1. 20-21 October 2004, Page 1 SAMCEF for Wind Turbines SL/04/SAM/MKG_ppt/37an_a

2. 20-21 October 2004, Page 2 Scope of the presentation  What is the problem of Wind Turbines?  Why is it difficult to analyze numerically Wind Turbines?  Methodology suited to compute dynamic behavior of vibrating mechanical systems  SAMCEF for Wind Turbines  Presentation of examples  Conclusions and future investigations&developments  Questions/Answers

3. 20-21 October 2004, Page 3  In order to increase the productivity, it is more and more necessary to increase the velocity of Wind Turbines moving parts  If maximum velocities are higher and higher, accelerations are increasing significantly  Accelerations prescribed by controllers are generating huge inertia forces due to the mass of moving parts  If the mass of moving parts is reduced due to the removal of material, there can be a loss of stiffness and thus of precision  Inertia effects are thus major excitations for vibrations of Wind Turbines in working conditions  Vibrations cannot be cut in pieces What is the problem of Wind Turbines ?

4. 20-21 October 2004, Page 4  Need of a coherent global model describing the whole dynamic flexible behavior of the Wind Turbine (mechanisms + structures + controllers)  Sensitive components are the ones transmitting the vibrations (guideways, linear motors, ball screws…)  These components have to be modeled using advanced numerical computation techniques able to manage the interactions between moving flexible bodies  NL Finite Element coupled with kinematical constraints and large relative motions contact techniques are appropriate to achieve this goal  SAMTECH has been working on this innovative approach since 15 years and brought recently this new powerful and integrated technology to the market, starting with the sector of Wind Turbines In conclusion: What is the problem of Wind Turbines ?

5. 20-21 October 2004, Page 5 Why is it difficult to analyse numerically Wind Turbines ? Examples of possible sources of “Non-Linearities” involved in FEM models of Wind Turbines: Involved “Non-Linearities” :  Rotor: Large deformations & rotations (in case of fully FEM), Anisotropy  Shafts: Large rotations (no problem)  Over-load clutch: very pronounced discontinuities in stiffness  Elastic Couplings: Generally progressive stiffness  Bearings and Gears: Frictional contact & impacts  the “worst of all”  Generator model: Discontinuities due to conection or disconection & instability if max. load/“Kippmoment” passed Implicit uncondinionally stable time integration:  Control of time integration error during discontinuities (impacts, stick-slip...)

6. 20-21 October 2004, Page 6 Augmented Lagrangian form of the equations of motion HHT-Form of discretised equilibrium equation Equations are solved iteratively each time integration step: “HHT-Form” Mixed system of equations in FEM-DOF and Lagrange Multipliers: M: Mass matrix B: Gradient of the constraint matrix  : Kinematical constraints Classical non-linear Finite Element equations : Lagrange multiplier p: Penalty factor k: Scaling factor Coupling mechanism and structural analysis Mathematical background Methodology suited to compute dynamic behavior of vibrating mechanical systems

7. 20-21 October 2004, Page 7 Dynamic forces can only be evaluated, if the complete wind turbine is modeled! Mechanism type Components: Gears Bearings Couplings Clutches Generator & Control Bushings Etc. Why is it difficult to analyse numerically Powertrains ? There are structural FEM-components like: Rotor: 3D/FEM, Super-Element, or concentrated mass & inertias Elastic shafts: Beams (stiffness & damping) Planetarycarrier/Planetenträger: 3D/FEM, Super- Element

8. 20-21 October 2004, Page 8 Mathematically bad conditioned: due to very different inertias & stiffness Why is it difficult to analyse numerically Powertrains ? Example: Ratio of Rotor  Gearbox Inertias/stiffness > 1E8

9. 20-21 October 2004, Page 9 Axial “Stick-Slip” effects in bearings, shafts, housings: Coulomb friction law Contact law Why is it difficult to analyse numerically Powertrains ?

10. 20-21 October 2004, Page 10 Integrated Multi-Disciplinary Solution in SAMCEF for Machine Tools General integrated environment giving access to different levels of model fidelity and different analysis types from the same CAD based model Methodology to compute dynamic behavior of vibrating mechanical systems

11. 20-21 October 2004, Page 11 Global solution for the modeling of Wind Turbines Very general computation scheme Methodology to compute dynamic behavior of vibrating mechanical systems

12. 20-21 October 2004, Page 12  Finite Element Analysis using SAMCEF SAMCEF Mecano: Mechanisms&Non-Linear Dynamic Structure/Contact FE Analysis SAMCEF Asef: Linear static FE analysis SAMCEF Dynam: Modal FE analysis + Super-element creation/restitution Interface between mechanical analysis (mechanisms&structures) and controller design: Import of MATLAB Simulink controllers inside SAMCEF Mecano Export of state space (A,B,C,D matrices) from SAMCEF Mecano to MATLAB Simulink SAMCEF Field: User Friendly CAD Based Modeling and Post- Processing Environment  Optimization of parameterized mechatronic systems using BOSS quattro Components of SAMCEF for Wind Turbines Methodology to compute dynamic behavior of vibrating mechanical systems

13. 20-21 October 2004, Page 13 NL Finite Elements (linear motor…) and S.E. Flexible kinematical joints (Guideways, flexible ball screw…) Digital Control (Sensors and Actuators) SIMULINK SIMULINK … Rigid kinematical joints SAMCEF Mecano: Integrated Non-Linear Mechanical Analysis A adapter Methodology to compute dynamic behavior of vibrating mechanical systems

14. 20-21 October 2004, Page 14  Description of all the parts and the connection devices of the mechanical system  Different Model Fidelity levels (rigid body, detailed meshed structure or Super-Element)  Different types of connection devices (rigid or flexible kinematical joints, contact/friction conditions)  “Sensors” and “actuators” to be connected to the digital control boxes  Output of linearized mechanical models for control design SAMCEF Mecano: Integrated Non-Linear Mechanical Analysis Methodology to compute dynamic behavior of vibrating mechanical systems

15. 20-21 October 2004, Page 15  Linear Static Analysis (SAMCEF Asef)  Modal Analysis (SAMCEF Dynam)  Super-Element creation and restitution (SAMCEF Dynam)  Integrated environment with MBS and NL FEA SAMCEF Asef & SAMCEF Dynam: Linear Analysis Modal Analysis (SAMCEF Dynam) Methodology to compute dynamic behavior of vibrating mechanical systems

16. 20-21 October 2004, Page 16 SAMCEF Mecano Control Box Next Sample Time Command Vector Updated State Variables Current Time Sensor Vector State Variable  Sensor Variables:  Positions, Displacements, Velocities, Accelerations and/or Reaction Forces  Command Vector:  Forces, Displacements, Velocities, Accelerations and/or functions Inputs and outputs of Controllers Communication between SAMCEF and embedded controllers SAMCEF-Digital Control Interface

17. 20-21 October 2004, Page 17  Controllers can be imported from a Controller Design Software (for example MATLAB Simulink using Real Time Workshop (RTW) to generate C subroutine)  They can also be fully written by the user (open to in-house controllers)  Linked with SAMCEF Mecano SAMCEF-Digital Control Interface

18. 20-21 October 2004, Page 18  Positions, displacements, velocities, accelerations as well as reaction forces are provided independently by SAMCEF Mecano to the controller  Activation and deactivation times  Several controllers can be used at the same time  The same controller can be used several times  Different schematic boxes (sensors, controllers and actuators) can be connected between them  Prescribed values and gains can be defined through the SAMCEF data set SAMCEF-Digital Control Interface

19. 20-21 October 2004, Page 19  The computation time step is bound to the time discretization of the Controller tComp <= t Controller  Automatic time step choice between instants of “Rendez-vous” Time step control SAMCEF-Digital Control Interface

20. 20-21 October 2004, Page 20 SAMCEF Field Integrated Modeling Environment A adapter

21. 20-21 October 2004, Page 21  Modeling defined directly on the CAD geometry  Easy modifications of the modeling fidelity for bodies Rigid Body Meshed body with any type of material, including composites Super-Elements generated by SAMCEF or imported from another FE package  Easy modifications of the modeling fidelity for connections Ideal kinematical joints Rigid-flexible contact conditions Flexible-flexible contact conditions  Easy switch from an analysis type to another one (SAMCEF Mecano, SAMCEF Dynam, SAMCEF Asef…)  Import of controllers from MATLAB Simulink SAMCEF Field Integrated Modeling Environment

22. 20-21 October 2004, Page 22 Five steps to manage the global to the detailed analyses SAMCEF for Wind Turbines

23. 20-21 October 2004, Page 23 SAMCEF for Wind Turbines

24. 20-21 October 2004, Page 24 CAD creation Geometry creation Geometry import

25. 20-21 October 2004, Page 25 Component creation Tower Generator Blades Gears

26. 20-21 October 2004, Page 26 Super-Element model reduction Component creation and use Placement and use Super-Element creation Parameterized components

27. 20-21 October 2004, Page 27 Anisotropic composite material model of rotor blades Material definition No. Materials: 2 No. “Plies”: 60 No. “Laminates”: 4

28. 20-21 October 2004, Page 28 Linear and non-linear data Data definition

29. 20-21 October 2004, Page 29 Linear and non-linear data definition on the geometry Data creation Definition and assembly of wind turbine components: blades, rotor-shaft, bearings, gearbox, couplings, clutch, generator, tower, etc.

30. 20-21 October 2004, Page 30 Wind Turbines Mechanical Elements Transmission models  Gears  Generators Common features  Definition on geometry  Multiple models for each physical component  Only additional parameters between different models of the same transmission

31. 20-21 October 2004, Page 31 Selection of rigid or flexible behavior Meshing: Rigid – Flexible

32. 20-21 October 2004, Page 32 Integration of a SAMCEF model into a functional simulation tool used to design controllers Controller design  Efficient choice of the right controller  Fine tuning of the initial values for the gains Controller design Model linearization Adapted controller

33. 20-21 October 2004, Page 33 Internal or external controller Integration of various Digital Controllers into SAMCEF Non-Linear Simulations in order to get a mechatronic FE model combining mechanisms, structures and controllers in one single model Controller interface with model

34. 20-21 October 2004, Page 34  Static stresses evaluation  Vibration Modes calculation  Super-Element reduction  Time response (static, kinematical or dynamic) Various analyses using same model description Analyses

35. 20-21 October 2004, Page 35 Non-linear structure analysis Non-Linear Results

36. 20-21 October 2004, Page 36 Pole placement controllerLQR controllerMIMO controllerMISO controllerHAC/LAC controllerPD controller Vibration Controller Types

37. 20-21 October 2004, Page 37 Synchronised time delay Tuning individual controllers Concurrent Tuning of all controllers Performance measures: Circle test

38. 20-21 October 2004, Page 38 BOSS quattro Task Management and Optimization

39. 20-21 October 2004, Page 39 Optimization iteration

40. 20-21 October 2004, Page 40 Optimization loop

41. 20-21 October 2004, Page 41 General SAMTECH Product Portfolio

42. 20-21 October 2004, Page 42 Product/Service Approach of SAMTECH Group

43. 20-21 October 2004, Page 43 Future Developments & Perspectives  Introduction of harmonic response (SAMCEF Repdyn) in SAMCEF Field  Introduction of thermal transient analysis (SAMCEF Thermal) in SAMCEF Field  Extension of the Mechatronic approach by Finite Element Analysis to other industrial sectors (robotics, textile machines, automotive, aeronautics, space, defense…)

44. 20-21 October 2004, Page 44 Conclusions  Wind Turbine manufacturers currently develop high speed machines were inertial effects are becoming critical  A Rigid Multi-Body model is clearly not sufficient  Super-Elements are not always sufficient (inefficient management of contact conditions between moving flexible bodies)  SAMCEF for Wind Turbines proposes an implicit NL Dynamic solution using NL Finite Element/Contact techniques  SAMCEF for Wind Turbines is also a global and integrated CAD based solution covering all the Machine Tools components (linear motor, guideways, ball screw, gears, rack-and-pinions…), the model fidelity levels and analysis types (linear or not) requested by this industrial sector