IEEE PES General Meeting July 13 - 17, 2003, Toronto

Transformer Modeling for Simulation of Low Frequency Transients

J.A. MARTINEZMARTINEZ-VELASCO Univ. Politcnica Catalunya Barcelona, Spain B.A MORK Michigan Tech. Tech. Univ. Houghton, Houghton, USA

Introduction

Large number of core designs Some of transformer parameters are both nonlinear and frequency dependent Physical attributes whose behavior may need to be correctly represented
core and coil configurations selfself- and mutual inductances leakage fluxes skin effect and

between coils

magnetic core saturation and eddy current losses in core capacitive

proximity effect in coils hysteresis

effects

Aim of this presentation

Transformer Models

Matrix representation
BCTRAN

model

Saturable Transformer Component (STC) Topology-based models

Duality

based models models

Geometric

Transformer Models
Matrix representation (BCTRAN model)

Branch impedance matrix of a multi-phase multiwinding transformer

Steady state equations Transient equations

[V ] = [ Z ] [ I ]

[ v] = [ R] [i ] + [ L] [ di / dt ]

[R] and j[L] are the real and the imaginary part of [Z], whose elements can be derived from excitation tests

The approach includes phase-to-phase couplings, models terminal characteristics, but does not consider differences in core or winding topology

Transformer Models
Saturable Transformer Component (STC model)
L1 R1
i

N1 : N2

R2

L2

Lm

Rm ideal N1 : NN RN . . .

LN

ideal

StarStar-circuit representation of singlesingle-phase NNwinding transformers

Transformer Models
Models derived using duality

Core design

Equivalent circuit

DualityDuality-derived model for a singlesingle-phase shellshellform transformer

MODEL

EQUATIONS
[R] [L] option

CHARACTERISTICS
These models include all phase-to-phase coupling and terminal characteristics. Only linear models can be represented. Excitation may be attached externally at the terminals in the form of non-linear elements. They are reasonable accurate for frequencies below 1 kHz. It cannot be used for more than 3 windings. The magnetising inductance is connected to the star point. Numerical instability can be produced with 3-winding models.

Matrix Representation (BCTRAN model)

[ v ] = [ R] [ i ] + [ L] [ di / dt ]
[A] [R] option

[di / dt ] = [ L]1[v ] [ L]1[ R][i ]

Saturable Transformer Component (STC model)

[ L] [v ] = [ L] [ R][i ] + [di / dt ]

Topology-based models

Duality-based models : They are de- Duality-based models include the effects of saturation in each individual leg of the rived using a circuit-based approach core, interphase magnetic coupling, and without a mathematical description leakage effects. Geometric models The mathematical formulation of geome[v ] = [ R][i ] + [d / dt ] tric models is based on the magnetic equations and their coupling to the electrical equations, which is made taking into account the core topology. Models differ from each other in the way in which the magnetic equations are derived.

Nonlinear and Frequency-Dependent Parameters

Some transformer parameters are nonlinear and/or frequency-dependent due to

saturation hysteresis eddy

currents

Saturation and hysteresis introduce distortion in waveforms Hysteresis and eddy currents originate losses Saturation is predominant in power transformers, but eddy current and hysteresis effects can play an important role in some transients

Nonlinear and Frequency-Dependent Parameters

Modeling of iron cores

Iron

core behavior represented by a relationship between the magnetic flux density B and the magnetic field intensity H Each magnetic field value is related to an infinity of possible magnetizations depending on the history of the sample To characterize the material behavior fully, a model has to be able to plot major and minor hysteresis loops
(minor loops can be symmetric or asymmetric)

Nonlinear and Frequency-Dependent Parameters

B Initial curve Anhysteretic curve Major loop Symmetric minor loop H Asymmetric minor loop

Magnetization curves and hysteresis loops

Modeling of iron cores

+

i R I

Equivalent circuit for reprerepresenting a nonlinear inductor V

Hysteresis loops have a negligible influence on the magnitude of the magnetizing current

Hysteresis losses can have some influence on some transients; the residual flux has a major influence on the magnitude of inrush currents

The saturation characteristic can be modeled by a piecewise linear inductance with two slopes, except in some cases, e.g. ferroresonance

Eddy current effects

Excitation losses are mostly iron-core losses

hysteresis

and eddy current losses they cannot be separated hysteresis losses are much smaller than eddy current losses

Eddy current models for

transformer

windings iron laminated cores

Eddy current effects

Models for windings
R1 R0 L1 L2 LN R2 RN

Series Foster equivalent circuit

Eddy current effects

Models for iron laminated cores
R1 L1 L2 R2 LN RN

Standard Cauer equivalent

LN

L1

L2

R1

R2

RN

Dual Cauer equivalent

Transformer Models
Matrix representation (BCTRAN model)
A B C a b c

Winding Leakages BCTRAN Model ZL ZL ZL

ZY

ZY
Transformer Core Equivalent

BCTRAN model for a threethree-phase threethree-legged stacked core transformer

Parameter Determination

Data usually available for any power transformer

power rating voltage rating excitation current excitation voltage excitation losses shortshort-circuit current shortshort-circuit voltage shortshort-circuit losses saturation curve capacitances between terminals and between windings

Excitation and shortshort-circuit currents, voltages and losses must be provided from both direct and homopolar measurements

Parameter Determination

An accurate representation for threethree-phase core transformers should be based on the core topology, include eddy current effects and saturation/hystere saturation/hysteresis representation A very careful representation and calculation of leakage inductances is usually required CoilCoil-capacitances have to be included for an accurate simulation of some transients Since no standard procedures have been developed, a parameter estimation seems to be required regardregardless of the selected model Temperature influence should not be neglected

Modeling - Case 1

Three-legged stacked-core transformer Cross section of core and winding assembly

Modeling - Case 1
L0 Rh Ll R0 Rm Lm Rl

Ly

L0 Rh

Ry Ll Rm Lm

Rl

R0

Ly

Ry L0 Rh Ll R0 Rm Lm Rl

Three-legged stacked-core transformer Duality-based equivalent circuit

Modeling - Case 1

Three-legged stacked-core transformer ATP implementation

Modeling - Case 1
10 Phase A Phase B Phase C

5 Current (A)

-5

-10 100

120

140 160 Time (ms)

180

200

Three-legged stacked-core transformer Excitation currents

Modeling - Case 1
Phase A 180 120 60 0 -60 -120 180 120 60 0 -60 -120 180 120 60 0 -60 -120 0 100 200 Time (ms) 300 400 500 Phase B Phase C

Three-legged stacked-core transformer Inrush currents

Ferroresonance - Case 2

CORE U SRC SRCX X1 Rw Leak Rc 98_8Seg

Magnetic Saturation - Case 2

0.8 Fluxlinked [Wb-T] 0.8 Fluxlinked [Wb-T] 0.6 0.6

0.4

0.4

0.2

0.2

0.0 0.0 62.7 125.3 188.0

I [A] 0.0 250.7 0.0 62.7 125.3 188.0

I [A] 250.7

8-Segment Curve vs. 2-Segment Curve

Steady-State Excitation - Case 2

200 [V] 150 100 50 0 -50 -100 -150 -200 0 10 20 c:SRC -XFMR 30 40 [ms] (file FR_Mart.pl4; x-var t) v:XFMR Excitation at Rated Voltage 8-Segment Curve 2.0 [A] 1.5 1.0 0.5 0.0 -0.5 -1.0 -1.5 -2.0 50

Ferroresonance - Case 2A
400 [V] 300 6.8 200 100 0 -100 -200 -8.8 -300 -400 0.00 -14.0 66.67 -3.6 Ferroresonance: 15uF, 8-Segment Magnetization Curve 12.0 [A]

1.6

11.11

22.22 v:SRCX

33.33 c:SRC -SRCX

44.44

55.56

[ms]

(file FR_Mart.pl4; x-var t) v:X1

Ferroresonance - Case 2B
350.0 [V] 262.5 175.0 87.5 0.0 -87.5 -175.0 -262.5 -350.0 0.00 -25 20 Ferroresonance: 15uF, 2-Segment Magnetization Curve 50 [A] 35

-10

11.11

22.22 v:SRCX

33.33 c:SRC -SRCX

44.44

55.56

[ms]

-40 66.67

(file FR_Mart.pl4; x-var t) v:X1

Effect of Mag Curve Representation

400 [V] 300 200 100 0 -100 -200 -300 -400 0.00 Ferroresonant Voltage: Red(2A): 8-Seg; Blue(2B): 2-Seg @ 1.0A; Green: 2-Seg @ 1.5A

11.11

22.22

33.33

44.44

55.56

[ms]

66.67

FR_Mart.pl4:v:X1 FR_MART1p5.pl4: v:X1 FR_MART1p0.pl4: v:X1

Effect of Mag Curve Representation

45 [A] 30 15 0 -15 -30 -45 0 10 20 30 40 [ms] 50
FR_Mart.pl4:c:SRC -SRCX FR_MART1p5.pl4: c:SRC -SRCX FR_MART1p0.pl4: c:SRC -SRCX Ferroresonant Current: Red(2A): 8-Seg; Blue(2B): 2-Seg @ 1.0A; Green: 2-Seg @ 1.5A

Conclusions

There is no agreement on the most adequate model Modeling difficulties

great

variety of core designs nonlinear and frequency dependent parameters inadequacy for acquisition and determination of some transformer parameters

Several modeling levels could be considered since not all parameters have the same influence on all transient phenomena