Anda di halaman 1dari 126

; A DETAILED DERIVATION OF A

-.".

NEWTON-RAPHSON BASED HARMONIC POWER FLOW/"


i

A Thesis Presented to
The Faculty of the Russ College of Engineering and Technology
Ohio University

In Partial Fulfillment
of the Requirements for the Degree
Master of Science

Heidt,
David Charles.--

i
.

June, 1994

ACKNOWLEDGMENTS

The author would like to thank all those who helped to provide the guidance
needed to complete this thesis, including Dr. Herman Hill and Dr. Brian Manhire of Ohio
University, and Dr. Jerry Heydt of Purdue University.
The author would also like to thank the Electrical and Computer Engineering
Department at Ohio University, for providing the financial support necessary to pursue the
Master of Science degree.
Lastly, the author would like to thank family and friends who were patient,
supportive and understanding through the many hours of sacrifice.

TABLE OF CONTENTS

Page
LIST OF TABLES..........................................................................................................vi
LIST OF FIGURES. ..... ...............................................,............................. . .......... ... ......vii
LIST OF SYMBOLS.......................................................................................................x

...

ABSTRACT............... ................... .............................. .................................................mil


CHAPTER 1 -- CONTRIBUTIONS OF THIS THESIS..................................................1

CHAPTER 2 -- CAUSES AND DETRIMENTAL EFFECTS OF POWER


SYSTEM HARMONICS............................................................ .... ........5
2.1 Definitions Related to Harmonics.......................................... ....... ...................5
2 . 2 The Cause of Harmonics in Power Systems......................... . ....... . ...................8
2.3 Detrimental Effects of Harmonics on Power Systems....................................18

CHAPTER 3 -- A NEWTON-RAPHSON BASED HARMONIC POWER


FLOW................................................. .. ............ ... ............... . ... .............22
3 . 1 Introduction............................................................. ... ...... . ....... , ...................22
3 . 2 Solving Nonlinear Algebraic Equations by Newton-Raphson
Methods......................................... ...................................... ................... .....25
3.3 A Brief Derivation of a Newton-Raphson Based Conventional
Power Flow.................................................................... ..............................33
3 . 4 A Detailed Derivation of a Newton-Raphson Based Harmonic
Power Flow............ ........... .................... ............................. ...... ... .... ........ .....39
3.5 Network Models for the Newton-Raphson based Conventional
and Harmonic Power Flows.................................................................. ........53
Introduction..................................................... ........................... ............53
. .
Transmission Lines................................................ ... ............. .......... ........53
Transformers............. .. ... ... ..... .................. ........................................ .......58
Generators............................................................ ....................... . ..........60
Induction Motors.. ............................................................. .......... . ..... ... ..6 1

Other Conventional Loads.......................................................................66


Gas Discharge Lighting Loads.................................................................68
3.6 Simplifications to the Newton-Raphson Based Harmonic Power
Flow Equation Set........................................................................................ 72
CHAPTER 4 ..DIRECT SOLUTION OF POWER SYSTEM HARMONIC
VOLTAGES......................................................................................... 76
CHAPTER 5 ..EXAMPLES......................................................................................... 79
5.1 Introduction..................................................................................................79
5.2 Example One................................................................................................90
5.3 Example Two............................................................................................... 92
5.4 Example Three............................................................................................101
CHAPTER 6 ..CONCLUSIONS AND FUTURE WORK ...........................................103
6.1 Conclusions................................................................................................
103
6.2 Future Work ...............................................................................................106
LIST OF REFERENCES .............................................................................................108
APPENDIX.................................................................................................................112

LIST OF TABLES

Table

Page

3.1

Sequence of a Fundamental Frequency Signal and its Harmonics in a


Completely Balanced Three-Phase ac Power System.......................................... 2 4

3.2

Skin Effect Table................................................................................................


57

5.1

Fundamental Frequency (60Hz)System Generator Data .....................................80

5.2

Data for the Conductors of the 230 Kv Transmission Line of Figures 5 . 2............8 2

5.3

Data for the Ground Wires of the 230Kv Transmission Line of Figure 5 . 2..........83

5.4

Impedance Data for the 230 Kv Transmission Line of Figure 5 . 2.........................83

5.5

Per-Unit Impedance Data for the 230 Kv Transmission Line of Figure 5 . 2........... 83

5.6

Data for Polynomial Curve Fits...........................................................................


96

LIST OF FIGURES

Figure

Page

2.1.

Generalized Six-Pulse Converter..........................................................................


.9

2.2.

A Simple Circuit Diagram of a Fluorescent Lamp Placed in Series with


.12
an Inductive Ballast...........................................................................................

2.3.

Typical B-H Curve used to Describe the Characteristics of the Material


,I5
used for a Transformer Core..............................................................................

2.4.

An Example of Actual B-H Curves....................................................................


.I6

3.1.

Geometrical Interpretation of the Newton-Raphson Method............................ .27

3.2.

An Example of How the Next Approximation to f ( x ) = 0 can be Far


Away from the Desired Solution x" if Iteration x r is Close to the
Minimum of a Function f ( x ).............................................................................28

3.3.

Illustration of the Two Systems Given by Equation Sets (3.7) and (3.8)..............30

3.4.

Single-Line Diagram at Bus i of a General n-Bus Power System.........................34

3.5.

The Equivalent Positive-Sequence-ImpedanceDiagram of Figure 3.4.................34

3.6.

General n-Bus Power System Arranged to Form the Fundamental


.....................................................
Frequency Admittance Matrix............. .
.

.36

3.7.

A Fundamental Frequency Admittance Diagram of a Two-Bus Power


System...............................................................................................................
.36

3.8.

Typical ac-Side Line Current Waveform of an ac Current to dc Current


Converter..........................................................................................................
.40

3.9.

(a) A Bus Containing a Generator and a Harmonic Producing (HP)


Device. (b) An Equivalent Representation of (a), Separated into Two
44
Buses Connected by a Short Circuit (SC)............................................................

General n-Bus Power System Arranged to form the kth Harmonic


Frequency Admittance Matrix.............................................................................47

A General Representation of a Power System at the kth Harmonic


Frequency, with a Harmonic Producing (HP) Load at Bus m.. .............................48
(a). A General Representation of a Power System at the kth Harmonic
Frequency, with a Conventional (C) Load at Bus i. (b). Equivalent
Representation of (a), with the Conventional Load Modeled as a Series
. .
R-L Combination................................................................................................50
Pi-Equivalent Model of a Long Transmission Line..............................................54
Pi-Equivalent Model of a Short Transmission Line. .............................................56
Power Flow Model of a Transformer..................................................................59
Harmonic Power Flow Model of a Synchronous Machine for the kth
Harmonic Frequency...........................................................................................62

A Per-Phase Fundamental Frequency Equivalent Model for an


Induction Machine (Phase a Shown)...................................................................63
Harmonic Power Flow Model of an Induction Motor for the kth
Harmonic Frequency...........................................................................................65
Suggested Harmonic Power Flow Model of an Unknown Conventional
Load for the kth Harmonic Frequency.................................................................67
General n-Bus Power System Arranged to form the kth Harmonic
Admittance Matrix. Shunt Impedances used to Model Conventional
Buses at the kth Harmonic Frequency, are Included when Forming
the kth Harmonic Frequency Admittance Matrix.................................................73
Representation of a Harmonic Producing Device at Bus m, as an
Ideal Current Source at the kth Harmonic Frequency..........................................77
Example Two-Bus Power System, with a Generator at Bus One and
a Fluorescent Lighting (FL) Load at Bus Two....................................................80
Configuration of the 230Kv Transmission Line in Figure 5.1..............................81
Impedance Diagram of the Power System in Figure 5.1, for
Fundamental Frequency (60Hz) Analysis...........................................................85

5.4.

Impedance Diagram of the Power System in Figure 5.1, for Third


85
Harmonic (180Hz) Analysis...............................................................................

5.5.

Fourier Series Components of Equation (5.10)...................................................88

5.6.

Summation of all Three Terms in Equation (5.10)...............................................


89

5.7.

Polynomial Curve Fits.........................................................................................


97

LIST OF SYMBOLS

In order to eliminate confbsion, frequencies of interest and iteration numbers will


be represented by superscripts. In addition, bus numbers, phase and phase sequences will
be represented by subscripts. For clarity, the most commonly used superscripts and
subscripts used in this thesis are explained below. Other superscripts and subscripts are
used, but are explained as they are presented.

superscripts

(1)

Fundamental frequency.

(3), (5),(7), ...... Harmonic frequencies.


(f)

The frequency of interest (f 2 1,f odd).

(k)

A harmonic frequency (k

(h)

The highest harmonic frequency of interest (h > 1,h o d )

A general iteration number.

2 3, kodd)

subscripts
i

A conventional (nonharmonic producing) bus.

A harmonic (producing) bus.

A conventional or harmonic bus ( j = i or m).

Phase a.

Phase b.

Phase c.

Direct.

Quadrature.

Neutral.

For this thesis, all power system values shall be expressed in per-unit, instead of
conventional SI units. In addition, all currents and voltages are rms values only. The most
commonly seen variables in this thesis are explained below. Other variables appear in this
thesis, but are explained as they are presented.

A voltage.
A current entering a conventional bus from ground.
A current entering a harmonic (producing) bus to ground.
Time.
A hndamental or harmonic component of a voltage Fourier Series.
A hndamental or harmonic component of a current Fourier Series.
Fundamental angular frequency.
Angular frequency.
Active power.
Reactive power.
Apparent power.
Complex power.
Distortion power.

A fundamental (f = 1) or harmonic (f > 1, f odd) frequency current entering bus j


( j = i or m) from ground into the transmission network. 7
p = I:)/#'.

A hndamental (f = 1) or harmonic (f > 1, f odd) frequency current entering bus j


( j = i or m ) from the transmission network to ground.

7'"

gg'/g-"
-.

A fundamental (f = 1) or harmonic (f > 1, f odd) frequency voltage at bus j


( j = i or m )

y(n = yw/q).
-

Impedance. Z = R + jX

Resistance.

Reactance.

Inductance.

Capacitance.

Admittance. 7 = y / g = g + jb

IFfll A fundamental (f

Iz~)I

= 1) or harmonic (f > 1, f odd) frequency admittance matrix.

A fundamental (f = 1) or harmonic (f > 1, f odd) frequency impedance matrix.

IlWI A fundamental (f = 1) or harmonic (f > 1, f odd) frequency current vector.

Ifl)I A fundamental (f

= 1) or harmonic (f > 1, f odd) frequency voltage vector

The total number of harmonic frequencies of interest.

The total number of harmonic (producing) buses in the power system considered.

Unknown value.

CL

Permeability.

An incremental change.

p.u.

Per unit.

rad.

Radians

ABSTRACT
Heidt, David Charles. M. S., Ohio University, June 1994. A Detailed Derivation of a
Newton-Raphson Based Harmonic Power Flow. Major Professor: Dr. Herman Hill.

The ideal electric power system contains only elements that generate, transmit, or
receive undistorted hndamental frequency voltages and currents. Several devices are
responsible for introducing harmonics of pure hndamental frequency waveforms into
power systems. If significant enough in amplitude, these harmonics can have a detrimental
effect on the performance and life of power system elements.
A tool commonly used to analyze the power system under normal balanced threephase sinusoidal steady-state conditions, is called a power flow. However, to this date, the
only power flows which are readily available in textbook form, assume pure hndamental
frequency voltages and currents throughout the power system.
It is the primary intent of this thesis, to take an available Newton-Raphson based
Harmonic Power Flow algorithm, and provide a "textbook" type derivation. This
algorithm eliminates the assumption of purely sinusoidal voltages and currents in a threephase power system operating under normal balanced steady-state conditions.

CHAPTER 1 -- CONTRIBUTIONS OF THIS THESIS

Chapter two focuses on the cause and detrimental effects of power system
harmonics. The chapter begins by discussing pertinent definitions related to power system
harmonics. Several equations in this introductory section and throughout this thesis paper,
are written in Fourier Series form. The method of Fourier Series is a convenient and
appropriate way to express the presence of harmonics in a power system.
The next section of chapter two describes some of the most common sources of
harmonics in power systems today. Power electronic devices are by far the most
significant source of harmonics in power systems today, not only because of the nature of
these devices, but also because of their rapidly growing usage [4]. Several other devices
such as gas discharge lighting, transformers, and arc furnaces are also significant harmonic
sources in power systems.
Chapter two is concluded by describing several of the detrimental effects that
harmonics can have on power systems. Here it is described how harmonics can cause
several devices such as relays and computers to misoperate. This section also discusses
why harmonics can lead to a reduction in the efficiency and life of several other devices
such as motors and transmission lines. It is also discussed how power system harmonics
can be amplified significantly depending upon the power system configuration and why
triplen harmonics (i.e., the third harmonic and its integer multiples) represent a special and
rather complicated problem. Chapter two provides sufficient information as to what power
system harmonics are, how they originate, and why they cannot be ignored.
The primary focus of chapter three is to derive a Newton-Raphson based
Harmonic Power Flow. Chapter three begins by defining both the conventional and

harmonic power flows. This introduction describes the differences between conventional
and harmonic power flows. In addition, this introduction states why a conventional power
flow cannot be used to obtain information about power system harmonics.
The next section of chapter three discusses the solution of nonlinear algebraic
equations by the Newton-Raphson method, along with some of the most common
problems encountered when using this method. One needs to be aware of how the
Newton-Raphson method can fail for the problem to be analyzed.
Chapter three continues by providing the derivation for the Newton-Raphson
based Conventional Power Flow. These equations are then used to help derive the
Newton-Raphson based Harmonic Power Flow. This is meant to show that the NewtonRaphson based Harmonic Power Flow is a logical extension of the Newton-Raphson
based Conventional Power Flow. The derivation begins by describing the number of
equations required for the Newton-Raphson based Harmonic Power Flow and explaining
briefly how they are obtained. The derivation continues by listing all required equations
and providing a detailed explanation for the logic behind each equation as they are
provided. The equation set obtained is derived in a general format, and is to be treated the
same as any other nonlinear algebraic equation set which is to be solved by the NewtonRaphson method. The detailed derivation of the Newton-Raphson based Harmonic Power
Flow, represents the author's main contribution to knowledge. This derivation is in a much
more general format than in reference [I]. In addition, intricate details provide more
insight than is available in reference [I]. To the best of the author's knowledge, reference
[ I ] had provided the best derivation of the Newton-Raphson based Harmonic Power Flow

to date.
The following section of chapter three provides network models for the NewtonRaphson based Conventional and Harmonic Power Flows. Network models for some of
the most common power system elements is provided. It is stressed that the hndamental
frequency power system representation will change for each harmonic frequency of

interest. A different admittance matrix is required for each frequency to be examined. The
equations for modeling one harmonic source (gas discharge lighting) is derived. All
harmonic producing loads cannot be modeled in the same way however, since the current
entering each different type of harmonic load will possess a different Fourier Series
The last section of chapter three discusses possible simplifications to the NewtonRaphson based Harmonic Power Flow equation set. Since this derivation is provided in a
general format, it is possible to model every bus in the power system the same way that
harmonic producing buses are modeled. In other words, for every bus in the power
system, the Fourier Series of the current entering each bus could be expressed as a
hnction of the Fourier Series of the voltage at the respective bus, and of any parameters
which describe this distorted waveform. The simplifications possible result when this
approach is used only for harmonic producing busses. The derivation of the NewtonRaphson based Harmonic Power Flow provided in this thesis does not assume that a
simplified approach will always be used, whereas reference [I] does.
Chapter four briefly discusses the most commonly used method for determining
power system harmonics voltages called the Current Injection Technique. In order to find
the power system harmonic voltages, this method employs the direct (non-iterative)
solution of the harmonic frequency admittance matrices. As the voltage distortion levels
increase throughout the power system, the less effective the Current Injection Technique
becomes at determining the power system harmonic voltages [I]. In addition, the Current
Injection Technique is generally effective only when the power system is a simple radial
network [l 1.
Chapter five provides three examples to help increase the reader's knowledge of
the Newton-Raphson based Harmonic Power Flow. The same two bus power system is
analyzed in all three examples. In addition, the goal is to obtain the same information in all
three examples. In the first example, the Newton-Raphson based Conventional Power
Flow is used to obtain the kndamental frequency voltages in the power system. The

Current Injection Technique is then used to determine the harmonic voltages in this power
system.
The second example in chapter five uses the Newton-Raphson based Harmonic
Power Flow to determine the hndamental and harmonic frequency voltages in this same
power system. For each bus in this example, the Fourier Series of the current entering the
bus, is modeled as a hnction of the Fourier Series of the voltage at this bus, and of any
parameters which describe this distorted current waveform. In this case, harmonic
frequency admittance matrices include only the admittance information for transmission
lines and transformers, as is true for the hndamental frequency admittance matrix.
The last example in chapter five uses the simplified approach which always
assumed to be true in reference [I]. Therefore, all buses which do not contain harmonic
sources are treated as an impedance load at harmonic frequencies. In this case, harmonic
frequency admittance matrices include admittance information for transmission lines,
transformers, and impedance loads (for nonharmonic producing loads). The hndamental
frequency admittance matrix, however, still only includes admittance information for
transmission lines and transformers. Chapter five also represents the author's contribution
to knowledge. To the best of the author's knowledge, detailed examples on how to
perform a Newton-Raphson based Harmonic Power Flow are not available.
Chapter six completes the thesis by listing appropriate conclusions and suggestions
for future work.

CHAPTER 2

-- CAUSES AND DETRIMENTAL EFFECTS OF POWER SYSTEM


HARMONICS

2.1 Definitions Related to Harmonics


Harmonic producing loads have a non-sinusoidal load current that is periodic, and
can be expressed as a Fourier Series. The distorted load current passing through the
power system results in distorted bus voltages that are periodic, and can also be expressed
as a Fourier Series. All frequency terms higher than the hndamental frequency, will be
referred to as the harmonics.
For this thesis, completely balanced power systems will be assumed, so that even
harmonics may be ignored. A property of a completely balanced power system is that even
harmonic currents and voltages will not exist [20]. For this thesis, harmonic frequencies
shall be assumed to be odd integer multiples of a 60Hz kndamental frequency. By
definition of a Fourier Series, any voltage v(t) or current i(t) in this balanced power
system, must be an odd function, as shown below ( f 2 1,f odd).

i(t) =

cV)sin (foot +

PA)

Also, for this completely balanced power system, the active power P , and reactive
power Q will be defined as follows (f 2 1,f odd).

In addition to the above equations, the apparent power S is a scalar quantity given
by equation (2.5).The root mean square of the voltage and current is defined by equations
(2.6) and (2.7)respectively ( f 2 1,f odd).

Harmonic distortion of a voltage v ( t ) and a current i(t) is defined by equations


(2.8) and (2.9)respectively ( f 2 1,f odd).

The definition of power factor is given below.

In the special case when both v ( t ) and i ( t ) contain fbndamental frequency


components only, equation (2.11)will be valid.
S2 = p2 + Q 2

(2.11)

However, when either v(t) or i ( t ) contains harmonic frequency components,


equation (2.11)is no longer valid. A term that is used to account for the discrepancy is
called the distortion power D, as given below.

D=,/m

(2.12)

The physical origination of D in power systems is an extremely controversial issue


[3], and will not be discussed in this thesis. The concept of distortion power is mentioned

here only for the sake of completeness.

2.2 The Cause of Harmonics in Power Systems

The generation of harmonics in power systems today, results from a number of


major sources. However, power electronic devices are by far the most significant sources
of harmonics [4] [ 5 ] .
For example, power electronic six-pulse converter devices used as rectifiers (to
convert ac current to dc current) and as inverters (to convert dc current to ac current) are
growing in usage because they are more efficient and economical than conventional
converter systems such as motor-generator sets [ 6 ] .A few of their applications include
he1 cells and batteries, static var generators, and adjustable-speed drives for motors.

A generalized six-pulse converter configuration is shown in Figure 2.1. When a


square wave switching scheme is used, each switch changes state only twice for a
designated switching time period. Whether the converter is used as an inverter or as a
rectifier, significant ac side harmonics of the hndamental current will result. Under
balanced conditions, the characteristic ac side harmonic currents produced by a six-pulse

converter are k = 6b 1(b = 1,2,3,...) [ 6 ] .The magnitude of these harmonics decrease


inversely proportional to their harmonic order, i.e.,
hndamental frequency current magnitude, and

= I"'/k, where I") is the

is a harmonic current magnitude.

Uncharacteristic harmonics may also be produced under unbalanced conditions, where the
magnitudes depend upon the amount of unbalance [ 7 ] .

A switching scheme for the six-pulse converter which is recently becoming more
popular is called the pulse-width modulated (PWM) scheme. For the PWM scheme, the
switches change state for a designated switching time period at a higher frequency than the
square wave switching scheme. Basically, the higher the switching frequency, the greater
the number of lower order harmonics that can be eliminated. However, the PWM scheme
results in the generation of higher frequency ac side harmonics. These harmonics can be
even greater in amplitude than the lower frequency ac side harmonics generated by the
square wave switching scheme. These higher frequency harmonics and their amplitudes

Figure 2.1. Generalized Six-Pulse Converter.


Source: Boost [8], p.273.

depend upon the switching frequency of the PWM converter. In addition, a PWM
converter produces less hndamental frequency current than a converter with a square
wave switching scheme.
However, the PWM scheme does have advantages over the square wave switching
scheme. For example, the main advantage of the PWM scheme is that a large number of
lower order harmonics can be eliminated. This results in a more sinusoidal ac side
waveform than a square wave switching inverter can produce. In addition, a square wave
switching inverter is not capable of regulating the ac output voltage magnitude. Therefore,
the dc input voltage must be adjusted in order to control the magnitude of the inverteroutput voltage [8]. Since the PWM inverter allows amplitude control of the ac output
voltage from within the device, this feature can result in a simpler and cheaper power
inverter [8].
It should be noted that 12-pulse converters are actually much more common,
particularly in new designs. One example of where 12-pulse converters are more
advantageous than 6-pulse converters is in High-Voltage DC (HVDC) transmission
applications. In order to meet high voltage requirements, and to reduce the number of
harmonics produced, it is necessary to use 12-pulse converters, instead of 6-pulse
converters. It is noted that under balanced conditions, the characteristic ac side harmonic
currents produced by a 12-pulse converter will be k = 12b

+ 1 (b = 1,2,3,......) [6]. The

magnitudes of the remaining harmonics will be the same as for a six-pulse converter.
Fuel cells and batteries are highly likely to grow in usage in power systems as
energy sources during utility peak loading conditions [6]. It is desirable to operate the
most efficient power generating plants (such as nuclear and the newer high-efficiency coalfired plants) at their rated capacity at all times. Due to the time of day and weather
conditions, the utility load demand will not remain constant. In order to meet peak loading
conditions, either oil- or gas-fired generators can be used, but are expensive to operate
because of the high cost of hel. An alternative is to store the energy generated by the

more efficientpower generating plants in batteries or fuel cells during low-load conditions,
and to use this surplus during peak-loading conditions. However, since both batteries and
fuel cells produce a dc voltage, an inverter is required to connect them to the utility
system.

A converter system can be designed such that the ac current can be quickly
controlled in magnitude and phase (leading or lagging) with respect to the ac voltage of
the device. Such a converter is commonly termed a static var generator, and can be used
for power factor correction applications [8]. Industrial loads such as arc furnaces can
cause very rapid changes in power factor. In this case, a fast and efficient static var
generator would be more desirable to use than a slower conventional power-factorcorrection capacitor bank.
Power electronic converter devices are commonly used as adjustable-speed drives
for motors [9]. For instance, a common configuration for an ac motor drive is made up of
a rectifier (ac to dc) supplying an inverter (dc to ac). By controlling the direction of the
power flow, this adjustable-speed drive configuration can allow precise control over the
motoring and braking of the ac motor [lo].
Gas discharge lighting (such as fluorescent, mercury arc, mercury vapor, neon,
xenon, and high pressure sodium) represent a major source of harmonics, particularly in
metropolitan areas [ll]. In particular, fluorescent lighting may constitute up to twenty-five
percent of the total customer load, and may exceed thirty percent [12]. The wide usage of
fluorescent lamps as compared to other lamp types is due to their high energy efficiency.
Greater stability in the operation of conventional 6OHz gas discharge lamps is
provided by an inductive ballast placed in series with the lamp [l 11. A simple circuit
diagram of a fluorescent lamp placed in series with an inductive ballast is shown in Figure
2.2, where vs is the ac source supply.

-+-

v ballast 4

Figure 2.2. A Simple Circuit Diagram of a Fluorescent Lamp Placed in Series with an
Inductive Ballast.
Source: Mohan [ 6 ] ,p.379.

The efficiencyof the lamp can be hrther increased twenty to thirty percent by
using a high frequency electronics (solid-state) ballast which converts an incoming 60Hz
ac input to a higher frequency in the 25 to 40kHz range [6].
Under balanced conditions, all odd harmonics of the hndamental ac current are
produced by a series lamp and ballast combination. The actual magnitudes of these
harmonics depends greatly upon the manufacturer models of both the lamp and ballast
type. In particular, solid-state ballasts tend to result in higher amplitudes for each of the ac
current harmonics produced [12]. Depending upon the lamp and ballast combination used,
a third harmonic ranging from seven to eighty-seven percent of the hndamental ac current
has been measured [12]. Many other odd harmonics can be also quite significant in
amplitude. Typical third and fifth harmonic currents produced are 21 and 7 percent of the
fbndamental ac current respectively [l 11.
Even if the harmonics produced by individual lamps are small in amplitude, each of
the respective harmonics produced by the individual lamps tend to be in phase and
additive. This is due to the fact that these harmonics arise from distortion of a 6OHz
fbndamental current [13].
Arc fbrnaces, which are most often used for the meltdown of scrap metals, are a
source of a number of problems in the power system today. The use of arc fbrnaces is
expected to increase, since the technology is improving [14].
In an arc fbmace, graphite electrodes are lowered into a basin containing scrap
metal, in order to strike electric arcs between the electrodes and the scrap. The heat
generated by the electric arcs results in meltdown of the scrap metal [15].
Under balanced conditions, typical arc fbrnace third, fifth, seventh, and ninth
voltage harmonics produced are 20, 10, 6, and 3 percent of the hndamental respectively.
Unbalanced conditions can lead to significant even harmonics, and additional
magnification of the odd voltage harmonics [14].

It is noted that as the pool of molten metal grows, the arc becomes more stable.
This in turn results in much steadier currents, with much less distortion, and less harmonic
activity.
The harmonic problem in power systems today is also due to a change in design
philosophy. In the past, power system devices tended to be underrated or overdesigned.
Now, in order to be more competitive, power system devices are more critically designed
[4]. Because of this design philosophy, iron core devices such as transformers are more

likely to become saturated.


The most common curve used to describe the characteristics of the material used
for a transformer core is the B-H curve as shown in Figure 2.3 [16]. Figure 2.4 provides
an example of some actual B-H curves.
The B-H curve represents the relationship between the magnetic field (B), and the
magnetic field strength (H). When there is little change in the value of B for large changes
in H, the transformer is said to be saturated.
Modern high-voltage transformers with grain oriented steel cores saturate typically
somewhere above 1.0 to 1.2 times the rated magnetic flux in the core [17]. Magnetic flux
represents the surface integral of the magnetic field. In general, manufacturer ratings are
given somewhere around the point of saturation. Operation of the transformer beyond
these ratings drives the core into saturation, and results in rapidly increasing levels of
harmonic currents [I].
It is impractical to design transformers without saturation, and as a consequence,
harmonic currents produced by a transformer are unavoidable.
The significant harmonic currents produced by a transformer under balanced
conditions, are all odd harmonics of the hndamental frequency exciting current. The
magnitudes of these harmonic currents depends greatly upon the magnitude of the voltage
at the transformer terminals. However, typical magnitudes for the third, fifth, seventh, and

Fig. 1-1 0. Hysteresis loop; hysteresis loss is proportional to the loop area (shaded).

Figure 2.3. Typical B-H Curve used to Describe the Characteristics of the Material used
for a Transformer Core.
Source: Fitzgerald 1161, p.2 1 .

-10

10

20

30
40
H,A turndm

50 70 90 110130150170

Fig. 1-6. 6 - H loops for M-5 grain-oriented electrical steel 0.012 in thick. Only the top halves
of the loops are shown here. (Armco I ~ c . )

Figure 2.4. An Example of Actual B-H Curves.


Source: Fitzgerald [16], p. 15.

ninth harmonic currents produced by a saturated transformer are 50, 20, 5, and 2.6
percent of the fundamental exciting current respectively [18].
Even in the absence of other nonlinearities in the power system, transformer
harmonic currents can reach significant levels [19].
Transformer harmonics can be of great concern, since transformers are widely used
in power systems, and play an integral role in power transmission.
Considerable effort is taken to design rotating machines such that the harmonic
currents they produce are negligible [16]. Special care is taken to eliminate triplen (i.e., the
third harmonic and its integer multiples) and fifth and seventh harmonics of the
fundamental frequency exciting current, where other harmonics are taken to be
insignificant [20].
Saturation of the stator and rotor teeth can lead to the generation of odd
harmonics, but these are also generally insignificant in amplitude [21]. For example, rotor
third harmonics produced under saturated conditions, typically have values lower than one
percent of the rated stator current 1211.
Unbalanced operating conditions may also result in odd harmonic voltages at the
machine terminals of a generator. For example, a third harmonic voltage of magnitude
greater than six percent of the fundamental can occur. However, these harmonics are
greatly dependent upon the length of the line that the generator is feeding [22].

2.3 Detrimental Effects of Harmonics

Harmonics have been linked to a number of problems in the power system,


including communication interference. Very often, communication lines used for the
transmission of signals share the same path as power lines used for transmitting and
distributing electrical energy [20]. Current flowing in the power lines results in magnetic
and electric fields which induce currents (or voltages) in the nearby communication lines
[26]. The interference introduced depends greatly upon the separation distance between

the power and communication lines (which are often closely in parallel with each other),
and the frequency (particularly the harmonics after the fundamental) and magnitude of the
induced currents (voltages) [20].
The inductive reactance of the ac power system (composed of generators,
transmission lines, transformers, etc.) and the capacitive reactance of widely used
capacitor banks (for power factor correction), and insulated cables, produce resonance
conditions [25]. The resonant frequency of an inductive-capacitive (LC) circuit occurs
when the inductive reactance equals the capacitive reactance.
If the power system appears to be a very low impedance (for instance, to a
harmonic source) at the resonant frequency, then this condition is termed series resonance.
Likewise, if the system appears to be a very high impedance at the resonant frequency,
then this condition is termed parallel resonance. In either case, if the resonant frequency of
the power system happens to be close to one of the frequencies generated by a harmonic
source in the system, the result may be the flow of high harmonic currents (or the
appearance of high harmonic voltages) [26].
Although a system resonance condition does not produce harmonic currents or
voltages, small currents (voltages) generated by a harmonic source in the power system
can be amplified significantly by a resonance condition.
Since switching the configuration of the capacitor banks (for a different power
factor) changes the capacitive reactance of the system, more than one resonant frequency

will exist [27]. In addition, since motor loads appear to be primarily inductive at harmonic
frequencies, changes in the motor loads on the system, can result in a shift of the resonant
frequency [25].
In a three-phase system, triplen harmonics (i.e., the third harmonic and its integer
multiples) represents a special and rather complicated problem. The objective here will be
to keep the discussion as simple as possible, yet to introduce a couple of realistic examples
of how triplen harmonics cause problems in power systems. Therefore, consider the case
of a wye-connected load, that of a delta-connected load.
Three-phase power systems are operated under balanced conditions (as much as
possible), where the hndamental frequency line currents are 120 degrees out of phase
with respect to each other. For example, in a balanced three-phase system, the
hndamental frequency line currents can be:
I:) = I(') cos( Wt)

I;) = I(') cos(Wt - 120)


I:) = I(') cos(0t + 120)

The third harmonic of the line currents above (assumed to be generated by a nonlinear
device in the power system) would be:

I:) =

COS(~(W~))

= I ( ~C) O S ( ~ ( W ~ ) )

I?) = I ( ~C) O S ( ~ (-W1200))


~
= I ( ~COS(~(W~))
)

I;) = I ( ~C) O S ( ~ (+W1200


~ 1) = I ( ~COS(~(WC))
)
In a wye-connected load, these third harmonic currents would be in phase and additive in
the neutral wire. Ideally, it is desired that all currents should be zero in the neutral wire. If
significant enough in amplitude, these third harmonic currents would lead to increased
heating of the neutral wire, and therefore shorten its lifespan. The absence of a neutral
wire will result in circulating triplen harmonic currents in a delta-connected load. Again, if
significant in amplitude, these third harmonic currents will lead to increased heating of the
elements in the delta-connected load, and therefore shorten their lifespan.

Motors and generators (both induction and synchronous) experience increased


heating due to iron and copper losses at harmonic frequencies. Harmonics can cause
rotating machinery to have a pulsating torque output. This results in mechanical
oscillations which can lead to rotor shaR fatigue and accelerated aging of the shaft and
connected mechanical parts. In particular, harmonics can cause or enhance phenomena
called cogging (a refbsal to start smoothly) or crawling (very high slip) in induction
machines [25]. The net effect of harmonics on rotating machinery is a reduction in both
efficiency and life.
The major effects of harmonics on transformers is that current harmonics cause an
increase in copper losses and stray flux losses, and voltage harmonics cause an increase in
iron losses [25]. As a result, there will be an increase in transformer heating, which in turn
can shorten the life of the unit.
Harmonics flowing through transmission lines (power cables) will lead to
additional power losses due to the "skin effect". As a result, the effective alternating
current resistance (R,,), will be raised above the direct current resistance (R,,), especially
for large conductors. Therefore, when the current flowing through the cable contains a
high harmonic content, the equivalent RaCfor the cable is raised even higher, which will
amplifjr the 12Ra,loss as illustrated in equation (2.19). Note that superposition of power is
valid as long as all terms to be added, are at different frequencies. On the right hand side
of equation (2.19), each successive term is at a frequency which is an odd integer multiple
of the first.

"Hot spots" are created along the cable by maximums of the overall current and voltage
due to standing wave phenomena. It is at these points, that excessive damage to the cable
and its insulation can occur [42].
In order to measure the effective power drawn by industrial loads, watt-hour
meters are typically used. Depending upon the type of meter used, and the harmonics

present (and their magnitudes) in the current and voltage waveforms, a significant
overestimation (or possible underestimation) of the real power drawn by the industrial
load can take place [30]. Therefore, large industrial loads that are significant harmonic
producers could appear to have a higher (or lower) power factor than they actually do.
For example, induction watt-hour meters are designed to operate in the presence
of purely sinusoidal voltage and current waveforms, and are calibrated in this way [3 11.
The watt-hour meter will indicate the real (average) power drawn by sensing the voltage,
the current, and the phase relation between the voltage and current. However, in the
presence of harmonics, the true phase relationship between the distorted current and
voltage waveforms becomes unclear, thus leading to errors in the power readings [32].
Several other devices are sensitive to the presence of harmonics, such as relays and
computers. A relay is a device which will perform one or more switching actions based on
the information received from the power system, and is often used in power system
protection schemes [2]. Negative-sequence current relays are oRen used to detect small
fault currents [33]. This negative-sequence overcurrent relay can be used to trip a breaker
coil (to interrupt operation) when the negative-sequence component of the fault current
exceeds a certain value. The presence of high negative-sequence harmonic currents, may
cause this relay to operate for a fault current with a negative-sequence component that is
much lower than the specified value. In a balanced power system, the fiRh harmonic is
entirely negative-sequence.
In order to prevent malhnction or damage to a computer, manufacturers typically
speci@ that the voltage distortion (due to harmonics) is to be less than 5 percent of the
hndamental voltage [34].

CHAPTER 3 -- A NEWTON-RAPHSON BASED HARMONIC POWER FLOW

3 . 1 Introduction

A conventional power flow is a tool for determining the voltage, current, real and

reactive power, and power factor at various points in an electric power system under
existing or proposed conditions of normal operation [35]. Only hndamental frequency
voltages and currents are considered when determining these power system quantities. It is
necessary to use a conventional power flow to analyze a power system, in order to
determine the effects of new generation sites, of new transmission line locations, and
interconnections with other power systems. A conventional power flow is also essential
for assessing the impact of projected load growth and various load conditions, such as
peak and off peak.
When using a conventional power flow, the assumption is made that all elements
of the power system will not produce harmonics of hndamental frequency voltages and
currents. This includes the transmission network, and any devices attached to generator or
load buses. Conventional power flows do not allow the inclusion of harmonic devices,
which are power system elements which will distort hndamental frequency voltages and
currents.
A harmonic power flow is a tool which is used to predict the level and propagation

of harmonics created by one or more harmonic device in the power system, the effects
thereof, and the exploration of corrective measures such as the use of filters [I].
The main focus of this chapter is to describe the extension of a conventional power
flow as solved by the Newton-Raphson method, so that a harmonic analysis of the power
system may be performed. For both the conventional and harmonic power flows discussed

in this chapter, only power systems operating in a completely balanced three-phase ac


mode will be considered. If each phase of the three-phase power system contains identical
fundamental and harmonic voltages and currents, the power system is said to be
completely balanced [I]. A property of a completely balanced power system is that even
harmonic currents and voltages will not exist [20]. With completely balanced conditions
assumed, a single line admittance (impedance) diagram of the power system may be used
for each harmonic considered, corresponding to the correct sequence given by Table 3.1.
In addition, the admittances (impedances) of each single line diagram must be scaled
according to the harmonic frequency.
If unbalanced conditions exist, fundamental and harmonic currents can each
contain positive-, negative-, and zero-sequence components. The method of symmetrical
components must then be used to simplifL analysis, but will not be considered in this
thesis.
For this thesis, buses which do not contain harmonic producing devices will be
referred to as conventional buses. In addition, buses which do contain harmonic producing
devices will be referred to as harmonic buses.

Harmonic

Sequence

Harmonic

Sequence

1
3

Pwitive

19
21

Poaitive
Zero
Negative
Poeitive
Zero
Negative
Poeitive

7
9
11

13
15
17

Zero
Negative
Poeitive
Zero
Negative
Positive
Zero
Negative

23
25
27
29
31
eta.

Table 3.1. Sequence of a Fundamental Frequency Signal and its Harmonics in a


Completely Balanced Three-Phase ac Power System.
Source: Westinghouse Electric Corporation 1201, p.759.

3.2 Solving Nonlinear Algebraic Equations by Newton-Raphson Methods


Numerical iterative techniques are used to solve a nonlinear algebraic equation set,
which has no formal method for direct solution of the unknowns. A numerical iterative
technique progressively computes more accurate estimates of the unknowns, until results
are achieved with the desired accuracy in a finite number of iterations. If such results are
obtained, the solution set is said to converge [2].
In order to explain the Newton-Raphson method, and the care required to use this
method, consider the general equation below, which has one unknown (i.e., x ) .

f (x) = 0

(3.1)

Expanding f (x) in a Taylor Series about a known value x r , results in equation


(3.2). The value xr is referred to as the value of x in the rth iteration.

Discarding all the terms after the first derivative, results in the approximation
below.
df (x') (X - Xr)
f (x) E f (xl) - -

dx

Because f (x) = 0 , equation (3.3) is rewritten as equation (3.4), or equation (3.5).

Because of the error introduced by neglecting the higher order terms in the Taylor
Series, the value x is not a true root off (x) = 0 . However, it is quite possible for x to
give a better estimate of a true root off (x) = 0 than xr will. therefore, the next
approximation to f (x) = 0 is given by solving equation (3.4) for x = x"'.

dx
The Newton-Raphson method for solving equation (3.1) begins by selecting a
starting value xO,from which Ax is calculated by equation (3.5). If Ax > E, where

is a

small allowable error, x1 is calculated using equation (3.6). Once again, Axis calculated,
and the procedure continues until one of three things happen. First, a solution to f ( x ) = 0
is said to converge, if Ax I E. Second, the solution to f (x) = 0 is said to diverge, if
Ax

A,,

where A-

is an unacceptably large change in x . Third, the process can also

terminate if the number of iterations has reached a maximum number allowed, i.e.,
r 2 .r,

Figure 3.1 represents the geometrical interpretation of the Newton-Raphson


method for finding a value of x which makes f (x) = 0 . In mathematical notation,
equation (3.3) is the equation of the tangent line at xr [36]. The point at which this
tangent line intersects the x axis (i.e., xr") is the point at which a new tangent line is
drawn on the curve of y = f (x) . Starting with the initial value of x r in Figure 3.1,
successive tangent lines will ultimately lead to the solution xs. The value xs is a value of
x which satisfies equation (3. I), i.e., f ( x ) = 0 .

However, if iteration xr is close to a minimum (or maximum) of a fbnction f (x), it


is quite possible that the next approximation to f (x) = 0 , will be far away from the
desired solution xs. This problem is illustrated in Figure 3.2, and if the solution does
eventually converge to xs, additional iterations are required. However, the solution may
converge to a more remote root, or may not converge at all. The Newton-Raphson
algorithm works best, when the initial guess for x is very close to the desired root of
f(x) = 0 .
A system of equations that are to be solved by the Newton-Raphson method, may

be unstable (ill-conditioned). A system of equations is called stable if relatively small


changes in the coefficients result in correspondingly small changes in the solution vector.

Figure 3.1. Geometrical Interpretation of the Newton-Raphson Method.


Source: Choma [37], p.3 14.

Figure 3.2. An Example of How the Next Approximation to f ( x ) = 0 can be Far Away
From the Desired Solution x s , if Iteration x r is Close to the Minimum of a Function f ( x ) .
Source: Choma [37], p.315.

For two systems, Ax = C and By = D, if the elements of A and B differ by little, and
those of C and D differ by little, then the elements of the solution vectors x and y will
also differ by little if the system is stable. As an example of an unstable system, consider
the equation sets (3.7) and (3.8) below.
x+y=l

(3.7a)

x + 1.00001y = 0

(3.7b)

x+y=l

(3.8a)

x + 0.99999~= 0

(3.8b)

Although these two systems above differ by little, the solution set for equations
(3.7) is (x = 100,001 and y = -100,000), and the solution set for equations (3.8) is
(x = -99,999 and y = 100,000). Each equation set represents an effort to find a position
(x,y) at which the two lines will intersect, as shown in Figure 3.3. Therefore, a slight shift
of either line, can move the point of intersection quite significantly.
The above problems with using the Newton-Raphson method are not necessarily
separable, and may occur simultaneously.
Therefore, the Newton-Raphson method for solving nonlinear algebraic equation
sets is not perfect, and one needs to be aware of how this method can fail for the problem
to be analyzed.
For n nonlinear algebraic equations with N unknowns as in equation (3.9) below,
the method of Newton-Raphson may be used to solve the equation set if n 2 N.
fr

= f(Ar,Br ,...........,N r ) = O

gr = g(Ar ,B',

............,N')

=0

;+ 0.99999~= 0

Figure 3.3. Illustration of the Two Systems Given by Equation Sets (3.7) and (3.8).

Using these equations, a Jacobian matrix

IJ")I

is formed by equation (3.10).

Numerical values for the partial derivatives are obtained by using the values of the
unknowns at the rth iteration.

The error in the unknowns, AA, AB,......., AN is found by equation (3.1 1). The
column matrix on the right side of equation (3.1 I), contains numerical values for equation
set (3.9) at the rth iteration.

The next approximation to the roots of equation set (3.9) is obtained from
equation (3.12).

The iteration process on the unknowns continues until IAAl I


....,and IAN1 5 E

~ unless
,
the

E,,

IABI I E,,

solution set diverges or the number of iterations has reached

the maximum number allowed.

3.3 - A Brief Derivation of a Newton-Raphson Based Conventional Power Flow


The equations for a Newton-Raphson based Conventional Power Flow (or
Conventional Power Flow) are derived by first observing the hndamental frequency
current flow at bus i of a general n -bus power system. The power system is shown in
Figure 3.4. According to Table 3.1, only the positive-sequence-impedance diagram is
required to properly represent this power system, as shown in Figure 3.5.
Using Kirchhoff s Current Law (KCL) at bus i , results in equation (3.13).

Taking the conjugate of equation (3.13), and multiplying through by the


-

hndamental frequency voltage at bus i , ~

( l=
)

~") / 4results
" , in equation (3.14) or its

equivalent, equation (3.15).

(fz))*

=~

~ ( 1 )

S,(1) -

Li

(7~+
) )~ ~( 1(?$I)*
)

( 1 )

+ f(1)
7i

(3.14)
(3.15)

The terminology of equation (3.15) is explained below.


-

s,")= The hndamental frequency three-phase complex generated power flowing into the
ith bus from the power source(s).
-

s:)

= The hndamental

frequency three-phase complex power flowing out of the ith bus

towards the load buses.

3;) = The hndamental frequency three-phase complex power flowing out of ith bus
towards the transmission network.
The individual hndamental frequency complex powers of equation (3.15) can be
resolved into their rectangular components as shown below.
-

s,(!)

p(1)
+ jQ(1)
Gi
Gi

s;)
= p(1) + jQ("
Li
xi

(1)

- p(1) + jQ(1)

STi -

Ti

Ti

Bus i

Transmission

T):
Generat ion

Figure 3.4. Single-Line Diagram at Bus i of a General n-Bus Power System.


Source: Gross [2], p.257.

Figure 3.5. The Equivalent Positive-Sequence-ImpedanceDiagram of Figure 3.4.


Source: Gross [2], p.257.

By inspection of equations (3.19, (3.16), (3.17), and (3.18), equations (3.19) and (3.20)
are formed.

P$
(1)

= pi) + f$')

(3.19)

(3.20)

(1)

Qa - Qfi + Q
':

A simple rearrangement of equations (3.19) and (3.20) results in the equations to


be written at each bus i in an n-bus power system for the Conventional Power Flow.
and

el)

a'')represent the net fundamental frequency real and reactive power flows at each bus

i.
= -P$)+

el)=

e)+ P:)

=0

(3.2 1)

+ Q;) + Q;) = 0

(3.22)

-&)

Depending upon the bus type, the terms

e)and Q$/ will be known or unknown

values in the Conventional Power Flow. Their final values will fall between a prespecified
minimum and maximum value for each.

q<," and Q;)

are always given or known values at each bus in the power system,

since Conventional Power Flow calculations are made at specified load conditions.
The terms

I$"

and Q;) are obtained by first rearranging the fundamental-frequency

positive-sequence power system network as in Figure 3.6. A fundamental frequency


admittance matrix

1 Po1 is then formed by writing KCL equations for the fundamental

frequency transmission current

f:)

entering every bus i using nodal analysis. Only the

transmission line and transformer fundamental frequency admittance values will be used to
create this matrix.
As an example of forming an admittance matrix, consider the simple fundamental
frequency admittance diagram of a two-bus power system, as in Figure 3.7. By using KCL
at each bus, equations (3.23) and (3.24) are formed.

1;) = (7:)

7;)

)~

( 1 )- (jjjl))q l )

(3.23)

Fundamental frequency
positive-sequence
transmission network,
which only contains
transformers and
transmission lines. All
loads and generation
are external.

Figure 3.6. General n -Bus Power System Arranged to Form the Fundamental Frequency
Admittance Matrix.
Source: Gross [2], p.258.

reference
Figure 3.7. A Fundamental Frequency Admittance Diagram of a Two-Bus Power System.
Source: Gross [2], p.259.

Equations (3.23) and (3.24) can be written in matrix form, to obtain equations (3.25),
(3.26), (3.27), and (3.28).

Every entry of the matrix IF" is in the general form of 7;:) = y F / a : )

Equation (3.18) may now be written in terms of the transmission current j:);",
which was formed when creating

1 F' 1.

Equation (3.18) above may be separated into its rectangular components, as shown
below.

Q
:
) =C

vl)~

q)

" sin y(8')


-~

at))

Therefore, the nonlinear algebraic equations that are to be written at each bus i in
the Conventional Power Flow, and solved using the Newton-Raphson technique, are
formed by rewriting equations (3.21) and (3.23) as shown below.

The above two equations describe the real and reactive power conditions at each
bus i in the power system. They will be referred to as the real and reactive power
mismatch equations.
The known and unknown variables in equations (3.3 1) and (3.32) depend upon the
bus type. As stated before,

e'and Q;) are always prespecified numerical values at every

bus in the power system. At the swing (or slack bus), the hndamental frequency voltage is
assumed to be

ql)= 1/@ p.u. as a reference, while e) and Q$ are the unknowns.

However, equations (3.31) and (3.32) do not need to be written at the slack bus, since

buses, both

E)and Q;)

e)and Q:,

ql)and 4"are known. At all load


are known and usually zero for each, while vl'and 8''are

there are no restrictions placed upon

and

unknown. Lastly, at generator (voltage-controlled) buses,

8'' and):Q

are unknown.

P<," and

vl)are known, while

3.4 A Detailed Derivation of a Newton-Raphson Based Harmonic Power Flow

Since nonlinear components are the source of harmonics in power systems, it is of


great importance in the Newton-Raphson based Harmonic Power Flow (or Harmonic
Power Flow), to accurately model the nonlinear relationship between the current and
voltage waveforms at each bus containing a harmonic device. For this purpose, it is
convenient to express distorted voltages and currents as Fourier Series. For the Harmonic
Power Flow, the Fourier Series of the current entering a bus containing a harmonic device
is expressed as a hnction of the Fourier Series of the voltage at this bus, and of any
parameters which describe this distorted current waveform. As an example of parameters
which describe the current entering a harmonic producing device, consider the typical acside line current waveform, for an ac current to dc current converter, as shown in Figure
3.8. The current i, represents the fundamental frequency component of the total current

i,. Briefly, the delay angle a describes the angle at which the line current i, lags the lineto-neutral voltage v,. Also, the commutation interval p represents the interval through
which i, goes from a value of zero to its maximum value. For the Harmonic Power Flow,
both a and p generally represent additional unknowns that must be found during the
iteration process. For the harmonic devices known to be modeled in the Harmonic Power
Flow, no more than two additional parameters must be found at each bus containing a
harmonic device [I].
Conventional buses can be modeled exactly the same way as harmonic buses are
modeled in the Harmonic Power Flow. This approach, however is generally not used in
order to simpli@the Harmonic Power Flow. Instead, each conventional bus is usually
modeled as a complex power demand at the fundamental frequency, and is modeled as an
impedance (that has a linear voltage-current characteristic) at the harmonic frequencies. At
each harmonic frequency, a different impedance is used to model the device(s) at the
conventional bus. This is the approach which is assumed in reference [I].

Figure 3.8. Typical ac-Side Line Current Waveform of an ac Current to dc Current


Converter.
Source: Mohan [ 6 ] , p.57.

However, the following derivation will show that conventional and harmonic buses
are basically treated the same way in the Harmonic Power Flow, even though the difficulty
of their bus models may not be the same. The goal here is to make this point as obvious as
possible for two reasons. First, once the simplified conventional bus model is understood,
the more detailed harmonic bus models will be easier to understand. Second, it will be
easier to understand how improved conventional bus models may be incorporated into the
Harmonic Power Flow.
Following this derivation, simplifications to the Harmonic Power Flow equation
set will be discussed. These simplifications are possible only when the simplified
conventional bus model (as discussed above ) is used.
In order to solve for the hndamental frequency unknowns in an n -bus power
system, 2 ( n - 1) equations are required. Like the Conventional Power Flow, there is no
need to solve for any hndamental frequency unknowns at the slack bus, since ql' and
are known, and because there are no restrictions placed upon

q'

el)and Q:!.

Also, 2(n)(q) equations are required to solve for every harmonic unknown in an

n -bus power system, when q harmonic frequencies are of interest. For this thesis, all
frequencies of interest greater than the hndamental, are referred to as the harmonics.
Up to two equations may be required to solve for the parameters which are used to
describe the distorted current waveform entering each harmonic producing bus m .
Therefore, for M harmonic buses, up to 2 ( M ) additional equations may be required for
the Harmonic Power Flow. Therefore, the total number of equations which may be
required to solve for all of the unknowns in the Harmonic Power Flow is

2 ( n - 1) + 2 ( n ) ( q )+ 2 ( M ) .
A brief description of the Harmonic Power Flow equations is now given, before
explaining them in further detail.
First, 2(n - 1) equations are provided by describing the real and reactive power
conditions at each bus in the power system, as a Fourier Series. Like the Conventional

Power Flow, these equations will be referred to as the real and reactive power mismatch
equations in the Harmonic Power Flow. These equations will allow complex power

(generated and load) to be specified at each bus in the power system, exactly as was
desired with the Conventional Power Flow.
As will be shown, a convenient way to obtain the additional 2(n)(q) + 2(M)
equations required for the Harmonic Power Flow is to apply Kirchhoff s current law
(KCL) at every bus in the power system.
Applying KCL at every bus j in the power system for each harmonic frequency of
interest, will result in (n)(q) equations. Separating each of these equations into its real and
imaginary components, results in 2(n)(q) equations for the Harmonic Power Flow. These
equations will be referred to as the harmonic frequency real and imaginary current
mismatch equations.
Lastly, 2 ( M ) equations are obtained from the real and imaginary components of
the fundamental frequency KCL equation formed at each harmonic bus. These equations
will be referred to as the fundamental frequency real and imaginary current mismatch
equations.
Like the Conventional Power Flow, the real and reactive power mismatch
equations will be formulated such that all complex power

3 will be treated as either

generation, transmission, or load complex power. Although these equations will contain
the hndamental and all harmonic frequency components, generated complex power will
be assumed to contain no harmonic components. Generators (synchronous machines) are
generally not considered to be significant harmonic producing devices, and are therefore
not modeled as harmonic sources in the Harmonic Power Flow [I].
Generators and harmonic devices both fulfill important, yet different roles in the
Harmonic Power Flow. Therefore, unnecessary complications may be avoided, if
generators and harmonic devices are placed at different buses in the power system
impedance diagrams. In addition, it will generally be easier if different types of harmonic

devices are placed at separate buses. For example, a gas discharge lighting load is not
treated the same as an ac current to dc current converter, in the Harmonic Power Flow.
When necessary, a single bus may be separated into two buses connected by a short circuit
(or a very small impedance). Figure 3.9 provides a simple example of how a single bus
containing a generator and a harmonic producing device, can be separated into two buses
connected by a short circuit.
Therefore, the general form of the real and reactive power mismatch equations at
every harmonic bus m are given by equations (3.33) and (3.34). The letter h refers to the
highest harmonic to be considered in the Harmonic Power Flow, and the letter f
represents the frequency of interest (fbndamental or harmonic).

In addition, the general form of the real and reactive power mismatch equations at
every conventional bus i are given by equations (3.35) and (3.36).

Only odd terms appear in equations (3.33) through (3.36) above, since even
harmonics do not exist in a completely balanced power system. Therefore, f 2 1,f odd.
The slack bus in the Harmonic Power Flow is treated essentially the same as for
the Conventional Power Flow. The hndamental frequency voltage is assumed to be

V''' = l / g p . u . as a reference, while El)and Qz are the unknowns. However,


equations(3.35) and (3.36) do not need to be written at the slack bus, since there are no
restrictions placed upon

and Q!/,

and

v)and 4')are known.

Figure 3.9. (a) A Bus Containing a Generator and a Harmonic Producing @P)
Device. (b)
An Equivalent Representation of (a), Separated into Two Buses Connected by a Short
Circuit (SC).

Generator (voltage-controlled) type buses in the Harmonic Power Flow, are also
treated essentially the same as for the Conventional Power Flow. Although Pi and Q,are
written as a Fourier Series at generator buses,
values, where

v)= ygmW,and 4') and Q):

and

vl)are still taken to be known

are unknown, where Q


): must fall between

a prespecified minimum and maximum value. If Q


): is not within these limits, it must be
set at the appropriate limit. For example, if Q:) > Q:LOX at the present iteration, Q$) must
): = EL,
be set to Q

again. Therefore,

This would require

vl),and Q):

vl)to become a variable at the generator bus

would exchange roles as knowns and unknowns. This

would be equivalent to changing the generator bus to a conventional load type bus. Note
that if

rises above the original fixed value

($deCIped) at this bus when Q):

= #LOX,

voltage control is regained, and the bus becomes a generator type again.
The transmitted real and reactive power terms in equation (3.33) through (3.36),
are obtained in the exact same way for both the fbndamental and harmonic frequencies. In
addition, these terms are formed exactly the same way as for the Conventional Power
Flow, and are to be used at every busj in an n -bus power system (except for the slack
bus). Therefore, the transmitted real and reactive power terms in equations (3.33) through
(3.36), are obtained from equations (3.37) through (3.40). Again, f 2 1,f odd.

In the above four equations, the "ymjIt, I yv It, a(?


, It, and It a:?

I'

terms represent

the magnitude (i.e.,y) and the angle (i.e.,a) of an element, of either the fbndamental or a

harmonic admittance matrix. The fundamental frequency matrix

IF"1, for the Harmonic

Power Flow is identical to the fundamental frequency admittance matrix obtained for the
Conventional Power Flow.
When forming the kth harmonic frequency admittance matrix lFk)I,it must be
noted that the power system representation will change for each harmonic frequency
considered. The correct sequence admittances (impedances) must be used according to
Table 3.1, and scaled with frequency. Also, the power system must be arranged as in
Figure 3.10. A comparison of Figure 3.10 with Figure 3.6, shows that only the
transmission network (composed of transmission lines and transformers) is used for
obtaining an admittance matrix at any frequency. Like the fundamental frequency
admittance matrix, the kth harmonic frequency admittance matrix is obtained by writing a
KCL equation, at every bus j using nodal analysis. However, in equations (3.33) through
(3.36), the terms

and Qg) ( j = i or m) are unknown for every harmonic frequency,

$)

and must be found during the iteration process.


For the Harmonic Power Flow, the Fourier Series of the current entering a
harmonic bus m is expressed as a function of the Fourier Series of the voltage at this bus,
and of any parameters which describe this distorted current waveform. Assuming this
Fourier Series has been obtained, the kth harmonic complex load power at bus m is
obtained by observation of Figure 3.1 1, to give equations (3.41) and (3.42).

P',k) -- Vk)g,(4cos(6',k' - p',kl)

(3.4 1)

- vWg2 Sin(&*)
elnt- p',"')

(3.42)

(k)

#,I

The voltage

rn

= v:)/$,"),
represents the kth harmonic voltage at bus m The current
7

~ 2= )gE/fl,"),
- represents the kth harmonic current entering the harmonic load bus m .
Due to the complexity involved in obtaining the Fourier Series for the current
entering a harmonic producing device, only one harmonic load will be discussed in detail

The kth harmonic


fiequency transmission
network, which only
contains transformers
and transmission lines.
The correct sequence
for the impedance
(admittance) is chosen
from Table 3.1. All
conventional and
harmonic loads and
generation are external.

Figure 3.10. General n -Bus Power System Arranged to form the krh Harmonic
Frequency Admittance Matrix.
Source: Gross [2], p.258.

Figure 3.1 1 . A General Representation of a Power System at the kth Harmonic


Frequency, with a Harmonic Producing (HP) Load at Bus m .

for this thesis.


As stated before, conventional buses can be modeled exactly the same way as
harmonic buses are modeled in the Harmonic Power Flow. This approach, however is
generally not used, in order to simplifjr the Harmonic Power Flow. Instead, each
conventional bus is modeled as a complex power demand at the hndamental frequency,
and is modeled as an impedance (that has a linear voltage-current characteristic) at the
harmonic frequencies. At each harmonic frequency, a different impedance is used to
model the device(s) at the conventional bus. These harmonic impedances are easily
estimated, and are discussed for several conventional devices in this thesis. Therefore, it is
easy to obtain the harmonic currents entering a conventional bus, and to obtain the
harmonic complex power drawn by a conventional bus.
For example, Figure 3.12(a), shows a general representation of a power system at
the kth harmonic frequency, with a conventional (C) load at bus i. The current

ZEI?'= gE'/pj*',
- represents the kth harmonic current entering the conventional load at bus
i. The kth harmonic voltage at bus i is represented as y"' = ~ ( ~
' / Assuming
4 ~ ) ' . that the
i

kth harmonic impedance model at conventional bus i is a series R-L combination as in


Figure 3.12(b), the current

g'=

g:'

is found by Ohm's law to give equation (3.43).

y(k)

R+j2@

By substitution of equation (3.43) into

(3.43)

f = V I', at the kth harmonic frequency, the

harmonic complex load power for the conventional load of Figure 3.12(b) is given by
equations (3.44) and (3.45).

bus i

1:'
4
7

bus i
j$)
4

Figure 3.12. (a). A General Representation of a Power System at the kth Harmonic
Frequency, with a Conventional (C) Load at Bus i . (b). Equivalent Representation of (a),
with the Conventional Load Modeled as a Series R-L Combination.

The additional 2(n)(q) + 2(M) equations required for the Harmonic Power Flow,
are conveniently obtained by applying KCL at every busj in the power system. As stated
before, applying KCL at every bus for each harmonic frequency of interest, will give
(n)(q) equations. Separating each of these equations into its real and imaginary
components, results in 2(n)(q) equations for the Harmonic Power Flow. These equations
are referred to as the harmonic frequency real and imaginary current mismatch equations.
Lastly, 2(M) equations are obtained from the real and imaginary components of
the hndamental frequency KCL equation formed at each harmonic bus. These equations
are referred to as the hndamental frequency real and imaginary current mismatch
equations.
By observation of Figure 3.1 1, application of KCL at a harmonic bus m for any
frequency of interest (hndamental or harmonic), results in equation (3.46), where f

1,

and f is odd ordered.

Also, by observation of Figure 3.12(a), application of KCL at a conventional bus


i , for any harmonic frequency of interest, gives equation (3.47), where k 2 3, and k is

odd ordered.

(j= i or m) is simply a KCL equation that is written at every bus j

The current

in the power system, using nodal analysis. These KCL equations are then used to produce
the hndamental (f = I), or harmonic (f > 1,f odd) frequency admittance matrix /PI.
Therefore, equations (3.46) and (3.47) can be written as equations (3.48) and (3.49)
respectively. Again, f

1,k 2 3 and f and k are odd ordered.

j=l

Separation of equations (3.48) and (3.49) into their real and imaginary
components, results in equations (3.50) and (3.51), where f

2 1,k 2

3 and f and k are

odd ordered.

(3.5 la)

To summarize, the equations to be used in the Harmonic Power Flow are


discussed below.
The general form of the real and reactive power mismatch equations to be used in
the Harmonic Power Flow, are given by equations (3.33) through (3.36). Equations (3.37)
through (3.45) help define the terms in equations (3.33) through (3.36).
Equations (3.46) through (3.5 1) represent the discussion of the hndamental and
harmonic frequency real and imaginary current mismatch equations. However, only
equations (3.50) and (3.51) need to be used in the Harmonic Power Flow.
Therefore, the general form of the Harmonic Power Flow equation set is given by
equations (3.33) through (3.36) and equations (3.50) and (3.5 1). This equation set was
derived in a general format, and is to be treated the same as any other nonlinear algebraic
equation set, that is to be solved by the Newton-Raphson method.

3.5 Network Models for the Newton-Raphson Based Conventional and Harmonic Power
Flows

Introduction
This section describes how to model a power system elements for the NewtonRaphson based Conventional and Harmonic Power Flows. The power system transmission
network composed of transmission lines and transformers, must be modeled at the
fundamental and harmonic frequencies. Harmonic frequency shunt impedance models, for
the most common conventional loads, such as synchronous machines and induction
motors is presented here.
The harmonic load model for a gas discharge lighting load is discussed. All
harmonic devices cannot be modeled in the same way, however, since the current entering
each different type of harmonic load will possess a different Fourier Series.
For the Conventional Power Flow, all impedances are positive-sequence.
However, for the Harmonic Power Flow, harmonic impedances may be positive-,
negative-, or zero-sequence, depending upon the harmonic fi-equency. The harmonic
frequencies presented in Table 3.1, apply to all power system network and device
impedances discussed here.

Transmission Lines
For the Harmonic Power Flow, three-phase transmission lines are modeled by a
single-phase pi-equivalent with the correct phase sequence, for both the fundamental and
harmonic frequencies. This model is also used for the Conventional Power Flow. The most
important factors to be considered for the pi-equivalent model in Figure 3.13, are line
length and skin effect. The "long line" pi-equivalent model in Figure 3.13 is recommended
for distances longer than five percent of the wavelength A of the highest harmonic of
interest[2] Knowing that A = c/ f , where c is the speed of light ( c = 3.00 x lo8mls), and

z,"=

sinh(@) ohms

RC" :Series resistance per unit length.


L~ :Series inductance per unit length.
d":Shunt capacitance per unit length
@:shunt conductance per unit length.
d:Length of the transmission line.

Figure 3; 13. Pi-Equivalent Model of a Long Transmission Line.


Source: Grady El], p.24.

assuming the fundamental frequency is 60Hz, (0.05)iE = 250km = 155.25miles.


Therefore, this critical length d is given by equation (3.52), where h is the highest
harmonic of interest.
d = (250lh)km = (155.251h) miles

(3.52)

For the 25th harmonic, the critical length d is approximately 6 miles. For distance shorter
than the critical length d , the "short line" pi-equivalent model of Figure 3.14 is adequate.
Cascaded sections of short line models may be used in order to model a long line. When
modeling a short line, a simple series impedance is generally sufficient for the fundamental
and lower harmonic frequencies (such as the 3rd, 5th, and 7th harmonics).
Skin effect is the tendency for an ac current to flow near a conductor's surface.
Skin effect increases the effective resistance of a conductor, as the frequency of the
current increases [2]. Very often, transmission line conductor tables present the resistance
at standard frequencies of 25, 50, and 60Hz. In order to obtain the resistance of the
conductor at other frequencies, equation (3.53) can be used [20].
Rac = A(R,,) ohms/mile

(3.53)

where:
Rac = The ac resistance of the conductor at the desired frequency in cycles per second.
R,,
A

= The

dc resistance of the conductor at any known temperature.

= Scaling factor given

in Table 3.2.

In Table 3.2, A is given as a function of B, as given by equation (3.54).

where:

= The

frequency in cycles per second.

= The

permeability of the conductor, which is taken to be a value of 1.0 for non-

magnetic materials.
Rmi, = The dc resistance of the conductor in ohms/mile.

Z,C" = ( p+ j d u ) ) d ohms

Em = j&d

mhos'

'Shunt conductance ignored

Figure 3.14. Pi-Equivalent Model of a Short Transmission Line.


Source: Grady [I], p.25.

Table 3.2. Skin Effect Table.


Source: Westinghouse Electric Corporation [20], p.53

Positive- and negative-sequence currents have no earth return components.


Therefore, the positive- and negative-sequence impedances of transmission lines can be
calculated without consideration to earth return paths for current. However, the return
paths for zero-sequence currents, are the earth and the tower ground wires. Therefore,
earth and ground wire paths can be very important when determining the zero-sequence
impedance of a transmission line. A number of different factors will effect the zerosequence current return paths, including tower geometry, wire conductivity, and the
number of ground wires. The equations which include these effects when calculating the
zero-sequence impedance of a transmission line, are called Carson's equations [20].
In the Appendix, the necessary equations are given, to calculate the positive-,
negative-, and zero-sequence impedances of the 230Kv transmission line in the Examples
chapter.

Transformers
Transformers in the Harmonic Power Flow, are modeled in the same way for both
the fbndamental and harmonic frequencies. This single-phase model is shown in Figure
3.15, and is the same standard model used in the Conventional Power Flow [2], where the

shunt magnetizing branch is ignored. The positive-sequence impedance is assumed to be


equal to the negative-sequence impedance. In a three-phase power system, the zerosequence impedance is equal to the positive-sequence impedance if three interconnected
one-phase transformers are used. However, zero-sequence impedance will be different if
the transformer is a single three-phase device. In addition, the zero-sequence model
depends upon the transformer connections (i.e., wye-connected, delta-connected, etc.)
The fbndamental frequency inductive reactance to resistance ratio (i.e., x"'/R"'
ratio) for each sequence is conservatively assumed to remain constant and is scaled
directly with frequency [25]. To the author's knowledge, no model is yet available,
whichallows the transformer to be treated as a harmonic source for the Newton-Raphson

Figure 3.15. Power Flow Model of a Transformer.


Source: Grady [I], p.30.

based Harmonic Power Flow.


As suggested by Figure 3.15, tap changing transformers must be used when

necessary, such as for modeling phase shift. For either wye-delta or delta-wye
connections, the high voltage side leads the low voltage side by thirty degrees for the
positive-sequence networks. Likewise, for the negative-sequence network, high voltage
side lags the low voltage side by thirty degrees. The zero-sequence network will introduce
no phase shift.

Generators
For each harmonic frequency, an acceptable model for a synchronous generator is
to directly scale the hndamental frequency negative-sequence inductance reactance of the
generator with frequency [I]. The resistive component of the hndamental frequency
negative-sequence impedance is generally much smaller in comparison to the reactive
component. In addition, in the absence of elaborate information, the resistive component is
assumed to remain constant for all harmonic frequencies, and negligible.
Negative-sequence hndamental frequency stator currents rotate at twice
synchronous speed, as seen from the rotor. The resultant flux is forced into paths of low
permeability, which do not link any rotor circuitry. These paths are characterized by the
direct and quadrature subtransient inductances L; and L:, respectively. By definition, the
hndamental frequency negative-sequence reactance of a synchronous machine is given by
equation (3.55) [20].

x:"= [(xi)"'
+ (x;)"']/2
where:

(xi)'"= Direct axis subtransient reactance


(xi)"'= Quadrature axis subtransient reactance

(3.55)

Then, by observation of equation (3.59, the findamental frequency negativesequence inductance of a synchronous machine is given by equation (3.56). This is the
inductance met by findamental frequency negative-sequence currents flowing into the
stator winding of a synchronous machine.

iq)= [(L; )"' + (~:)"']/2

(3.56)

When harmonic currents flow from the network into the stator windings of a
synchronous generator, they also create a flux rotating at multiples of synchronous speed,
where the direction may be the same or in opposition to the rotor's direction of rotation
[38]. In addition, these fluxes are also forced into paths characterized by the subtransient
inductances. Therefore, the average inductance which appears to be met not only by
negative-sequence fbndamental frequency stator currents, but also by harmonic frequency
stator currents is given by equation (3.56) [I].
Then, the reactance used to model the synchronous machine as in Figure 3.16, for
the kth harmonic frequency is given by equation (3.57), where k is the harmonic frequency
of interest, and

x!i

x!) is given by equation (3.55).

k q

(3.57)

Induction Motors
The induction machine is one of the most commonly used motor loads in power
systems today. An adequate harmonic frequency model of the induction machine is the
same one proposed for the synchronous machine [l]. However, if the direct axis and
quadrature axis subtransient reactances are unknown, then a harmonic impedance model
can be obtained, based on the per-phase fbndamental frequency model shown in Figure
3.17 [I].
The magnetizing inductance L, is generally ignored, since it is large in comparison
with the other terms.

Figure 3.16. Harmonic Power Flow Model of a Synchronous Machine for the kth
Harmonic Frequency.
Source: Grady [I], p.35.

S)T':

=The per - phase fundamental frequency stator voltage.

Tz) =The per - phase fundamental frequency stator current.


r, =The resistance of the stator winding.
L, =The leakage inductance of the stator winding.
L, =The magnetizing inductance
L, =The leakage inductance of the rotor winding.
r, =The resistance of the rotor winding.
s =The slip =(a,- o r ) / @ ,
w =The angular frequency in radians per second.

Figure 3.17. A Per-Phase Fundamental Frequency Equivalent Model for an Induction


Machine (Phase a Shown).
Source: Grady [I], p.33.

When harmonic currents flow from the network into the stator windings of an induction
machine, the resultant flux will rotate in the air gap at a speed of o, = ko,, where the
direction may be the same or in opposition to the rotor's direction of rotation [6]. As an
approximation, the rotor speed or is assumed to be roughly the same as the fbndamental
frequency, synchronous speed o,. Therefore, using the definition of slip given in Figure
3.17, the rotor slip relative to the (fundamental frequency) synchronous speed at the kth
harmonic frequency is approximately given by equation (3.58).
S,

= ( a , +w,)/w, z ( k o , +o,)/ko, ~ ( k l)/k-1


f

(3.58)

Therefore, the term q /s in Figure 3.17 is assumed to be approximately rr for all


harmonic frequencies. At the fbndamental frequency, the rotor and stator resistances are
generally small in comparison to the reactances. Therefore, this approximation will not
introduce significant error.
The resulting combined series resistance and reactance can be determined the
fbndamental frequency active and reactive power demand of the motor. Therefore, the
values of

and L: are obtained by equation (3.59) below, and are used in the

Harmonic Power Flow model of Figure 3.18. Note that if Q:) is negative, then a
capacitance

62 should be used instead of an inductance L;.

In the absence of elaborate information, the fbndamental frequency resistive


component is assumed to remain constant for all harmonic frequencies [I]. The
fbndamental frequency reactance is scaled directly with frequency.
Therefore, the impedance used to model the induction motor at all harmonic
frequencies is given by equations (3.60) and (3.61), where k is the harmonic frequency of
interest.

Figure 3.18. Harmonic Power Flow Model of an Induction Motor for the kfh Harmonic
Frequency.
Source: Grady [I], p.36.

Other Conventional Loads


It is difficult to represent many conventional load buses at harmonic frequencies,
since their exact composition is usually unknown. It is suggested, that in the absence of
specific information, a conventional load bus must be modeled as a shunt resistor in
parallel with an inductor or a capacitor [I]. The resistance and the reactance can be
determined by the hndamental frequency active and reactive power demand of the
conventional load. Therefore, the values of

R ~ and
L LS~,$

are obtained from equations

(3.62), and (3.63) below, and are used in the Harmonic Power Flow model of Figure
3.19. Note that if Qz' is negative, then a capacitance

cL should be used instead of an

inductance LS~,;.

In the absence of elaborate information, the hndamental frequency resistance is


assumed to remain constant for all harmonic frequencies [I]. The hndamental frequency
reactance is scaled directly with frequency.
Therefore, the impedance used to model an unknown conventional load at all
harmonic frequencies is given by equations (3.64) and (3.65) below.
~ f ;=

REL

E
L = kX(l)

load

(3.64)
(3.65)

Figure 3.19. Suggested Power Flow Model of an Unknown Conventional Load for the
kth Harmonic Frequency.
Source: Grady [I], p.37.

Gas Discharge Lighting Loads


The following harmonic load model, has been used to approximate the
characteristics of gas discharge lighting loads (such as fluorescent, neon, mercury arc, and
high pressure sodium) [l 11.
The Fourier Series of the total current entering a gas discharge lighting load is
assumed to take on the general form given by equation (3.66). The letterf refers to the
frequency of interest (fbndamental or harmonic) and the letter h refers to the highest
harmonic of interest.

Also, the Fourier Series of the voltage waveform measured at this bus is assumed
to have the general form given by equation (3.67).
V, ( t ) =

v z ) sin(fwot +

c))

In equations (3.66) and (3.67), f and h are odd, and f 2 1 and h 2 3


The proposed model for this harmonic load is the power series expansion given by
equation (3.68).

In the above equation, both d and w are odd, where d represents a polynomial power,
and w represents the highest polynomial power to be used in the expansion.
Equation (3.68) represents a polynomial series of odd powers of the voltage given
by equation (3.67). Even powers of the voltage may be excluded, if a completely balanced
power system is assumed. In order for the current (given by equation (3.66)) to be a true
hnction of the voltage ,the polynomial coefficients bd need to be obtained by curve fitting
the current versus voltage using measured data. In general, the curve fit obtained is only
valid for the particular loading conditions in the power system, at which the measurements

were taken. The term " B" in equation (3.68) is simply a scaling variable which is
multiplied across the entire equation.
The Fourier Series expansions for odd powers of voltage is completely given by
equation (3.69). Note that equation (3.69) is simply a shortcut method of raising equation
(3.67) to an odd power, and expanding and rearranging this equation to a form suitable for
the Harmonic Power Flow.

(3.69)
The terms L, and A, are given by equations (3.70) and (3.71) respectively.

AL =

+ 8i2)+...............+did)

(3.71)

Since even harmonics are not considered here, the values of A,, A 2 , .......,Ad and h , are
odd integers only. In addition, since -h 5 Ld 5 h, negative values of Ld are converted to
positive values via the trigonometric identities below.
cos(x) = cos(-x), and - sin(x) = sin(-x)

(3.72)

After directly substituting equation (3.69) into equation (3.68) and expanding,
equation (3.68) will assume the form given by equation (3.73). Note that all harmonic
terms greater than h , are simply ignored.

are directly substituted into


From equation (3.73), the terms gZVrand g2A,
equations (3.50a), and (3.50b) respectively.
Two suggested methods are given by equations (3.74) and (3.79, which can be
used to initialize the fbndamental and harmonic voltage magnitudes at each bus in the
power system, in the absence of a better method. For both initialization methods, the angle

of the hndamental and harmonic voltages in the power system, are all assumed to be zero
degrees to start.

yl)= 1.0, and yk)= 0.1


yl)= 1.0, and yk)= Ilk
In the above two equations, j

i or m, and k 2 3.

Recall that up two additional equations are available in the Harmonic Power Flow,
to solve for the parameters which describe the distorted current waveform entering a
harmonic load, such as in Figure 3.8. For the gas discharge lighting load, only one of these
two equations will be needed to solve for the scaling variable "B" in equation (3.68).
Since this is true, it is only necessary to specifL the real power for a gas discharge lighting
load (i.e., Pmas specified by equation (3.33) ). The reactive power (i.e., Qm as specified by
equation (3.34) ), and the apparent power (i.e., S,,,)
are left unspecified, and will assume a
value consistent with the network solution set. In general, Qmwill be nonzero when
harmonic voltages are present at the gas discharge lighting bus.

An initial estimate for the scaling variable " B" of equation (3.68), can be obtained
by use of equations (3.67) and (3.68). By initially assuming no harmonic voltages at the
gas discharge lighting bus, equation (3.67) becomes equation (3.76).
vm(t) = v:) sin(w,t

+ 8:))

(3.76)

For example purposes, equation (3.68) will be expanded to only two polynomial
coefficients (i.e., w = 3), to give equation (3.77) below.
g,

('1

= '(b1vm (f) + 4 [vm(t)]9)

By substituting equation (3.76) into equation (3.77), and rearranging using


equation (3.78), the result is equation (3.79).

A'
[ Asin (x)]' = -(3sin (x) - sin (3x))
4

(3.77)

Equation (3.76) is multiplied by equation (3.79). This product is then integrated


over the period of either the voltage or current, and divided by the period. This gives the
average load power absorbed by the gas discharge lighting load (i.e.,

pz ). The resulting

equation is rearranged to yield equation (3.80) below. Note that equation (3.80) requires
rms per-unit values for both

P'z and V',".

3.6 Simplifications to the Newton-Raphson Based Harmonic Power Flow Equation Set

As stated in the previous section, since nonlinear components are the source of
harmonics in power systems, it is of great importance in the Harmonic Power Flow, to
accurately model the nonlinear relationship between the current and voltage waveforms at
each bus containing a harmonic device. For this purpose, it is convenient to express
distorted voltages and currents as Fourier Series. For the Harmonic Power Flow, the
Fourier Series of the current entering a harmonic bus is expressed as a function of the
Fourier Series of the voltage at this bus, and of any parameters which describe this
distorted current waveform.
Conventional buses can be modeled exactly the same way as harmonic buses are
modeled in the Harmonic Power Flow. This approach, however is generally not used, in
order to simplify the Harmonic Power Flow. Instead, each conventional bus is modeled as
a complex power demand at the hndamental frequency, and is modeled as an impedance
(that has a linear voltage-current characteristic) at the harmonic frequencies. At each
harmonic frequency, a different impedance is used to model the device(s) at the
conventional bus. This is the approach which is assumed in reference [I].
In the previous section, the Harmonic Power Flow was derived in a general
format. Therefore, both harmonic and conventional buses were approached in the same
way. However, when the simplified modeling approach (discussed above) is used for
conventional buses, two basic simplifications can be made when forming the Harmonic
Power Flow equation set. These simplifications result when the kth harmonic power
system impedance diagram is arranged according to Figure 3.20 (instead of Figure 3 . lo),
when forming the kth harmonic frequency admittance matrix (i.e., lY(*'I ) Although
simplified, the resulting equation set will be equivalent to the set formed by following the
previous section. Note that no information is lost as a result of these simplifications.
Rather, the same information is simply restated in a different way.

The kth harmonic


frequency power system
representation, which
includes the transmission
network made up of all
transformers and
transmission lines. In
addition, all generators
and conventional loads
modeled by a shunt
impedance, are included.
The correct sequence for
the impedance
(admittance) is chosen
from Table 3.1.Harmonic
producing devices and
any conventional bus
device not modeled by a
shunt impedance, are
external.

Figure 3.20. General n -Bus Power System arranged to Form the kfh Harmonic
Admittance Matrix. Shunt Impedances used to Model Conventional Buses at the kth
Harmonic Frequency, are Included when Forming the kth Harmonic Frequency
Admittance Matrix.
Source: Gross [2], p.258.

First, at every conventional bus modeled as a shunt impedance (for each harmonic
frequency, as in Figure 3.12), the real and reactive power mismatch equations only need to
be written with fbndamental frequency components. Therefore, equations (3.35), (3.36),
(3.39), and (3.40) are combined to give equations (3.8 1) and (3.82) below.

In addition, by forming the kth harmonic frequency admittance matrix according to


Figure 3.20, the currents i$) and

g:)

of Figure 3.12 and equation (3.47), are both

automatically included in the KCL equations which are written. Therefore, the harmonic
frequency real and imaginary current mismatch equations to be written at each
conventional bus modeled as a shunt impedance (for each harmonic frequency), are also
simplified. Hence, equations (3.5 1a) and (3.5 1b) are rewritten as equations (3.83) and
(3.84) respectively, where k 2 3, and k is odd ordered.

CyF)y(k)
sin (a!) + y)= 0
Note that all other equations discussed in the previous section, will remain
unchanged.
In the Examples chapter, the equations discussed in the previous section are used
to form the Harmonic Power Flow equation set, for a two bus power system. Then, this
equation set is again formed, using the two simplifications discussed in this section. It will
be shown that both equations sets have an identical solution set.
Although it may be then concluded that the two equation sets are equivalent, it is
very possible that they may converge at a different rate. In addition, depending upon the

initial conditions, the two equation sets may converge to a different solution set. However,
no attempt will be made by the author of this thesis, to prove this hypothesis.

CHAPTER 4 -- DIRECT SOLUTION OF POWER SYSTEM HARMONIC


VOLTAGES

The most common method to determine power system harmonic voltages is the
direct (non-iterative) solution of the harmonic admittance matrix (i.e., irk)l
), at each
harmonic frequency, using equation (4.1) [25].

1 V k1 )= pk)1 Fk)1

(4.1)

This method is often referred to as the Current Injection Technique. The matrix

Iz'*)I

refers to the inverted lrk'lmatrix. The hndamental frequency voltages in the power
system, however, are determined by the Conventional Power Flow
For the Current Injection Technique, the power system is arranged as in Figure
3.20, for each harmonic frequency, in order to obtain each harmonic admittance matrix.

Note however, the harmonic producing devices in the power system, are modeled as ideal
current sources at each harmonic frequency, as shown in Figure 4.1. This is a much
simpler method used to model harmonic devices, as compared to the models proposed by
the Harmonic Power Flow. The magnitude and the angle, of each harmonic current
injected into the power system, by each harmonic device, are assumed to be constant and
known. The vector

I '*)I

in equation (4. l), represents a column matrix which includes the

magnitude and the angle of the harmonic current, produced by each harmonic device, at
the kth harmonic frequency. The vector Ivk)1
in equation (4. l), represents the column
matrix of the unknown harmonic voltages, at every bus in the power system, for the kth
harmonic frequency.

bus m

-(k)

1%
L

Figure 4.1. Representation of a Harmonic Producing Device at Bus m, as an Ideal Current


Source at the kth Harmonic Frequency.

In the examples chapter, the Current Injection Technique is used to find the third harmonic
voltages, in a two bus power system with a gas discharge lighting load.
Project IEEE-5 19 [25] states that the assumption which permits the Current
Injection Technique to be used is that the power system voltages are not distorted.
However, Project IEEE-5 19 also states that the Current Injection Technique is generally
accurate, if the voltage distortion levels in the power system are less than ten percent.
Basically, as the voltage distortion levels in the power system increase, the less effective
the Current Injection Technique becomes, at modeling the relationship between harmonic
voltages and currents in the power system.
The purpose of this chapter was not meant to be a comparison between the
Harmonic Power Flow, and the Current Injection Technique. Rather, the idea was to
present a simple method which can be used in place of the Harmonic Power Flow, in
certain instances. Very often, a simplified solution is adequate in simple radial networks.
However, this is usually not the case in nonradial networks or when the magnitudes of
power system harmonic voltages are significant [I].

CHAPTER 5 -- EXAMPLES

5.1 Introduction

The example system under consideration is as shown in Figure 5.1. The complex
power generated at bus one is carried over a one half mile, 230Kv transmission line, to a
large fluorescent lighting load drawing 6 M W , and OMVARS at bus two. For simplicity, it
will be assumed that the presence of the fluorescent lighting load will only introduce third

harmonic currents and voltages into the power system.


It is noted that this example power system is not entirely realistic. The author of
this thesis constructed this example power system based on readily available information.
In addition, simplicity was desired in order to keep the examples manageable. However,
the solution methods presented in the examples are correct.
For all impedance data, the system base values are given below in equation (5.1).

The generator data of Table 5.1 is supplied in per-unit form on the system bases
chosen.
The configuration of the 230Kv transmission line is as shown in Figure 5.2. Note
that due to lack of information, some dimensions had to be approximated. In Tables 5.2
and 5.3, the data for the transmission line conductors and ground wires is given. Using
Tables 5.2 and 5.3, and the equations in the Appendix, the necessary impedance data for
the 230Kv transmission line was obtained, and is given in Table 5.4. Because the

bur

bur 2

generaor
13.8Kv

0.5 mile - 230Kv transmission line


r

j 0 . 0 2 ~u..

*
Figure 5.1. Example Two-Bus Power System, with a Generator at Bus One and a
Fluorescent Lighting (FL) Load at Bus Two.

s3+",4

(MVA)

VL",
(kV)

~fratsa f r a t 4

(kg)

Rotor

Q*

Q-

(Hz) (Mvar) (Mvar)

Poles (s)

Type
Salient

Xd

XL

Xi

X,

Xi

Xi

X2

X,

Xm

(impedance data in % on generator ratings)

Table 5.1. Fundamental Frequency (60Hz) System Generator Data.


Source: Gross [2], p.25 1.

R,

R,

R,

Figure 5.2. Configuration of the 230Kv Transmission Line in Figure 5.1.


Source: Westinghouse Electric Corporation [20], p.590.

Electrical characteristics of overhead ground wires


Part A : Alumoweld strand
Res~stancc.Rim1
Small currents

60-Hz reactance
for I -lt radiw

75 % of cap.

60- Hz

Strand

25C

25C

75C

75C

(AWG)

oc

60Hz

oc

60 HZ

Induct~ve,
n/ml

Capac~tive,
M n . mi

geometric
mean radius,
it

3 NO. 10

8.870

8.870

10.440

10.670

0.777

0.1 392

0.001650

Table 5.3. Data for the Ground Wires of the 230Kv Transmission Line of Figure 5.2.
Source: Gonen [41], p.658.

freauencv

positive-sequence impedance

zero-sequence impedance

Table 5.4. Impedance Data for the 230Kv Transmission Line of Figure 5.2.

frequency
(Hz)
60

positive-sequence impedance
(per unit)
0.04341 5 + j0.258128

zero-sequence impedance
(per unit)
0.08251 4 + j0.507372

Table 5.5. Per-Unit Impedance Data for the 230Kv Transmission Line of Figure 5.2.

transmission line of Figure 5.1 is only one half mile in length, and the highest harmonic
(i.e., current or voltage) of interest is the third, the short transmission line equations of
Chapter 3 can be used, with shunt capacitance ignored. The transmission line data of Table
5.4 is converted to per-unit, by using the chosen system base values, and is given in Table
5.5. As an example, the 60Hz positive-sequence impedance from Table 5.4 is converted
to per-unit as shown below.

Because no information is given in reference [20] on power system ground wires


for Figure 5.2, a set of ground wires were chosen based on the IEEE Green Book [39].
The criterion used from the IEEE Green Book is that systems above 15Kv should be
effectively grounded. For an effectively grounded system, the hndamental frequency
ratios given by equations (5.3) and (5.4) should be true.

xt)/xf)< +3
#)/xf) < +l

(5.3)
(5.4)

By using Table 5.4, and equations (5.3) and (5.4), the transmission line of Figure
5.2 was determined to be effectively grounded, as shown below.

X:)/X?

= 1.6104/0.8193 = 1.9655 < +3

@"/x(')
= 0.2619/0.8193 = 0.3196 < +1

(5.5)
(5.6)

Three examples are given in this chapter, using the simple two-bus power system
of Figure 5.1. The first example will use the Conventional Power Flow and the Current
Injection Technique to find the hndamental and third harmonic voltages in the power
system. The second example, will use the Newton-Raphson based Harmonic Power Flow,
to determine the hndamental and third harmonic voltages in this power system. The final
example, will discuss how to simplifjr the equation set obtained in the second example. A
comparison will then be made between the equation sets of examples two and three, in
order to show that they are equivalent.

bus 2

0.043415 + j0.258128 p.u.

bus 1
s

Figure 5.3. Impedance Diagram of the Power System in Figure 5.1, for Fundamental
Frequency (60Hz) Analysis.

bus 2

0.140422 + j1.474952 p.u.

bus 1

--

j0.6 p.u.
j0.18 p.u.

Figure 5.4. Impedance Diagram of the Power System in Figure 5.1, for Third Harmonic
(180Hz)Analysis.

The impedance diagrams necessary for all three examples to follow, are given in
Figure 5.3 and 5.4. Only the 60Hz, positive-sequence impedance of the transmission line
will be required, to represent the power system at the hndamental frequency. In addition,
the 180Hz zero-sequence impedance of both the generator and the transmission line will
be required, to represent the power system for third harmonic analysis. Note that in Figure
5.4, the shunt impedance at bus one is composed of two separate impedances. First, the
third harmonic impedance of the generator is estimated by directly scaling its hndamental
frequency negative-sequence reactance (i.e., from Table 5.1) with frequency. The
generator impedance is obtained as shown below.
Zf,?, = 3jx:" = f(j0.2p.u.) = jO.6p.u.

(5.7)

Second, at the hndamental frequency, the generator will be grounded by 3~:) in


the zero-sequence case [2]. Although reference [l] does not discuss inclusion of this
neutral impedance in the harmonic impedance diagrams, it will be assumed here, that this
impedance is not negligible, and should therefore be included. It is noted that reference [I]
assumes that a wye-delta transformer at the generator bus, will isolate the generator from
the power system, in the zero-sequence case. However, since no transformer appears at
the generator bus of Figure 5.1, there will be no isolation of the generator from the power
system, in the zero-sequence case. Therefore, at the third harmonic frequency, the
generator grounding impedance is estimated as shown below, using Table 5.1.

zgd = (3)(3)jX'," = (3)(3)(jO.O2p.u.)

= jO.18p.u.

(5.8)

The typical current waveform entering a fluorescent lighting load, will roughly be
of the form given by equation (5.9), when a nearly sinusoidal voltage is applied [l 11.
gLZ(t) = g 2 c0s(wot + P i ) ) + g g cos(3(wot+Pi'))

+.. ........+gE cos(k(w,t +pi)))+...........+ g z cos(h(w,t + &)))

(5.9)

As an example, typical magnitudes of the third and fifth harmonic currents are 21 % and

7 % ,respectively, of the fbndamental. Using theses typical magnitudes, and assuming that

the fundamental frequency current is phase shifted by R radians, the resulting equation is
given below.
g, (t) = l.Ocos(oot + n) + 0.21cos(3(wot + x)) + 0.07cos(5(oot + n))

(5.10)

Figure 5.5 represents the Fourier Series components of equation (5. lo), while
Figure 5.6 represents the summation of all three terms in equation (5.10).

Figure 5.5. Fourier Series Components of Equation (5.10).

Figure 5.6. Summation of all Three Terms in Equation (5.10).

5.2 Example One


In this example, the hndamental and third harmonic voltages in the power system
of Figure 5.1, will be found by using the Conventional Power Flow, and the Current
Injection Technique.
First, the fbndamental frequency voltages in the power system are obtained by the
Conventional Power Flow. The hndamental frequency admittance matrix is obtained by
arranging Figure 5.3 according to Figure 3.6, and forming KCL equations using the node
voltage method. The resulting admittance matrix is shown below.

The fluorescent load of Figure 5.1, draws a load of 6MW and 0 MVARS . This
complex load is converted to per-unit below, using the system base values given by
equation (5.1).

ei) 6MW / 120MW = 0.05p.u.

(5.12)

Using equations (3.31) and (3.32), and equations (5.11) and (5.12), the real and
reactive power mismatch equations for this power system, are given below. Note that the
fundamental frequency voltage at bus one is assumed to be

Q?

= (3.82034) el)sin

v")

= 1JQp.u.

(&' - 1.73742) + (3.76732)[@')]2

(5.14)

The Newton-Raphson method is now used to solve equations (5.13) and (5.14).
By initially assuming
found to be

K'')

= l/Qp.u., the fbndamental frequency voltage at bus two is

F(')= 0.9977381-0.01352p.u.

Next, the third harmonic voltages in the power system are obtained by using the
Current Injection Technique. The third harmonic admittance matrix is obtained by
arranging Figure 5.4, according to Figure 3.20, and forming KCL equations using the
node voltage method. The resulting admittance matrix is shown below.

It will be assumed, that the measured (or estimated) third harmonic current
entering the fluorescent lighting load at bus two, has a magnitude which is 20% of the
fbndamental current, and is phase shifted according to equation (5.9). Using the
information obtained from the Conventional Power Flow, and equation (5.1I), the
fbndamental frequency fluorescent lamp current is given below.

Therefore, by inspection of equation (5.9) and (5.16), the third harmonic current entering
bus two is given by equation (5.17) below.

Equation (5.17), and the inverted third harmonic admittance matrix given by
equation (5.19, are now substituted into equation (4. I), and the result is shown below.

Direct substitution of equation (5. IS), results in the estimated solution of the third
harmonic voltages for the power system of Figure 5.1, as given below.

5.3 Example Two


For this example, the hndamental and third harmonic voltages in the power system
of Figure 5.1, will be found by using the Newton-Raphson based Harmonic Power Flow.
For the two bus power system of Figure 5.1 (i.e., where n = 2), there are
2(n - 1) = 2(2 - 1) = 2 findamental frequency unknowns for the Harmonic Power Flow.
These hndamental frequency unknowns
bus one is always assumed to be

ql)and d$",where it is noted that the voltage at

V'') = 1lQp.u.

In addition, since the only harmonic frequency of interest is the third harmonic
(i.e., q = I), there are 2(n)(q) = 2(2)(1) = 4 harmonic frequency unknowns for the
Harmonic Power Flow. These third harmonic frequency unknowns are v 3 ) ,

q),
q 3 ) ,and

Lastly, since the gas discharge lighting model represented by equation (3.68) will
be used at bus two, it is necessary to solve for the scaling variable " B ",during the
iteration process.
Therefore, in order to solve for these seven unknowns using the Newton-Raphson
method, seven different algebraic equations are needed.
The first set of equations that are to be formulated for the Harmonic Power Flow,
are the real and reactive power mismatch equations at bus two. However, it is noted that
the gas discharge lighting model (i.e., equation (3.68)) only requires that the real power be
specified at bus two (i.e., P, as given by equation (3.33) ). The reactive power (i.e., Q,)
and the apparent power (i.e., S,) are left unspecified, and will assume values consistent
with the network solution set.
Therefore, according to equation (3.33), the general form of the real power
mismatch equations at bus two is given by equation (5.20) below.

From equation (3.37), the first two terms in equation (5.20) are represented below.

p3) = v(3)v(3)
(3) COS($;)
T2
2
1 ~ 2 1

q)- a: ) + [q3)12
yg) cos(- ay2)

(5.22)

In equation (5.20), the hndamental frequency real load power is given by equation (5.12),
and is repeated below.

From equation (3.41), the last term in equation (5.20) is given below.

fi;)= ~ ' 3 ' (3) cos(@'


2

g ~ 2

(5.23)

- fl))

As will be shown, the gas discharge lighting model of equation (3.68), produces the real
and imaginary components of the current gg) = gz/p;).
- Therefore, the terms g:

and

p';) for equation (5.23) are calculated as shown below.

Special care must be used with equation (5.25), to ensure that the proper sign will result.
The next set of equations to be determined for the Harmonic Power Flow, are the
hndamental frequency real and imaginary current mismatch equations at bus two. Using
equations (3.50a) and (3.50b), the desired equations are given below.

y;?y(') cos(ay,) +

4')
) + yf2q1)cos(a$; + g))
+ g::,

yg)?") sin(a2 + @)) + yfiql) sin(a$


Very shortly, the terms g:,

+ 6<:') + g::,

=0
=0

(5.26)
(5.27)

and g f ' jwill be obtained from the gas discharge lighting

model of equation (3.68).


The last set of equations to be determined for the Harmonic Power Flow, are the
harmonic frequency real and imaginary current mismatch equations at each bus. Using
equations (3.50a) and (3.50b), the desired equations for bus two are given as equations

(5.28) and (5.29). Likewise, the equations for bus one are obtained by using equations
(3.5 1a) and (3.5lb), and are given as equations (5.30) and (5.3 1).
y:i)~;(~)
cos(ag) + p))+ y g ) q )cos(af2 +
sin(($3)

(3) ~ ( 3 )
I

YZI

+ q))
+ y g ) v )sin(ag + g))+ ggi

(5.29)

=0

(5.30)

,, + ~ ) ) + y : ~ ) \ ( 3 ) s i n+$')+gg
(a~~
=0

(5.3 1)

,, + @I)+ y l i ' v ) cos(a/: + 6:))+: :g

sin(a'3'

(3) ~ ( 3 )
I

YII

(5.28)

=0

21

(3) ~ ( 3 cos(a(3)
)

YII

6:')+ gf?,, = 0

The gas discharge lighting model of equation (3.68) will be used to obtain the last
terms in equations (5.26), (5.27), (5.28), and (5.29). However, it will first be shown how
to obtain the last term in equations (5.30) and (5.3 1). By direct observation of Figure 5.4,
and Figure (3.12(a)), the current

g;)

is found by Ohm's Law to give equation (5.32).

Therefore, separation of equation (5.32) into its real and imaginary components, results in
equation (5.33) and (5.34) below, where the angles are assumed to be in radians.

g ~ =\

(g)
(q)
sin

7T

- -)

Equations (5.33) and (5.34) are directly substituted into equations (5.30) and (5.3 1)
respectively.
It will now be shown, how to use the gas discharge lighting model of equation
(3.68), in order to model the fluorescent lighting load at bus two. For this example, it will
be assumed that the current and voltage waveforms measured at bus two, are given below,
where the phase shifts are in radians.
g,, (t) = (0.0522632) cos(o,t

+ 3.13535) + (0.0104525) cos(3(wot+ 3.13535))

v2(t) = (0.997738)cos(oot - 0.01352) + (0.0236155)cos(3wot + 4.63151)

(5.35)
(5.36)

Note that the current and voltage waveforms above, were obtained by using the
information in Example One.
Using a simple and well known curve fitting method called the Least Squares
Approximation [40], the current in equation (5.39, was curve fitted as a hnction of the
voltage in equation (5.36). Appropriate data for the curve fit was obtained by varying a
in equations (5.39, and (5.36) from zero to 2 n radians, in increments of ~ 1 1 2This
. data
is given in columns one, two, and three of Table 5.6. By curve fitting column two in
Table 5.6, as a hnction of column three in Table 5.6, with only two polynomial
coefficients, the following equation was obtained.

Also, by curve fitting column two in Table 5.6, as a hnction of column three in Table 5.6,
with three polynomial coefficients, the equation below was obtained.

Each row in column four of Table 5.6, was generated by using equation (5.37),
and the adjacent value of voltage in column three of Table 5.6. Likewise, each row in
column five of Table 5.6, was obtained by using equation (5.38), and the adjacent value
of voltage in column three of Table 5.6.
A plot of columns two, four, and five in Table 5.6 is shown in Figure 5.7. Carehl

observation reveals that the plots of columns four and five are relatively the same, where
both plots are slightly shifted to the right of the curve given by column two, by the same
amount. It was observed by trial and error, that a perfect curve fit was attainable only
when the current is of the form given by equation (5.9), the voltage is of the form given by
equation (5.39) below, and when E )= 8;).

v2 (t) = ql)c0s(Wot+ g))


+ q3)
C O S ( ~ (+
Cg
C)
))
~)
~

.+v)
cos(k(w,t +

+.........

#I))+..

........+v)
cos(h(m0t + g)))

Table 5.6. Data for Polynomial Curve Fits.

gLZ(f)

gL'l(f)

w t (radians)
0

gLZ(fi'

Polynomial C u r v e Fits

The peak values (and their signs) of the hndamental and harmonic components of the
current and voltage had very little bearing on how good a curve fit could be attained.
With these last few points in mind, the fluorescent lighting model for bus two is
obtained from equations (3.68) and (5.37), and is given below.
g ~ (20 = ~(b1v2
( 0 + b3[v2(01')
It, = -0.0215250,b3 = -0.0412504

(5.40)

Note that in equation (5.40), the terms v2( t )and [v2(t)]' are to be represented in Fourier
Series form by using equation (3.69). Note that equation (3.69) is simply a shortcut
method of raising equation (3.67) to an odd power, and expanding and rearranging this
equation to a form suitable for the Harmonic Power Flow. After substituting equation
(3.69) into equation (5.40), and collecting like terms together, the resulting equation will
be of the form given by equation (3.73), which is repeated below.

The terms g::,r,

g!;,

gl):,, and g z j are directly obtained fiom this last step, and are

shown below, where b, = -0.0215250, and b, = -0.0412504 in all four equations.


g:;,

= Bb,

ql)cos($')

3
+ ~b,(--[h("]'

+3[vi]2
2
v,(l))~~s($))

+ Bb, -3 [g1)]2
V:)' cos(@) - 2$))
4

g:;,

= ~ b , sin ($I)

;(3

,
3

+ [ b(3)]' sin (a))


+~ b , [
3
+ Bb, -[q1)I2
5"' s i n ( p - 2$')

gfiSr= Bb, 3(3)


cos($) )
+ Bb, -1
cos(3q')

[v)I3

4
3
+ Bb3(l[v"']'

+2[q"r
2
c3))cos($')

Equations (5.41), (5.42), (5.43), and (5.44) are then directly substituted into equations
(5.26), (5.27), (5.28), and (5.29) respectively.
In equations (5.21), (5.22), and equations (5.26) through (5.3 I), the " y W ", and
II

am terms, represent the magnitudes and the angles respectively, of elements from either
II

the fbndamental (i.e., f = 1) or the third harmonic (i.e., f = 3) admittance matrix.


The fbndamental frequency admittance matrix for the Harmonic Power Flow is
identical to the hndamental frequency admittance matrix obtained for the Conventional
Power Flow in Example One. Therefore, the matrix given by equation (5.11) is to be used
to complete equations (5.2 l), (5.26), and (5.27).
The third harmonic admittance matrix is obtained by arranging Figure 5.4,
according to Figure 3.10, and forming KCL equations using the node voltage method. The
resulting admittance is shown below.

The matrix of equation (5.45) is to be used to complete equation (5.22), and equations
(5.28) through (5.3 1). Note that this matrix is not the same as the one given by equation
(5.15).
In order to solve for the Harmonic Power Flow equation set using the NewtonRaphson method, it is first necessary to supply a reasonably good initial condition for each
unknown. For this example, the magnitudes and the angles of all bus voltages, were
initialized by using the information from Example One. These initial conditions are
presented below.

In addition, it is necessary to provide an initial condition for the scaling variable


"B" in equations (5.41) through (5.44). Therefore, using equation (3.80) with the

magnitude of the fbndamental frequency voltage at bus two given by equation (5.46), the
load power given by equation (5.12), and the values for b, and b3 from equation (5.40),
the initial value for " B" is given below.

Using the Newton-Raphson method to solve the Harmonic Power Flow equation
set, with the initial conditions given by equations (5.46) and (5.47), the final results are
given below. To solve the Harmonic Power Flow equation set for this example, a
spreadsheet software package (Microsoft Excel) and a separate matrix inversion software
package (MathCad 4.0) was used.

It is noted that the final results above, closely match the initial conditions given by
equations (5.46) and (5.47). However, note that it was not expected that the final results
would match up perfectly with the initial conditions, even for this extremely simple power
system. This is true, simply because a perfect curve fit of equation (5.35) as a function of
equation (5.36), could not be obtained (i.e., as was observed in Figure 5.7).

5.4 Example Three


In this example, it will be discussed how to simplie the Harmonic Power Flow
equation set obtained from Example Two. A comparison will then be made between the
Harmonic Power Flow equation sets of Examples Two and Three, in order to show that
they are equivalent.
In Example Two, a simplified approach was used to model the generator bus (a
conventional bus) for the Harmonic Power Flow. Since bus one is the slack bus, it does
not have to be modeled as a complex power demand at the hndamental frequency. In the
introduction, it was stated that all power generated at bus one, was delivered to bus two.
At the third harmonic frequency, the slack (generator) bus is simply modeled as a shunt
impedance. This impedance represents a model of the generator at the third harmonic
frequency. Therefore, this approach allows the Harmonic Power Flow equation set from
Example Two, to be simplified. These simplifications result by obtaining the third
harmonic admittance matrix according to Figure 3.20, instead of Figure 3.10. Note that
the third harmonic admittance matrix which is to be used for this example is given by
equation (5. IS), instead of equation (5.45). The hndamental frequency admittance matrix
given by equation (5.1 l), however, remains unchanged.
The first simplification now possible is to write the real and reactive power
mismatch equations at the conventional bus as hndamental frequency equations only.
However, since the conventional bus is also the slack bus, these equations were not
needed in Example Two.
The second simplification possible is to rewrite the harmonic frequency real and
imaginary current mismatch equations at the conventional (slack) bus using equations
(3.83) and (3.84). Therefore, equations (5.30) and (5.3 1) are rewritten as equations (5.49)
and (5.50) respectively.

y,':) v;") cos(ai:)+ q))


+ yI(;)4(')cos(4;)+ 6(,3)) = 0

(5.49)

sin

(3) ~ ( 3 )
Y11 1

+ 43)
) + y:;) 1/:3)sin (a(3)
12 + @)) = 0

(5.50)

Again, it is to be understood, that no information has been lost as a result of these


simplifications. Rather, the same information has been simply restated in a different way.
Therefore, the total simplified Harmonic Power Flow equation set is given by
equations (5.20) through (5.29), equations (5.41) through (5.44), and equations (5.49)
and (5.50). These equations are to be used with the matrices given by equations (5.11) and
(5.15). Note that equations (5.30) through (5.34), are no longer needed.
Substitution of the results given by equations (5.48), into the simplified Harmonic
Power Flow equation set, will show that each equation will cancel to approximately zero
(i.e., less than 1.0 x 10-'p.u.). Therefore, the equation sets for Examples Two and Three,
have an identical solution set, and have been determined to be equivalent.

CHAPTER 6 -- CONCLUSIONS AND FUTURE WORK

6.1 Conclusions

It has been shown in this thesis that the Newton-Raphson based Harmonic Power
Flow is a logical extension of the Newton-Raphson based Conventional Power Flow. The
equations used to determine the fundamental frequency voltages in the power system for
the Newton-Raphson based Harmonic Power Flow, are very similar to the equations
derived for the Newton-Raphson based Conventional Power Flow. The only difference is
that the equations derived for the Newton-Raphson based Conventional Power Flow are
written in Fourier Series form for the Newton-Raphson based Harmonic Power Flow. If
the conventional buses in the power system are modeled as a shunt impedance at the
harmonic frequencies, then the equations used to determine the fbndamental frequency
voltages in the power system are exactly the same for both power flows.
The additional equations required for the Newton-Raphson based Harmonic Power
Flow result from the need to determine harmonic frequency voltages in addition to the
fbndamental frequency voltages in the power system. The fbndamental frequency
equations in both power flows are based on the real and reactive power flows in the power
system. The harmonic frequency equations for the Newton-Raphson based Harmonic
Power Flow are based on the harmonic current flows in the power system. For both the
fbndamental and harmonic frequency equations, all equations are written as a fbnction of
the power system impedances and as a fbnction of the voltages throughout the power
system.

Therefore, although there are more equations that must be solved simultaneously
for the Newton-Raphson based Harmonic Power Flow, the equations are constructed
similarly to those derived for the Newton-Raphson based Conventional Power Flow.
It has also been shown in this thesis that a major drawback of using the NewtonRaphson based Harmonic Power Flow is the large number of nonlinear algebraic equations
that must be solved simultaneously. For the Newton-Raphson based Conventional Power
Flow, the number of nonlinear algebraic equations that must be solved simultaneously are

2 ( n - l ) , where n is the number of buses in the power system. For every bus in the
power system, it is desired that the hndamental frequency voltage magnitude and angle be
determined. For the slack bus, the fbndamental frequency voltage is always assumed to be

V,"'

= 1/Qp.u.as a reference.

For the Newton-Raphson based Harmonic Power Flow, the number of equations
required to solve for the hndamental frequency voltages in the power system is again

2(n - 1 ) . The number of equations required to solve for the harmonic frequency voltage
magnitudes and angles is 2(n)(q) where q is the number of harmonic frequencies of
interest. Up to an additional 2 ( M ) equations is required to solve for the parameters which
describe the current waveform entering each harmonic bus m,where M is the number of
harmonic buses. Therefore, the total number of equations which may be required to solve
for all of the unknowns in the Newton-Raphson based Harmonic Power Flow is

2(n- 1 ) + 2 ( n ) ( q )+ 2 ( M ) .
For a modestly small power system composed of ten buses, the Newton-Raphson
based Conventional Power Flow requires the simultaneous solution of 2(10 - 1) = 18
nonlinear algebraic equations. If is desired that all odd harmonics up to the 25th harmonic
are to be analyzed, this would require an additional 2(n)(q)= 2(10)(12)= 240 equations,
not including the 2 ( M ) equations which may be required for harmonic buses.
The major difficulty results when inverting the Jacobian matrix. For the NewtonRaphson based Harmonic Power Flow, a 2 ( n - 1 ) x 2 ( n - 1) Jacobian matrix must be

inverted at each iteration. For the Newton-Raphson based Harmonic Power Flow, it is
necessary to invert a [2(n- 1 ) + 2(n)(q)+ 2 ( M ) ]x [2(n- 1 ) + 2(n)(q)+ 2 ( M ) ] Jacobian
matrix at each iteration. This requires a very large amount of computer memory in
comparison to the Newton-Raphson based Conventional Power Flow. Because of the
large number of equations that must be solved simultaneously, the rate of convergence can
be expected to be slower in comparison to the Newton-Raphson based Conventional
Power Flow. In addition, the chances for having equations diverge increases as the number
of equations increase.

6.2 Future Work

Better load and device models are needed to obtain more accurate results when
using the Newton-Raphson based Harmonic Power flow. Although the Newton-Raphson
based Harmonic Power Flow is a logical method for determining power system harmonic
voltages, the results obtained are critically dependent on how the power system is modeled
at each frequency. Several of the models derived for representing conventional devices at
harmonic frequencies in this thesis were approximative. In addition, it was discussed that
the model used to represent a gas discharge lighting load is critically dependent upon
curve fitting the current entering the load as a function of the voltage at this bus. In an
example two in this thesis it was shown that it was not possible to obtain a perfect curve
fit unless the current at each frequency was perfectly in phase with the voltage at that
respective frequency. This is a highly unlikely occurrence, particularly when several
different harmonic sources are present in the power system. Therefore, it is recommended
that if more accurate results are desired when using the Newton-Raphson based Harmonic
Power Flow, that more research be put into obtaining more accurate load and device
models.
Another way to increase the accuracy of the results obtained from the Harmonic
Power Flow, would be to obtain better initialization methods for the unknowns in the
equations. For this thesis, gas discharge lighting loads were the primarily focus of
discussions concerning harmonic sources in power systems. Recommendations were
provided for initializing the fundamental and harmonic frequency voltages in a power
system when the only harmonic sources present were gas discharge lighting loads. These
recommendations do not account for the effect that the rest of the power system will have
on the magnitudes and phase shifts of the respective voltages. This becomes of particular
concern when several different harmonic sources are present in the power system.
The results obtained from the Newton-Raphson based Harmonic Power Flow are
critically dependent upon the initial conditions for the unknowns in the equations. It is

noted that the Newton-Raphson method works best when the initial conditions provided
are close to an actual solution set. Because of the large number of equations that must be
solved simultaneously, the chances for having equations diverge increases dramatically in
comparison to the Newton-Raphson based Conventional Power Flow. In order to increase
the usefulness of the Newton-Raphson based Harmonic Power Flow, a more accurate
method for initializing the unknowns in the equations is necessary.

LIST OF REFERENCES

W. M. Grady, Harmonic Power Flow Studies, Ph.D. Dissertation, Purdue


University, West Lafayette, Indiana, August 1983.
C.A. Gross, Power System Analysis, Second Edition, John Wiley & Sons, Inc.,
New York, 1986.
A.E. Emanual, "Powers in Nonsinusoidal Situations-A Review of Definitions and
Physical Meaning", IEEE Transactions on Power Delivery, Vol. 5, No.3, pp. 13771389, July 1990.
IEEE Working Group on Power System Harmonics, "Power System Harmonics:
An Overview", IEEE Transactions on Power Apparatus and Systems, Vol. PAS102, No.8, pp. 2455-2460, August 1983.
A.A. Girgis, et al, "Measurement and Characterization of Harmonic and High
Frequency Distortion for a Large Industrial Load", IEEE Transactions on Power
Delivery, No.70, pp.427-434, April 1989.
N. Mohan, T.M. Undeland, W.P. Robbins, Power Electronics-Converters,
Applications, and Design, John Wiley & Sons, Inc., New York 1989.

K. Olejniczak, G.T. Heydt, "Basic Mechanisms of Generation and Flow of


Harmonic Signals in Balanced and Unbalanced Three-Phase Power Systems",
IEEE Transactions on Power Delivery, Vol. 4, No. 4, pp.2 162-2170,
October 1989.
M.A. Boost, P.D. Ziogas, "State-of-the-Art Carrier PWM Techniques: A Critical
Evaluation", IEEE Transactions on Industry Applications, Vo1.24, No.2, pp. 271280, MarchIApril 1988.
J.H. Eto, A.D. Almeida, "Saving Electricity in Commercial Buildings with
Adjustable-Speed Drives", IEEE Transactions on Industry Applications, Vo1.24,
No.3, pp. 439-442, May/June 1988.

R.A. Hanna, "Harmonics and Technical Barriers in Adjustable-Speed Drives",


IEEE Transactions on Industry Applications, Vo1.25, No. 5, pp.894-900,
SeptemberIOctober 1989.
W.M. Grady, G.T. Heydt, "Prediction of Power System Harmonics Due to
Gaseous Discharge Lighting", IEEE Transactions on Power Apparatus and
Systems, Vol. PAS-104, No.3, pp.554-561, March 1985.
S.E. Kalinowsky, J.J. Martello, "Electrical and Illumination Characteristics of
Energy-Saving Fluorescent Lighting as a Function of Potential", IEEE
Transactions on Industry Applications, Vo1.25, No.2, pp.208-2 15, MarchIApril
1989.
D. Xia, G.T. Heydt, "Harmonic Power Flow Studies. Part 11-Implementation and
Practical Application", IEEE Transactions on Power Apparatus and Systems,
Vol.PAS-101, No.6, pp. 1266-1270, June 1982.
S.R. Mendis, D.A. Gonzalez, "Harmonic and Transient Overvoltage Analyses in
Arc Furnace Power Systems", IEEE Transactions on Industry Applications,
Vo1.28, No.2, pp.336-342, MarcWApril 1992.

G. Manchur, C.C. Erven, "Development of a Model for Predicting Flicker from


Electric Arc Furnaces", IEEE Transactions on Power Delivery, Vo1.7, No. 1,
pp.416-426, January 1992.
A.E. Fitzgerald, C. Kingsley Jr., S.D. Umans, Electric Machinery, McGraw-Hill
Book Co., Inc., New York, 1983.
H.W. Dommel, A. Yan, S. Wei, "Harmonics from Transformer Saturation", IEEE
Transactions on Power Systems, Vol. PWRD-1, No.2, pp.209-215, April 1986.
M.F. McGranaghan, R.C. Dugan, J.A. King, W.T. Jewell, "Distribution Feeder
Harmonic Study Methodology", IEEE Transactions on Power Apparatus and
Systems, Vol. PAS-103, No. 12, pp.3663-367 1, December 1984.
A. Semylen, E. Acha, J. Arrillaga, "Harmonic Norton Equivalent for the
Magnetizing Branch of a Transformer", IEE Proceedings, Vol. 134, Pt.C, No.2,
pp. 162-169, March 1987.
Westinghouse Electric Corporation, Electrical Transmission and Distribution
Reference Book, East Pittsburgh, Pennsylvania, 1964.
J.C. Moreira, T.A. Lipo, "Modeling of Saturated ac Machines Including Air Gap
Flux Harmonic Components", IEEE Transactions on Industry Applications,
Vo1.28, No.2, pp.343-349, MarcWApril 1992.

A. Semylen, J.F. Eggleston, J. Arrillaga, "Admittance Matrix Model of a


Synchronous Machine for Harmonic Analysis", IEEE Transactions on Power
Systems, Vol. PWRS-2, No.4, pp.833-840, November 1987.
F.A. Dewinter, L.G. Grainger, "A Practical Approach to Solving Large Drive
Harmonic Problems at the Design Stage, "IEEE Transactions on Industry
Applications, Vol. 26, No.6, pp. 1095-1101, NovemberDecember 1990.
W.M. Grady, M.J. Samatyj, A.H. Noyola, "The Application of Network Objective
Functions for Actively Minimizing the Impact of Voltage Harmonics in Power
Systems", IEEE Transactions on Power Delivery, Vo1.7, No.3, pp. 1379- 1386, July
1992.
Update of Harmonic Standard IEEE-5 19, "IEEE Recommended Practices and
Requirements for Harmonic Control in Power Systems", October 1991.
J. S. Subjak Jr., J.S. McQuilken, "Harmonics-Causes, Effects, Measurements, and
Analysis: An Update", IEEE Transactions on Industry Applications, Vo1.26,
pp. 1034- 1042, NovemberDecember 1990.

D.T. Rizy, E.W. Gunther, M.F. McGranaghan, "Transient and Harmonic Voltages
Associated with Automated Capacitor Switching on Distribution Systems", IEEE
Transactions on Power Systems, pp.7 13-723, Vol. PWRS-2, No.3, August 1987.
P.K. Sen, H.A. Landa, "Derating of Induction Motors Due to Waveform
Distortion", IEEE Transactions on Industry Applications, Vo1.26, No.6, pp. 11021107, NovernberDecember 1990.
A.P.S. Meliopoulos, M.A. Martin Jr., "Calculation of Secondary Cable Losses and
Ampacity in the Presence of Harmonics", IEEE Transactions on Power Delivery,
Vo1.7, No.2, pp.45 1-459, April 1992.
T.H. Ortmeyer, K.R. Chakravarthi, A.A. Mahrnoud, "The Effects of Power
System Harmonics on Power System Equipment and Loads", IEEE Transactions
on Power Apparatus and Systems, Vol. PAS- 104. No.9, pp.2555-2563, September
1985.
A.A. Girgis, T.L. Baldwin, E.B. Makram, H.S. Fortson Jr., "Testing the
Performance of Three-Phase Induction Watthour Meters in the Presence of
Harmonic Distortion", IEEE Transactions on Industry Applications, Vo1.26, No.4,
pp.689-695, July/August 1990.

I. Purkayastha, P.J. Savoie, "Effects of Harmonics on Power Measurement", IEEE


Transactions on Industry Applications, Vo1.26, No.5, pp.944-946,
SeptemberIOctober 1990.
Y.Ohura, T. Matsuda, M. Suzuki, F. Andow, Y. Kuosawa, A. Takeuchi", A
Digital Distance Relay Using Negative Sequence Current", IEEE Transactions on
Power Delivery, Vo1.5, No. 1, pp.79-84, January 1990.
J.J. Burke, D.C. Grifith, D.J. Ward, "Power Quality-Two Different Perspectives",
JEEE Transactions on Power Delivery, Vo1.5, No.3, pp. 1501- 1513, July 1990.
G.D. Flinn, The Development of a New Power Flow Program, Masters Thesis,
Ohio State University, Columbus, Ohio, 1978.
R.W. Hamming, Numerical Methoh for Scientists and Engineers, Second
Edition, McGraw-Hill Book Co., Inc., New York 1973.
J. Choma Jr., Electrical Networks-Theory and Analysis, John Wiley & Sons, Inc.,
New York, 1985.
A.A. Mahmoud, R.D. Shultz, "AMethod for Analyzing Harmonic Distribution in
A.C. Power Systems", IEEE Transactions on Power Apparatus and Systems, Vol.
PAS-101, No.6, pp. 1815-1824, June 1982.
ANSI/IEEE std 142-1982, "IEEE Recommended Practice for Grounding of
Industrial and Commercial Power Systems".
M.C. Kohn, Practical Numerical Methoh-Algorithms and Programs, Macmillan
Publishing Co., Inc., New York, 1987.
T. Gonen, Electric Power Distribution System Engineering, McGraw-Hill Book
Co., Inc., New York 1986.
Y. Baghzouz, O.T. Tan,"Maximum Heating and Insulation Stress on
Untransposed Lines with Unbalanced Harmonic Distortion", JEEE Transactions on
Power Delivery, Vol. PWRD-2, No. 1, pp.256-261, January 1987.

APPENDIX

This appendix supplies the equations used to calculate the positive-, negative-, and
zero-sequence impedances of the 230Kv transmission line of Figure 5.2, found in the
Examples chapter. These impedance values are used to create a single phase pi-equivalent
model of the transmission line for the hndamental and harmonic frequencies (i.e., Figures
5.3 and 5.4 respectively). Note that the correct phase sequence is given by Table 3.1.

Because the transmission line is only one half mile in length, and the highest harmonic of
interest is the third, the "short line" pi-equivalent model of Figure 3.14 is used, with shunt
capacitance ignored.

Positive- and negative-sequence series impedance


The transmission line positive- and negative-sequence series impedance for each
frequency is obtained from equation (A. I), where the result is in Q/conductor/mile.

zy = z-?= ra + j(0.2794)(f/60)loglo(l/GMR,,)+ j(0.2794)( f /60)loglo(GMD)


(A.1)

The terminology of equation (A. 1) is explained below.

ra = The conductor resistance obtained from Table 5.2, that is varied with frequency in
order to account for skin effect.

f = The frequency of interest, in cycles per second.


GMR,,

= The conductor geometric mean

GMD = The geometric mean distance =

-d

radius in feet.
.

dl, ,d,, ,d3, = The distances between the conductors in feet.

Zero-sequence series impedance


The calculation of the transmission line zero-sequence series impedance, for each
frequency is more complicated than the positive- or negative-sequence calculations. The
return paths for the zero-sequence currents, are the earth and tower ground wires.
Therefore, earth and ground wire paths can be very important when determining the zerosequence impedance of a transmission line. A number of different factors will effect the
zero-sequence current return paths, including tower geometry, wire conductivity, and the
number of ground wires. Carson's equations [20] include these factors, and are used to
calculate the zero-sequence impedance of the 230Kv transmission line of Figure 5.2.
The total zero-sequence impedance of one three-phase circuit consisting of three
conductors and two ground wires (with earth return) is given by equation (A.2) below.

The term

zxl refers to the zero-sequence self-impedance of one three-phase circuit

consisting of three conductors (with earth return) but without ground wires. This
impedance is calculated with the use of equation (A.3), where the result is in

The term 2g)refers to the zero-sequence self-impedance of two ground wires


(with earth return) and is given by equation (A.4) below, where the result is in

RJphaselmile.
-

(A.4)
zg) = 3(ra)/2 + (0.00477)f + j(0.0 1397)f loglo(D,/,/(GMR,,, )(d,) )
The term Zx,, refers to the zero-sequence mutual impedance between the threephase circuit as one group of conductors and the two ground wires as the other group of
conductors. This impedance is calculated with the use of equation (A.5), where the result
is in RJphaselmile.

= (0.00477)f + j(0.01397)flogl,

( D e / ~ d ~ ~ d b g l d ~ 8 1 d ~ 8 2 d b g 2 d c 8 2)

For equations (A.3), (A.4), and (AS), the terms ra,J GMR,,

W.5)
GMD, dl,, d2,,

d,,, are the same as defined for equation (A.1). The other terms in equations (A.3), (A.4),
and (AS), are explained below.

p = The earth resistivity is in meter-ohms, and is usually taken to be a value of 100 meter-

ohms, when the actual composition of the earth is unknown.


GMR,,,

= The ground wire geometric mean

d,

separation distance between the two conductors, in feet.

= The

radius in feet.

dag1= The distance between conductor a, and the first ground wire. The terms dbgl,dcgl,
dag2,dbg2,and d,,, are defined similarly.

Anda mungkin juga menyukai