Conference programme

Back to the programme printer.gif Print



Thursday, 19 November 2015
14:30 - 16:00 Innovation in wind turbine drive trains
Turbine technology  
Onshore      Offshore    


Room: Montmartre

This session will share new concepts to reduce failures/improve behavior cost-effectively:

  • New configuration of bearingless permanent magnet machine with multiple modules for direct drive to reduce bearing failure
  • Coupled magnetoelastic model of a multipole permanent magnet generator in no-load conditions to reduce weight
  • Potential of MgB2 superconductors on direct drive generators
  • Alternative Wind Turbine Drive Train with Power Split and High-speed Generators 

Learning objectives

  • Delegates will be able to list problems associated to generator systems in direct-drive WTG and identify a new bearingless concept
  • Delegates will be able to list design problems inherent to scaleability and identify a model to capture dynamic phenomena and derive design criteria
  • Delegates will be able to explain cost considerations limiting superconducting direct drive technology and identify potential optimization
Lead Session Chair:
Anand Natarajan, DTU, Denmark
Henk-Jan Kooijman, GE Power & Water, Germany
Jaume Betran Alstom Wind, Spain
Co-authors:
JAUME BETRAN (1) F ROGER BERGUA (1) JAUME MIRO (1)
(1) ALSTOM WIND, BARCELONA, Spain

Share this presentation on:

Printer friendly version: printer.gif Print

Presenter's biography

Biographies are supplied directly by presenters at EWEA 2015 and are published here unedited

Mr Betran has been working in the wind industry for almost 15 years in several different disciplines. He studied Industrial Electric Engineering at the Polytechnic University of Catalonia. He worked in gearbox design and failure investigations, holistic wind turbine modelling and engineering studies for the main OEM companies and research on aerodynamic modelling. He is currently doing research on coupled computation applied to aeroelasticity and mangetoelasticity as well as optimizationa pplied to tower dynamics and holistic models.

Abstract

Capturing Resonance of a Multipole Permanent Magnet Generator with a Magneto-Mechanical Model

Introduction

During the last few decades engineers have been designing wind turbines of increasing size seeking lower values of cost of energy. Coupling finite element (FE) models with in-house tools is a powerful means to customize models, increase computation capabilities and enhance designs. These techniques are especially meant to assess scalability problems for the increasingly slender structures incurring in lower modes and bigger deflections.

Approach

The present work explains a computation method to couple a surrogate model of air-gap magnetic forces to a FE based multibody model of a permanent magnet generator. The resulting coupled simulations are aimed to capture the magnetoelastic behavior of an off-shore wind turbine multipole permanent magnet generator in no-load start-ups.
The model is the result of coupling simplified representations of both structural parts and magnetic loads in a flexible multibody simulation environment so that the magnetoelastic behavior of the set is naturally captured. Both rotor and stator structures are condensed in superelements by substructuring techniques. Bearings are represented with a stiffness matrix and rotation constraints. Finally, magnetic forces are included by means of a surrogate model from FE magnetic simulations.


Main body of abstract

The magnetic interaction between rotor magnets and stator teeth is of a static nature for no-load conditions, that is, no dynamic phenomena exist for magnetic forces due to changes of the air-gap or magnets-teeth relative position. To build a coupled magnetoelastic model, magnetic forces are computed beforehand using the commercial 2D finite element code for electromagnetic simulation FLUX, including non-linear magnetic loads among stator teeth and rotor magnets. These data are included in the magnetic forces model as a look up table. A custom surrogate model of the magnetic forces was created including an algorithm that identifies the configuration of magnets and teeth and applies magnetic forces in 6 DOFs at every magnet and every tooth depending on their relative azimuthal position and local air-gap. The model was built in FORTRAN and coupled to the SAMCEF model as a USER element.
The structural models of both rotor and stator are modelled by means of the commercial FE based flexible multibody tool SAMCEF. Superelements of the said structures are obtained by means of component mode synthesis and modal damping is included and tuned by matching their frequency response functions with experimental ones.
Finally, a fully coupled model is created by putting together superelement representations of the structures and the surrogate magnetic loads model. The bearing is modeled with a set of constraints and Lagrange multipliers to account for rotation and a coupled stiffness matrix to account for its elastic behavior in the 6 DOFs. The coupled model is then solved in a time discretized integration scheme, the exchange of coordinates and forces between submodels is done with non-converged magnitudes and therefore a simultaneous convergence and dynamic equilibrium is guaranteed at every time step.
The results obtained at the moment are promising as the generator behavior is validated against unbalanced magnetic pull, electric field rotation speed and cogging frequency are verified, and the assembly dynamic behavior is validated against experimental frequency response functions.


Conclusion

A model of multipole permanent magnet generator has been set up with a view to capture relevant dynamic phenomena and derive design criteria to reduce generator weight while satisfying safety margins. The model performance matched Alstom objectives as the behavior in normal operation and extreme conditions were validated with measurements.


Learning objectives
A method to model the magnetoelastic performance of a permanent magnet generator is presented. The model with a small number of DOFs captures the dynamic behavior by means of FE based fully coupled magnetoelastic simulations.