approach is extremely computation-intensive and very precise in principle. Still, we can only speak of a fatigue estimation. Keep in mind that with today's technology, computer-aided fatigue prediction can always only be a fatigue estimation and cannot give absolute data, e.g. maximum number of years of lifetime, etc. Nevertheless, computer-aided fatigue prediction is a valuable tool for the engineer to quickly compare variants and to identify "weak points" of a mechanical design during the design process. During the development of FAT4FEM, primary focus was given to the user-friendliness of the graphical user interface. Operating the user interface of FAT4FEM is something you can learn in a matter of hours, which makes the program very suitable for non-experts, see Fig. 2 and 3.

LIFESPAn ESTIMATIon —THE MoST IMPoRTAnT ISSuESRoughnesseffect: The crankshaft is made of cast iron (GGG-60). Therefore, roughness effect does not necessarily apply because GGG-60 is a material with internal notches, and surface is smoothed at critical places so no surface effect is taken into account. Temperature effect: At oil temperatures up to 140 °C, the temperature effect must not be considered for GGG-60. The follow-ing factors are of importance for the correct assessment of the crankshaft with regard to lifespan:

“FAT4FEM is an easy-to-use and easy-to-learn fatigue tool

that is based on the critical plane approach.”


Stefan Dietz, SIMPACK AG; Wolfgang Erhardt, SAFE-FEM GmbH | SoFTWARE

Fig. 4: FAT4FEM results in the SIMPACK PostProcessor — Equivalent stress amplitude (red corresponds to large stresses)

Fig. 5: FAT4FEM results in the SIMPACK PostProcessor — Safety factor (green indicates critical spots)

1. Correctly determined stresses via MBS and FEM It is assumed that the stresses calculated with SIMPACK and NX Nastran are suf-ficiently accurate. 2. Correctly defined S/n curve The material properties for GGG-60 were taken from the publicly accessible website of the TU Darmstadt. These properties are based on data pools of Ch. Boller and T. Seeger and are publicly accessible on the internet: www.wm.bauing.tu-darmstadt.de

3. Suitable equivalent stress hypothesisGGG-60 has a breaking strain A5 of 7 %. This means that the equivalent stress hypothesis according to Van Mises can be applied.

4. Mean stress correction Mean stress sensitivity is approx. 0.3 %, which indicates a ductile material.

5. Size effect This takes the size of the component into account. The size effect was not considered in the present study. There-fore, we get more conserva-tive values for safety. With comparative calculations, as is also customary for crankshafts, it is, of course, necessary to always use a specific method, either with or without notch sensitivity.

ILLuSTRATIon oF THE RESuLTSAfter the lifespan estimation, the results are automatically uploaded to the SIMPACK PostProcessor by FAT4FEM. Fig. 4 shows the mean stresses and Fig. 5 the factor of safety for the examined rpm.

SuMMARYThe SIMPACK FAT4FEM functionality de-scribed here is available as a licensed module in addition to the SIMPACK Stress module with version 8904. The seamless integration of MBS, FEM and Fatigue comes with clear time and cost advantages for the experi-

enced user. The process described here is available for automated calculations in batch operations. The fatigue estimation only causes minimal extra time

effort in the overall process chain. Defining the FAT4FEM project takes just a few min-utes. Regarding CPU times, the calculation of FAT4FEM for the areas shown in Fig. 4 and 5 and for 360 load steps is about one hour. If only every third load step is activated, one gets nearly the same results for a fatigue estimation, but evaluation time amounts to only about six minutes. This means that lifespan estimation for dynamically stressed components, even for a great number of operating modes, is possible in an extremely fast and simple manner.

“Operating the user interface of FAT4FEM is

something one can learn in a couple of hours.”


SoFTWARE | Wolfgang Trautenberg, SIMPACK AG

200 new Features and Improvements introduced with SIMPACK Version 8904

GEARWHEEL SIMuLATIonS GoInG 3DThe fast and accurate SIMPACK Gear Pair Force Element was greatly enhanced in numerous ways. Functionalities such as automatic slicing and effects like force shuttling can now be used to include 3D effects into the gearwheel simulation. Additionally, different methods were added for taking into account profile and flank modifications of the gears. The list of supported geometry pairings was expanded to facilitate the simulation of rack and pinion pairs. To enable easier postprocessing and greater insight, the generated outputs were completely reworked, including drastically improved 3D visualizations of the forces acting in a gear pair. Another important new capability is the data check facility that lets the user see all important gear pairing data in a central location. For more details, see the separate article on gearwheel simulation in this edition of the SIMPACK news (see page 25).

BRAnD nEW BELT DRIVE MoDuLEThe list of the major drivetrain coupling elements of gearwheels and chains was complemented by adding the Belt Drive module (Fig. 1). Starting with SIMPACK 8904, detailed and fast belt drive simulations can be performed via a modal belt description. A belt drive is defined by the pulleys, the optional tensioner system and the belt force element that connect the former elements to a belt drive. The belt itself is defined via its geometry and material properties. From these inputs, SIMPACK automatically computes a modal description of the belt. Different friction models are available.

CHAIn DRIVE EnHAnCEMEnTSThe Chain Drive module, the third of the three drivetrain coupling modules, was enhanced with the capability to include an additional centrifugal ring when using smart chains (Fig. 2). Now, in addition to the contact of the chain plate with the chain

wheel shoulder, the chain plate can have contact with these additional rings.

DYnAMIC BuSHInG PARAMETER FITTInG AnD nEW oPERATInG MoDEA new operating mode was added to the dynamic bushing force element to enable a hysteresis computation that provides for a memory effect typically seen in bushing measurements. Finding the right set of parameters for the dynamic bushing was made much simpler and quicker by introducing a parameter fitting utility. This utility allows the user to tune the force element parameters interactively via sliders while a plot shows the comparison of measured and computed data.

TIRE nEWSCDTire (Fig. 3), a tire model for comfort and durability applications, owned and now developed by the Fraunhofer Institute for Durability and System Reliability LBF, is now available as one of the tire modules shipped with SIMPACK. CDTire can automatically adapt to changing road situations by switching between different internal representations. These include detailed models for large deformations and allow for tire rim contact. In addition, the CDTire implementation in SIMPACK can take full advantage of multi core and multi CPU computers by utilizing one CPU core per tire if requested by the user.The DELFT Tire module integrated in SIMPACK is now version 6.1.2. In addition

The SIMPACK development team is strongly focused on creating SIMPACK 9000, the next major SIMPACK release due in 2011. nevertheless, a lot of attention and development effort was put into SIMPACK 8904, the latest SIMPACK version, released in September 2010. Almost 200 new features and improvements have been added to SIMPACK 8904.These include brand new SIMPACK modules such as the belt drive module or the Bio-Mechanics module, as well as major new features for existing modules, such as the newly added Gear Pair Force Element capabilities. The most important of these new features and improvements are summarized in the following sections. For in-depth information on all changes, please refer to the release notes of SIMPACK 8904 which are available on the SIMPACK Download Server.

Fig. 1: Belt drive

Fig. 2: Centrifugal contact ring for chains


Wolfgang Trautenberg, SIMPACK AG | SoFTWARE

to many improvements, support for a new road type was added with this DELFT Tire version. The curved regular grid road (or CRG road), which describes the road via a grid of measured road heights lateral to a spine, can now be used with this tire model.

nEW RAIL nEWSThe New Rail module was expanded to enable simulations of untrue (out-of-round) wheels. In addition, the simulation of large yaw angles was reworked and significantly improved leading to greater accuracy and robustness.

MoRE SIMuLATIon PoWER FoR WInD EnERGYIn addition to many new features offered for drivetrain simulations which are also applicable for wind turbines, more wind specific functionality was added to this SIMPACK release. This includes a SIMPACK native integration of NREL’s aerodynamic library AeroDyn for simulating the wind forces acting on rotorblades of horizontal axis wind turbines.For simulating large non-linear deformations in rotorblades an option was added to SIMPACK’s rotorblade generator to automatically split the rotorblade into different flexible bodies.

BIoMECHAnICS PoWERED BY BIoMoTIon SoLuTIonSFor analyzing biomechanic systems such as a human being interacting with a power drill, a seat restraint system or a motorcycle, a completely new module was added to SIMPACK (Fig. 4). This module consists of special force and control elements for modelling muscles, tendons and their respective controllers, spinal discs and so called wobbling masses. These elements are used in full or partial models of human beings with an adaptable level of detail. The models can be generated by a model generator that works off a database of typical human beings. The Biomechanics solution is contributed by Biomotion Solutions (www.biomotion-solutions.com), a company that specializes in measuring and simulating the biomechanics of human beings.

TIME EXCITATIonS SPED uPThe time excitation types 15 - for importing measured data on either position, velocity or acceleration level — and 20 — for importing a speed profile given over position — were significantly sped up and reworked to offer more options on dealing with the measured data. In addition, the time excitation type 17 — for importing velocity dependent periodic signals — was expanded to enable the definition of the Fourier Series via Re and Im as well as magnitude and phase.

3D PRIMITIVESNew primitives were added for displaying belt pulleys and the belt path. The gear wheel primitive was reworked to provide for a much more realistic display of bevel gears, and cylinder primitives can now have a hole.

FLEXIBLE BoDIES AnD FATIGuEThe flexible body interface FEMBS now supports cdb files of ANSYS 12 models for the graphical representation of flexible bodies. Also, the solver performance for flexible bodies with a large number of modes has been greatly improved. A solution for performing fatigue analysis was added directly into SIMPACK. This solution is based on the FAT4FEM technology (Fig. 5) developed by SafeFEM GmbH (www.safe-fem.com). Please see the separate article about this module in this SIMPACK news edition (see page 27).

STRESS AnD STRAIn CoMPuTATIonSIMPACK 8904 now offers the possibility to compute and display strains for selected flex body nodes. The strains can be plotted as

Fig. 3: CDTire

Fig. 5: FAT4FEM

standard 2D curves and can be exported in various formats.The stress display capability of SIMPACK has been greatly sped up for ABAQUS and improved so that much larger FE-result files can now be processed.

CoDE EXPoRTThe Code Export module was reworked to provide a simpler and cleaner interface by supplying the runtime library as convenient DLL. Also, the Code Export licensing is now completely based on OLicense.

PoSTPRoCESSoRA new spectrum filter was added to the PostProcessor, providing many interesting features such as overlapping averaging windows. The integration and differentiation filters were both redesigned and extended and can now also handle signals with non-equidistant x-axis values. Displaying and scaling 3D force arrows was improved by adding a smarter scaling algorithm and more scaling options .New commands were added to the scripting engine. The script execution now only requires a license if PostProcessing functionality is triggered by the script. In addition, the size of curve markers is now user definable.

MATLAB® AnD AMESIM InTERFACESThe MATLAB® interfaces SIMAT and MatSIM now support MATLAB version 2009b and 2010a.The AMEsim interface SIMAS now supports AMEsim version 8 and 9.

PLATFoRM nEWSWindows 7 is now a fully supported and certified SIMPACK platform. Official support for RedHat Enterprise 5.3 was also added with this SIMPACK version.On 64 bit operating systems, SIMPACK 8904 supports up to 4 GB of memory even with 32bit binaries. This is useful for simulation tasks that require huge amounts of memory, e.g. due to large result files that should be generated. It is therefore highly recommended to always install SIMPACK on machines with a 64bit operating system.Fig. 4: Biomechanics

SIMPACK NewsSIMPACK News

InSIDE THIS ISSuE

ConTACTS

GERMAnY wOrlDwiDe HeaDQuarterSSIMPACK AG Friedrichshafener Strasse 1 82205 GilchingGermanyPhone: +49 (0)8105 77266-0 Fax: +49 (0)8105 77266-11 [email protected]

uSASIMPACK US Inc.46701 Commerce Center DrivePlymouth Michigan 48170, USAPhone: +1 734 233-3080Fax: +1 734 207-3165 Mobile: +1 251 923 [email protected]

InDIAProSIM R&D Center#21/B, 9th Main, Shankarnagar Mahalakshmipuram Bangalore – 560096, Karnataka, IndiaPhone: +91 (0)80 233-79686 / 23020Fax: +91 (0)80 233-23304 / 235-78292 [email protected] / [email protected]

GREAT BRITAInSIMPACK UK Ltd. The Whittle EstateCambridge RoadWhetstoneLeicester LE8 6LHUKPhone: +44 (0)116 27513-13Fax: +44 (0)116 27513-33Mobile: +44 (0)7767 416 [email protected]

FRAnCESIMPACK France S.A.S.Immeuble "Le President", 4eme étage40, Avenue Georges Pompidou69003 Lyon, FrancePhone: +33 (0)437 5619-71 Mobile: +33 (0)673 881 [email protected]

BRAzILVirtualCAE Serviços de Sistemas Ltda.Rua Tiradentes, 160 – Sala 22São Caetano do Sul – São PauloCEP: 09541-220, BrasilPhone: +55 (0)11 4229-1349Mobile: +55 (0)11 [email protected] www.virtualcae.com.br

JAPAnSIMPACK Japan K.K.5F Okubo Bldg.2-4-12 YotsuyaShinjuku-ku Tokyo 160-0004 JapanPhone: +81 (0)3 5360-6631Fax: +81 (0)3 [email protected]

CHInAGlobal Engineering Technology Group (GET Group)Tri-Tower, C Building, Room 1802, 1804Zhongguancun East Road 66 Haidian District Beijing 100190, P.R. ChinaPhone: +86 (0)10 626 [email protected]

SIMPACK News

impreSSum:

PUBLICATION PERIOD:

1996 – 2010

EDITORIAL, DESIGN & LAYOUT:

Steven Mulski, Nicole Blum

© SIMPACK AG

SIMPACK AGFriedrichshafener Strasse 182205 Gilching, GermanyPhone: +49 (0)8105 77266-0 Fax: +49 (0)8105 77266-11 [email protected] www.SIMPACK.com

CIRCULATION: 5.300

All previous SIMPACK News articles can be found on www.SimpaCK.com: downloads, newsletters/articles

If you would like to sign up for free delivery of the SIMPACK News please visit: www.simpack.com: downloads, newsletters/articles, subscription

1. Simulation of the Dynamic Behavior of Aircraft Landing Gear Systems

2. Electronic Stability Program (ESP) for Trucks on the Daimler Driving Simulator

3. Simulating Tank Vehicles with Sloshing Liquid Load

4. Development of a SIMPACK User Routine for Dynamic Light Rail Vehicle Gauging Simulations

5. Determining Reliable Load Assumptions in Wind Turbines using SIMPACK

6. Simulation of Drivetrains on Wind Turbines within the Framework of Certification — with SIMPACK

7. SIMPACK Academy Events in 2010

8. Coupling of MBS and CFD: an Oscillating Aeroelastic Wing Model

9. Gear Pair Enhancements with SIMPACK Version 8904

10. Fatigue Analyses with FAT4FEM in SIMPACK's PostProcessor

11. 200 New Features and Improvements introduced with SIMPACK Version 8904

simpack ag, friedrichshafener strasse 1, 82205 gilching ...€¦ · simpack ag, friedrichshafener...

Documents