danhong presentation

Upload: harshan-arumugam

Post on 05-Apr-2018

220 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/2/2019 Danhong Presentation

    1/43

    Finite Element Analysis of Electric Machines

    ------The Solver and Its Application

    Danhong Zhong

    Department of Electrical Engineering

    The Pennsylvania State University

  • 8/2/2019 Danhong Presentation

    2/43

    Motivation

    Steady-state finite element solver

    Low rotor loss permanent magnetmachine design

    Summary and future work

    Outline

  • 8/2/2019 Danhong Presentation

    3/43

    Electric Machines in New Technologies

    Flywheel for frequency regulation for

    renewable and distributed generation

    (Credit: Beacon Power)

  • 8/2/2019 Danhong Presentation

    4/43

    How do we design a machine that

    suits our need?

  • 8/2/2019 Danhong Presentation

    5/43

    Flywheel energy storage system

    Motor/Generator test setup for flywheel system

  • 8/2/2019 Danhong Presentation

    6/43

    Design Objectives

    Flywheel motor/generator forsmall satellite application

    Power level of 100W

    Ultrahigh speed operation(150-300krpm)High frequency

    electromagnetics

    Continuously charging anddischarging

    Thermal Constraints important

    Electrical losses

    Rotorlosses

  • 8/2/2019 Danhong Presentation

    7/43

    Design Methods

    Rules-of-thumb, empirical tables, design equations, a designers

    intuition are all valuable

    Classic Way----prototyping

    Large-scale numerical simulation

  • 8/2/2019 Danhong Presentation

    8/43

    Magnetic excitation is supplied by high-energy permanent magnet

    No power loss is associated

    with machine field excitation

    High power/weight ratio

    ----Popular choice for

    high-speed motor/generator

    applications

    Synchronous Permanent Magnet (PM) Machines

  • 8/2/2019 Danhong Presentation

    9/43

    Rotor losses

    Rotor exposed to high electrical frequency harmonics

    Heat generated

    Rotor spins in vacuum, supported by magnetic bearings

    Only method of heat transfer is through blackbody

    radiation, which is a relatively poor heat transfer mechanism

  • 8/2/2019 Danhong Presentation

    10/43

    Temperature dependence of Neodymium Iron Boron

    The properties of manymaterials in the flywheel rotor will

    degrade with increasing

    temperature

    E.g., the intrinsic coercivity of

    Neodymium-Iron-Boron decreases

    significantly with temperature,

    creating a risk of demagnetization

    It is therefore important to

    minimize rotor losses in this

    application

    Change of intrinsic coercivity of

    Neodymium Iron Boron

    with respect to temperature

    (courtesy Dexter Magnetics)

  • 8/2/2019 Danhong Presentation

    11/43

    Solver overview

    Programmed in MATLAB environment 2-D finite element analysis

    Software capable of:

    model building

    mesh generation

    steady-state solution solving (time stepping)

    rotor losses calculation

    Uses GMRES method to speed up steady-state solution

    Allows study of rotor losses when insulating barriers existin the 2-D plane of the rotor (e.g., segmenting permanentmagnets)

  • 8/2/2019 Danhong Presentation

    12/43

    Road Map

    1) Maxwells equation --The governing PartialDifferential Equation(PDE) of the problem

    2) PDE - Finite Element Equations

    3) Main techniques used in achieving the steady state

    solution.

  • 8/2/2019 Danhong Presentation

    13/43

    Mathematical model

    Maxwells equations:

    t

    BE

    JH

    B

    0

    Constitutive Laws:

    MHB

    BvEJ

    00

    )(

    Governing Partial Differential Equation (PDE)

    (1)

  • 8/2/2019 Danhong Presentation

    14/43

    Finite Element Method

    The object is divided into FiniteElement mesh

    A simple relationship is used torepresent the variables anywhere in

    the element by variables on thenodes of the element.

    Within the element, approximatefunctions in terms of nodal valuesare then derived from the PDE

    Mesh (16761 points 33232 elements)

  • 8/2/2019 Danhong Presentation

    15/43

    element

    The method of weighted residuals

    Define a residual function:

    (1)

    The solution of the PDE should satisfy that, for a given weight function w,

    (2)

    (3)

  • 8/2/2019 Danhong Presentation

    16/43

    Finite Element Formulation

    The finite element formulation is given as:

    Where

    D, K :global matrix decided by material properties and element shape

    M, M : represents magnetization in ferromagnetic materials orpermanent magnets, can be a nonlinear function of x

    I s : forced current flowing in stator windings

    (4)

    sysxs

    yrxr

    s

    r

    ssrs

    srrr

    s

    rrr

    IMM

    MM

    a

    a

    KK

    KK

    s

    a

    aS 0

    00

    0

    3

    ~00

    000

    00

    0~

    3

    10

    (5)

  • 8/2/2019 Danhong Presentation

    17/43

    Steady-State Analysis

    Under steady-state assumptions, after one period T we wish to achievethe same primary variables, i.e.

    If we define the nonlinear state transition function as , this becomes:)(

    (6)

  • 8/2/2019 Danhong Presentation

    18/43

    Steady-State Solution Techniques

    The state transition function in (6) is determined by (5)

    To achieve the steady-state solutions, we used the following numericaltechniques:

    o Backward Euler Integration

    o

    Shooting-Newton Algorithmo Matrix-free GMRES

    (5)(6)

  • 8/2/2019 Danhong Presentation

    19/43

    Shooting Newton Method

    ModifyNO

    Time integration of the differential equation

    YES

    Initial Value Final Value

    Error < Tolerance?

    Target Value (=Initial Value)End Iteration

    Calculate

    Modification

    The steady-state solution in the time domain is obtained byusing a shooting Newton method:

    Applying the Newton-Raphson method, we get:

    WhereJis the Jacobian of the nonlinear state transition function

    (8)

    In our solver we define (7)

  • 8/2/2019 Danhong Presentation

    20/43

    Matrix-free GMRES Method

    Generalized minimum residual method (GMRES) :

    A Krylov-subspace method

    When solving Ax=b, no direct access of matrix A is used, A onlyneed to be accessible via a subroutine that returns y=Az

    The Jacobin 1

    1

    1

    )0(

    )(),0,)0((

    j

    jj

    x

    TxTxJ can not be explicitly written

    (8)

    A x b

  • 8/2/2019 Danhong Presentation

    21/43

    Efficiency Improvement

    By using GMRES, the computation time of the shooting-Newtonmethod is dramatically reduced.

  • 8/2/2019 Danhong Presentation

    22/43

    Flux Density Distribution

  • 8/2/2019 Danhong Presentation

    23/43

    Rotor Loss Distribution

  • 8/2/2019 Danhong Presentation

    24/43

    Harmonic analysis

    0 20 40 60 80 1000

    0.05

    0.1

    0.15

    0.2

    0.25

    harmonics

    rotorlosses(w)

    After we achieve the steady-state solution, we perform a Discrete FourierTransform of x(t) and calculate the eddy current rotor losses for eachharmonic.

  • 8/2/2019 Danhong Presentation

    25/43

    Rotor Loss Design Study

    Design parameters are changed to study the effecton rotor losses

    The stator current peak is adjusted to maintainconstant steady-state mechanical power

  • 8/2/2019 Danhong Presentation

    26/43

    Segmenting the PM poles

  • 8/2/2019 Danhong Presentation

    27/43

    Other Rotor Design Aspect

    Laminated backiron

    Different permanent magnet materials

  • 8/2/2019 Danhong Presentation

    28/43

    Stator Design: 36 Stator Slots

  • 8/2/2019 Danhong Presentation

    29/43

    Stator Design: Open Slot vs. Closed Slot

    open slot

    closed slot

    0 20 40 60 80 1000

    0.05

    0.1

    0.15

    0.2

    0.25

    harmonics

    los

    s(w)

  • 8/2/2019 Danhong Presentation

    30/43

    Techniques for Reducing Rotor Losses

    Laminating rotor backiron

    Segmenting the Permanent magnet poles

    Increasing slot number

    Closing the stator slots

  • 8/2/2019 Danhong Presentation

    31/43

    Summary

    Purpose of Finite element analysis for electric machines

    The flywheel energy storage system

    The steady-state nonlinear finite element solver

    Its application

  • 8/2/2019 Danhong Presentation

    32/43

    Further research into numerical modeling of electric

    machines

    Special machine design and analysis

    Future Work

    2011 Chevrolet Volt Propulsion System

    (Credit: GM)

  • 8/2/2019 Danhong Presentation

    33/43

    Thank you!

    This work has been supported by NASA Grant NAG3-2598

  • 8/2/2019 Danhong Presentation

    34/43

    Currents flowing in permanent magnet are eddy currents. In machinedesign, magnets can be electrically insulated from each other and the

    rotor backiron.

    Eddy currents meet large impedance at the

    end of the machine, surface charge will

    accumulate and electric potential is built

    across the magnet

    Permanent Magnets

    )(

    1 2

    0AMAt

    A

  • 8/2/2019 Danhong Presentation

    35/43

    Electric Scalar Potential Dynamics

    t

    AJeddy

    (2)s

    dsJs

    eddy

    c

    c

    dst

    AS

    s

    )(

    The charge relaxation time constant / is extremely small inpermanent magnets, and so our system consists of a set of fast

    dynamics (electric scalar potential) and slow dynamics (magneticvector potential). Therefore, singular perturbation techniques can

    be used to analyze the system.

  • 8/2/2019 Danhong Presentation

    36/43

    Singular Perturbation Analysis Techniques

    Under singular perturbation theory, when analyzing the fastdynamics, the slow variables can be assumed to be essentially

    constant

    If the fast dynamics are stable under this condition, the fast

    variables will converge to a quasi-steady-state value, which is a

    function of the slow variables When analyzing the slow dynamics, it can be assumed that the

    fast variables have converged to their quasi-steady-state value

    discussed above.

    0

    Sdst

    A

    S

    m

    n

    Rzztztzxgz

    Rxxtxtzxfx

    ,)(),,,,(

    ,)(),,,,(

    00

    0

    0

    small

    ),0,

    ~

    ,

    ~

    (0 tzxg

    (3)dst

    AS

    s

    )(

  • 8/2/2019 Danhong Presentation

    37/43

    Applying Finite Element Method

    To solve the Partial Differential Equations

    use 2D finite element methodPrimary variable : A and Element type : triangular

    Shape functions : linearProcedure used : weighted residuals

    Error distribution principle :

    Galerkins method

    Boundary conditions:

    0zA

  • 8/2/2019 Danhong Presentation

    38/43

    Newton-Raphson Method

    Applying the Newton-Raphson method, we get:

    WhereJis the Jacobian of the nonlinear state transition function

    )()(11

    1

    jjj

    xxfxx

    x

    fj

    Define : and its called the Jacobian ofjxf xfJ

    )(xf

    (5)

    In our solver we define

    For a nonlinear equation, the Newton-Raphson Method provides thefollowing iterative procedure to solver for x:

    0)( xf

  • 8/2/2019 Danhong Presentation

    39/43

    Equation solving

    Now we look closely at the steady state iterative equation :

    1

    1

    1

    )0(

    )(),0,)0((

    j

    jj

    x

    TxTxJ

    In (5), is not known explicitly but is determinedthrough Backward Euler Integration. Thus:

    11)(),0,)0(( jj TxTx

    (5)

    can not be explicitly written either and would have to be calculated using aform of numerical integration

  • 8/2/2019 Danhong Presentation

    40/43

    Equation solving (2)

    One way of solving for the Jacobian is to differentiate both sides of (4)with respect to

    J

    Where and are matrices derived from (4). can then be computed by

    repeatedly solving (6) starting from the initial conditionfC pC

    J

    Computation load analysis of this method:Matrix-Matrix multiplication on the right side and LU-factorization on the left

    side. If the vector has components, the computation work per time step is

    of at least orderN

    2N

  • 8/2/2019 Danhong Presentation

    41/43

    The Backward Euler Integration Method calculate y at step

    n+1 implicitly by :

    Backward Euler Integration

    Consider the problem:

    )()(2

    11 hOyhfyy nnn

    In the solver we separate the period T into a number of time

    steps h. Provided a solution at time t1, x(t1), we then cancompute the solution x(t1+h) at time t1+h, by applying the

    backward Euler integration to (5) over the time interval h:

    (5)

    (7)

  • 8/2/2019 Danhong Presentation

    42/43

    Application of the GMRES

    Computation load analysis of this method:

    Now we consider solving the problem using GMRES method. By usingGMRES method, we dont need to provide explicitly. Instead, we only needto provide . We apply the procedure (6) to a vector

    J

    Matrix-vector multiplication on the right side and LU-factorization on the left

    side. Typically the number of iterations required by GMRES to achieve a

    sufficiently low relative error is substantially smaller than problem size. So by

    using GMRES, the numerical efficiency is improved.

  • 8/2/2019 Danhong Presentation

    43/43

    The eddy currents in rotor are :Post-processing:

    rotor loss calculation

    After we achieve the steady-state solution at time 0, the entire response

    x(t) can be calculated by integrating (5) for one period. We perform aDiscrete Fourier Transform of x(t) and calculate the eddy current rotorlosses for each harmonic in each element.

    Se

    iiis

    Se

    ie

    s

    ei

    dSAjl

    dSJl

    P

    2

    2

    )()(

    )(

    t

    AJ