Lecture Notes in Electrical Engineering 412

Andrzej Kłos

Mathematical
Models of
Electrical Network
Systems
Theory and Applications - An
Introduction
Lecture Notes in Electrical Engineering

Volume 412

Board of Series editors

Leopoldo Angrisani, Napoli, Italy
Marco Arteaga, Coyoacán, México
Samarjit Chakraborty, München, Germany
Jiming Chen, Hangzhou, P.R. China
Tan Kay Chen, Singapore, Singapore
Rüdiger Dillmann, Karlsruhe, Germany
Haibin Duan, Beijing, China
Gianluigi Ferrari, Parma, Italy
Manuel Ferre, Madrid, Spain
Sandra Hirche, München, Germany
Faryar Jabbari, Irvine, USA
Janusz Kacprzyk, Warsaw, Poland
Alaa Khamis, New Cairo City, Egypt
Torsten Kroeger, Stanford, USA
Tan Cher Ming, Singapore, Singapore
Wolfgang Minker, Ulm, Germany
Pradeep Misra, Dayton, USA
Sebastian Möller, Berlin, Germany
Subhas Mukhopadyay, Palmerston, New Zealand
Cun-Zheng Ning, Tempe, USA
Toyoaki Nishida, Sakyo-ku, Japan
Bijaya Ketan Panigrahi, New Delhi, India
Federica Pascucci, Roma, Italy
Tariq Samad, Minneapolis, USA
Gan Woon Seng, Nanyang Avenue, Singapore
Germano Veiga, Porto, Portugal
Haitao Wu, Beijing, China
Junjie James Zhang, Charlotte, USA
About this Series

“Lecture Notes in Electrical Engineering (LNEE)” is a book series which reports

the latest research and developments in Electrical Engineering, namely:
• Communication, Networks, and Information Theory
• Computer Engineering
• Signal, Image, Speech and Information Processing
• Circuits and Systems
• Bioengineering
LNEE publishes authored monographs and contributed volumes which present
cutting edge research information as well as new perspectives on classical fields,
while maintaining Springer’s high standards of academic excellence. Also
considered for publication are lecture materials, proceedings, and other related
materials of exceptionally high quality and interest. The subject matter should be
original and timely, reporting the latest research and developments in all areas of
electrical engineering.
The audience for the books in LNEE consists of advanced level students,
researchers, and industry professionals working at the forefront of their fields. Much
like Springer’s other Lecture Notes series, LNEE will be distributed through
Springer’s print and electronic publishing channels.

More information about this series at [Link]

Andrzej Kłos

Mathematical Models
of Electrical Network
Systems
Theory and Applications - An Introduction

123
Andrzej Kłos
Electrical Department
Warsaw University of Technology
Warsaw
Poland

ISSN 1876-1100 ISSN 1876-1119 (electronic)

Lecture Notes in Electrical Engineering
ISBN 978-3-319-52176-3 ISBN 978-3-319-52178-7 (eBook)
DOI 10.1007/978-3-319-52178-7
Library of Congress Control Number: 2016963175

© Springer International Publishing AG 2017

This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part
of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations,
recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission
or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar
methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this
publication does not imply, even in the absence of a specific statement, that such names are exempt from
the relevant protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this
book are believed to be true and accurate at the date of publication. Neither the publisher nor the
authors or the editors give a warranty, express or implied, with respect to the material contained herein or
for any errors or omissions that may have been made. The publisher remains neutral with regard to
jurisdictional claims in published maps and institutional affiliations.

Printed on acid-free paper

This Springer imprint is published by Springer Nature

The registered company is Springer International Publishing AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
Preface

The electricity and electrical networks are used now in so many fields of science
and engineering practice that the further development of theoretical background and
practical applications of electrical networks is necessary. The recently used network
(circuits) theory and methodology of linear network system analysis and solution
are based, mainly, on rather simple mathematical models. The majority of text-
books, handbooks, and lecturing materials concerning electrical network systems,
which are used in a high- and medium-level technological universities and technical
schools, are outdated. Nowadays the electricity supply utilities have serious prob-
lems concerning security, economy, ecology, etc., e.g. power systems have serious
problems with development of high voltage networks and building new power
stations. This book is needed which is widening the mathematical background
(network theory), not so much as to be difficult to the readers, but useful as a tool
for the new practical applications. This book presents a modern and
non-conventional network theory and its practical applications in network analysis
and solution. In the first part of this book, an advanced mathematical (linear
algebra) approach to the modeling of time-constant networks is given. The alge-
braic model of network system topology is defined, and topological equations are
derived and expressed in the form of a linear space. It was shown that modeling
network graph in terms of linear algebra leads to the non-singular topological
transformation matrix T, which appears to be a useful tool of network analysis and
solution. The algebraic models of the Kirchhoff’s current and voltage laws and
of the Ohm’s law are derived and expressed in the form of linear spaces. It makes
possible the derivation of various, commonly not known, equations, which are
widening the methodology of network analysis. The combined current–voltage
vector is defined, and it leads to the unexpected result; it was proved that the
summation of current and voltage values is reasonable from the mathematical
viewpoint. Using the algebraic models of currents and voltages, the generally not
known mathematical formulations of fundamental Kirchhoff’s laws are derived and
discussed. The classical Ohm’s law is supplemented by introducing the system
parameters, which enable using the current and voltage sources in network system.

v
vi Preface

In the second part of this book, using the algebraic network model, the various
applications of this model are presented. The connection between theory and
practical network problems is shown by solving the selected examples of network
problems in which the variety of input data, solvability conditions, not known
solution method, computation efficiency etc., are to be taken into account. The
general algebraic model of network solution (generalization of classical method) is
presented which may be applied to variety of technical and non-technical fields.
Particularly, some examples of practical applications in the field of power system
network analysis and solution are given. The usefulness of the new formulations of
Kirchhoff’s and Ohm’s laws is shown using simple examples of network. The
solution method of the arbitrary input data problems is given using algebraic net-
work model. Using topological matrix T, the equations of network system analysis
and solution are derived and discussed. The not known solution methods of load
flow in power system network are derived and illustrated using the example of real
network.
The text of this book includes the mathematical derivations and formulas, but it
is understandable for engineers and students. Mathematically, more difficult parts,
e.g., linear space terminology, are illustrated and described in a way understandable
for non-mathematicians. The book level and contents are addressed to researchers,
university lecturers, software developers, and advanced undergraduate and post-
graduate students involved in power system network analysis and network
development.

Warsaw, Poland Andrzej Kłos

Contents

Part I Mathematical Model of Network System

1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.1 Network System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2 Modeling Network System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.3 Solving Network System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.4 Network Laws . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2 Basic Notions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.1 Network Graph Notions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.2 Electrical Network Notions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
3 Algebraic Model of Network Graph . . . . . . . . . . . . . . . . . . . . . . . . . 13
3.1 Cut-Set and Loop-Set Vectors . . . . . . . . . . . . . . . . . . . . . . . . . . 13
3.2 Topology of Network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
3.3 Topological Model of Loop-Sets . . . . . . . . . . . . . . . . . . . . . . . . 15
3.4 Topological Model of Cut-Sets . . . . . . . . . . . . . . . . . . . . . . . . . . 16
3.5 Orthogonality of Cut-Sets and Loop-Sets . . . . . . . . . . . . . . . . . . 18
3.6 Linear Space Model of Network Topology . . . . . . . . . . . . . . . . . 19
3.7 Topological Transformation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
3.8 Topological Matrix T . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
4 Algebraic Model of Network Currents . . . . . . . . . . . . . . . . . . . . . . . 25
4.1 Matrix Model of Network Currents . . . . . . . . . . . . . . . . . . . . . . 25
4.2 Linear Space Model of Network Currents. . . . . . . . . . . . . . . . . . 26
4.3 Topological Transformation of Current Space . . . . . . . . . . . . . . 28
4.4 Current Relations Using Topological Matrix T . . . . . . . . . . . . . . 29
5 Algebraic Model of Network Voltages . . . . . . . . . . . . . . . . . . . . . . . . 31
5.1 Matrix Model of Network Voltages . . . . . . . . . . . . . . . . . . . . . . 31
5.2 Linear Space Model of Network Voltages . . . . . . . . . . . . . . . . . 33

vii
viii Contents

5.3 Topological Transformation of Voltage Space . . . . . . . . . . . . . . 34

5.4 Voltage Relations Using Topological Matrix T . . . . . . . . . . . . . . 35
6 Algebraic Model of Current–Voltage Vectors . . . . . . . . . . . . . . . . . . 37
6.1 Orthogonality of Current and Voltage Vectors . . . . . . . . . . . . . . 37
6.2 Current–Voltage Vector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
6.3 Linear Space Model of Current–Voltage Vectors . . . . . . . . . . . . 39
6.4 Current–Voltage Vectors’ Relations Using Matrix T. . . . . . . . . . 40
6.5 Comment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
7 Kirchhoff’s Laws Using Matrix T . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
7.1 Basic b-Dimensional Formulation . . . . . . . . . . . . . . . . . . . . . . . . 43
7.2 Pseudo-Unit Formulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
7.3 Current–Voltage Formulation . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
8 Current–Voltage Functional Relations . . . . . . . . . . . . . . . . . . . . .... 47
8.1 Branch Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .... 47
8.2 Ohm’s Law of Network System—Algebraic Model . . . . . . .... 50
8.3 Linear Space Model of Current–Voltage State Vectors
in Network System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .... 52
8.4 Formulation of Ohm’s Law Using Matrix T . . . . . . . . . . . . .... 55
8.5 Matrices of System Admittance and System Impedance . . . .... 57
9 General Comments to Part I . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
9.1 Part II Application Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . 60

Part II Application Examples

10 Applications of b-Dimensional Formulation of Kirchhoff’s
Laws . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
10.1 Basic b-Dimensional Formulation . . . . . . . . . . . . . . . . . . . . . . . . 63
10.2 Pseudo-Unit Formulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
11 Network Solution Method Using Algebraic Network Model . . . . . . 69
11.1 Derivation of Method Using Matrix T . . . . . . . . . . . . . . . . . . . . 69
11.2 Classical Network Solution Methods . . . . . . . . . . . . . . . . . . . . . 71
12 Method of Current and Voltage Sensitivity Analysis . . . . . . . . . . . . 73
12.1 Derivation of Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
12.2 Numerical Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
13 Method of Arbitrary Input Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
13.1 Description of Arbitrary Input Data . . . . . . . . . . . . . . . . . . . . . . 77
13.2 Derivation of Solution Equations . . . . . . . . . . . . . . . . . . . . . . . . 78
13.3 Solvability Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
13.4 Numerical Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
Contents ix

14 Logical Optimization Method of Network System . . . . . . . . . . . . . . 87

14.1 Example of Network System Matrices . . . . . . . . . . . . . . . . . . . . 87
14.2 Logical Analysis of Network System Equations . . . . . . . . . . . . . 89
14.3 Application Possibilities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
15 Current-Based Method of Load Flow Solution . . . . . . . . . . . . . . . . . 97
15.1 Current State-Oriented Topology of Network . . . . . . . . . . . . . . . 97
15.2 Current State Vector and Current Flow Method . . . . . . . . . . . . . 98
15.3 Current-Based Load Flow Method . . . . . . . . . . . . . . . . . . . . . . . 101
15.4 Verification of Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
15.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
 Part I
Mathematical Model of Network System
Chapter 1
Introduction

1.1 Network System

The electricity is an important and widely used product of the today’s civilization.
The electrical power technology, electronics, information technology, and various
kinds of electronic utilities are based on the application of electrical networks.
The term “electrical system” has been used to describe a vast range of physical
or abstract systems which differ in appearance, function, and size. In this book, the
electrical network system, called simply network system, is concerned. It has fol-
lowing characteristic features:
1. Network systems are concerned with the generation, transportation, transfor-
mation, and utilization of medium what is called generally “electricity.” Physical
medium is the electric charge, identified sometimes with electrical power and
energy.
2. Network systems involve a number of interconnected routes, along which the
medium is generated, transported, transformed and utilized, and involve the
geographically distributed points, which interconnect these routes. This structure
is visualized geometrically by a graph whose branches represent the routes and
whose nodes represent interconnection points.
3. The routes are called the network system branches, and the points are called the
nodes or the terminals. The network branches represent parts of the network
system which are technological subsystems, earth, and air.
4. The network system can be physically described and mathematically modeled
by following quantities: current, voltage, electrical energy, electrical power, and
electrical parameter. These physical quantities are interrelated with each other
and with the topological structure of network graph. The basic interrelations are
called the network laws.

© Springer International Publishing AG 2017 3

A. Kłos, Mathematical Models of Electrical Network Systems,
Lecture Notes in Electrical Engineering 412,
DOI 10.1007/978-3-319-52178-7_1
4 1 Introduction

5. The theory of networks is still evolving basing on the matrix theory. The main
aim of this book was to widen the theoretical background of network system
knowledge and showing the examples of possible practical application of this
knowledge.

1.2 Modeling Network System

The term “mathematical modeling” is often used in physics and other fields to
describe various objects, events, and actions. The term, mathematical modeling of
electrical network system, used in this book means the description of the physical
states and working of an electrical network system, using mathematical notions and
expressions to solve the theoretical and practical network problems.
The main step in the formulation of the mathematical models of network systems
is to replace the real network by a physical model of the network, which possesses
the major characteristics of the original network but in abstract form. Such an
abstract form is the starting point to the mathematical model of network. It is
assumed that the reader has the knowledge and ability to formulate the abstract
model of electrical network. This book starts from such model, which is referred
hereafter as the mathematical model of network system. The mathematical model of
network system consists of two sets as follows:
– The set of mathematical quantities representing and describing the network sys-
tem. Each of these quantities represents (is modeling) a physical network quantity
and consist of the given data concerning the topology of network, voltages,
currents, powers, and branch parameters.
– The set of relations between mathematical quantities representing the physical
network laws, and other relations, all expressed in the form of equations,
inequalities, and logical expressions.
Network system models are classified into different types depending on the kind
of data and assumed relationships, using the following categories:
(a) A network system model is deterministic if the data and assumed relationships
are known to be correct with no significant errors. The model is non-deter-
ministic if there are some degrees of uncertainty associated with some of data or
relationships. A non-deterministic model is called probabilistic if the data are
treated as random variables with certain probability distributions.
(b) A network system model is linear if the relations between mathematical
quantities are linear, and nonlinear if among them there are nonlinear relations.
The relations in nonlinear model may be continuous or may be step functions.
(c) A network system model is dynamic or time-independent if the mathematical
quantities are of dynamic character or are fixed in time, and dynamic or
time-dependent if they are of dynamic character or varies in time.
1.2 Modeling Network System 5

The above classification subdivides the models of network system into eight main
classes according to the eight subsets of properties formed by the two major choices
in each of (a), (b), and (c) classifications. This is a general classification which does
not include models for some special kind of problems, such as those which are mixed
with respect to this classification. For example, the important technique of “state
estimation” is concerned with finding a “best possible” approximate solution for
networks where the data may be incomplete, inaccurate, and contradictory.
In each of the above main classes, the problem can be further subclassified
according to the amount and kind of information (data) available. For each of the
data categories, topology data, state of network data, and current–voltage relation
data, the given data may include either complete information or incomplete, partial
information.
In what follows the basic linear and continuous mathematical model of the
interconnected, closed electrical network system (called in what follows the
mathematical model of network system) is derived. Such model can be used as a
part in the modeling of any other kind of network models.

1.3 Solving Network System

It is quite usual to encounter the problem of “solving a network system,” though

this may mean many different things. In this book, network solution problems of
general form are considered. They are classified in manner most useful for the
solution process as follows.
Suppose certain information about a network is given, or is assumed, which is of
a quantitative nature, and suppose that certain quantitative information about the
network is required. If the given information (input data) does not include the
required information, or if the information is given but is ambiguous or incorrect,
then arises the obvious problem of finding or estimating the required information. It
is assumed here that the required information can be found by calculations using the
given data rather than by other methods, such as measuring the quantities. In the
following, the term solution of the electrical network system is used to denote both
the set of calculated values of the required information and also the process of
calculation involved.
The definition given here for “solution of the electrical network” is very general,
since it allows many possible pairings of “information given” with “information
required.” In order to keep the situation reasonable, it is further assumed that the
information required about a network is complete, or nearly complete. More
specifically, it is supposed that the information is required on each of the following
which is complete as possible:
(a) The network topology. The number and the interconnections of all branches and
the subdivision of branches into a tree and a cotree. This is equivalent to
knowing the network graph.
6 1 Introduction

(b) The electrical state of network. The numerical values of all branch currents and
branch voltages, or in other words, complete knowledge of the network cur-
rents, voltages, and other required information.
(c) The interrelations of current and voltages (numerical values of all branch
parameters), or in other words the admittance or impedance branch coefficients
and the values of voltage and current sources.
The network system is regarded as solved if the above information is known, and
the term of “solving the network” is that of finding the above information when it is
not part of known data.

1.4 Network Laws

The values of branch currents and branch voltages in the network system are
restricted in the sense that the states of network which are physically possible must
fulfill the fundamental network laws.
The Kirchhoff’s Current law: The sum of branch currents in a node (cut-set),
taken into account the directions of branch currents is equal to zero. Note that a
current directed away from a node is considered to be the same that the current
directed towards the node which has the same numerical value but opposite sign.
The law, stated in simple terms means that the current flowing into the node
(cut-set) equals the current flowing out of it.
The Kirchhoff’s Voltage law: The sum of all branch voltages in a loop-(loop-set)
taken into account the directions of branch voltages is equal to zero. This law stated
in simple terms, mean that the voltage value between any two nodes of network is
independent of the loop (path of branches) between these nodes along which the
voltage is measured.
The Ohm’s law: Generally, it is certain linear function between the currents and
the voltages of network branches. This law has a non-topological nature and is
reflecting the interaction in a physical network between the currents/voltages and
the physical structure of the branches of the network. The interaction of Ohm’s law
is done using constant branch coefficients of current/voltage functions. Note that
each real network, called in this book the network system, must include at least one
current or voltage source. Without any source, the network is empty (I = 0, V = 0)
and the classical network Ohm’s law in such network can be used to the parts of
network only. The Ohms law of network system, derived in this book, refers to the
network system.
The Power law: The sum of powers in all branches of network is zero, where the
power in a branch is defined as a product of the branch current and branch voltage.
Note that these values may be positive or negative according to their relation to the
branch direction. This is equivalent to saying that each set of all branch voltages
1.4 Network Laws 7

written as a vector is orthogonal to any corresponding set of all branch currents,

written as a vector (that is the scalar product of two vectors is zero). The law is
equivalent to the property that each cut-set of network has an even number of
branches in common with each loop-set. This in turn can easily be shown to follow
automatically from the definition used here for cut-set and loop-set, so then in fact
the power law is a consequence of the current and voltage laws.
Chapter 2
Basic Notions

2.1 Network Graph Notions

Classical modeling of network systems uses the graph theory notions to define the
graph of network. The following fundamental notions are used:
The branch (edge) is a basic axiomatic notion and is defined as any element,
represented by a line segment, with specified two different ends: terminals plus (+)
and terminal minus (−) designating positive branch direction from terminal minus
to terminal plus.
The node (vertex) is a basic axiomatic notion and is defined as a point desig-
nating one or more branch terminals.
The connection or interconnection of two or more branches means that they have
one common terminal.
The planar graph is a graph, which if drawn in planar plane has not crossing
branches.
The tree is defined as any minimal set of interconnected branches, called tree
branches, which contain all nodes in network system. From the definition of a tree,
it follows that it is an open subgraph of a network system graph.
The cotree is defined as a set of branches, which does not belong to any tree. The
sets of tree branches and cotree branches are complementary if they are disjoint.
Cotree is a dual concept, in any sense, to the tree.

2.2 Electrical Network Notions

Graph theory provides an excellent tool for visualizing and developing the model of
network graph. However, graph theory is of little use in modeling the physical
states of the network system and interrelations of branch currents and branch
voltages. Furthermore, even representation of network by a graphical model has

© Springer International Publishing AG 2017 9

A. Kłos, Mathematical Models of Electrical Network Systems,
Lecture Notes in Electrical Engineering 412,
DOI 10.1007/978-3-319-52178-7_2
10 2 Basic Notions

serious drawbacks from the point of view of development of the mathematical

model of network. Hence, graph theory is used as an aid in developing the model.
In what follows, the final model presented is algebraic in character and is expressed
in terms of linear algebra. Taking into account the above definitions, the basic
notions used in mathematical modeling of network system are defined as follows:
The directed electrical branch (called branch in what follows) is a basic notion
and is defined as any electrical element (object or earth or air), represented by a line
segment, which has two specified end terminals: plus terminal and minus terminal,
designating branch direction.
The electrical network system (called network system) is a set of connected
branches, which are electrically interconnected and electrically active. Electrical
interconnection means that the electrophysical quantities of all branches (e.g., branch
currents) are dependent on each other. In other words, in network system, the
electrophysical quantity of every branch (e.g., branch current) depends on electro-
physical quantities of all other branches, and the system has the ability to be active
(possesses electricity source).
Electrical network system is closed if each electrical branch is connected at least
with two or more other branches and is open if there are branches connected with
one terminal only to the other branches.
The branch current is a quantity, which characterizes the amount of medium
(electric charge) transmitted along the branch in a unit time interval. It can be
measured at any one point of branch, assuming that there are no losses or dissi-
pation of the medium in the branch.
The branch voltage is a quantity, which characterizes the difference of electric
“pressure” or “stress” being exerted and measured on the branch between plus
terminal and minus terminal.
The above quantities are fundamental notions used in the mathematical modeling
of electrical systems. In order to derive the algebraic relations precisely, it is nec-
essary to define the algebraic sign of the branch current and branch voltage values
in relation to the branch direction. The choice of sign is optional. In what follows,
we assume:
Branch current has plus sign if current is flowing from branch terminal plus to
branch terminal minus. Branch voltage has plus sign if the voltage value measured
along the branch is increasing in the direction from terminal minus to terminal plus.
The branch parameters are the constant coefficients of branch current as a
function of branch voltage, called branch admittances or the constant coefficients of
branch voltage as a function of branch current, called branch impedances, and are
the constant coefficients of electricity sources: the ideal current sources and the
ideal voltage sources.
Real electrical branches are often combinations of the above main kinds, e.g., in
power system analysis, there are sending (generating) branches, which are com-
bination of ideal voltage sources and impedances and receiving (load-consuming)
branches, which are treated as current sources and admittances.
The node-set is defined as a set of branches connected together to one common
terminal (node of a graph). The directed node-set can be defined as a set of network
2.2 Electrical Network Notions 11

branches, which connect one node with the rest of network. The branches defining a
directed node-set are connected to the node (branch terminal) either by the terminal
plus or by the terminal minus. So there are two options and consequently two
possible definitions of the direction of node-set. In the first option, the direction of
node-set is the same as the direction of branches directed toward common terminal,
and in the second option, it is the same as the direction of branches directed away
from the common terminal. From the definition of the node-set, it follows that the
sum of all node-sets is equal to zero. This follows because each branch is repre-
sented in one node-set by plus terminal and in another node-set by minus terminal.
Removing any one node-set in the network must leave a sum of the remaining
node-sets not equal to zero. The number of all node-sets minus any one is denoted
by n and is a number of linearly independent node-sets.
The cut-set is defined as a set of network branches connecting two parts of
network. It is the generalization of node-set. Note that a node-set is also connecting
two parts of network, because it connects one node with the rest of network, so is a
special kind of cut-set. Graphically, cut-set is a set of branches cut by a line
disconnecting two parts of network. From the definition of a cut-set, it follows that
the number of cut-sets is equal to the number of all subsets of the n + 1 node-sets,
excluding the empty subset and the whole subset, so the number of cut-sets is equal
to 2n+1 − 2.
The path of branches is intuitively defined as any set of two or more branches
and nodes (branch terminals) connected in series (one branch after another). The
number of nodes is equal to the number of branches plus one. The path of branches
is closed if two ends of path terminals are one terminal. The direction of path is
optional and may be defined equal to the direction of any branch in path.
The loop is generally defined as a closed path which is not crossing itself.
Intuitively, it is a loop or a mesh in planar networks.
Chapter 3
Algebraic Model of Network Graph

The important aspect of the mathematical modeling of network system is the rep-
resentation of electrical network graph in terms of mathematical quantities (scalars,
vectors, matrices, linear spaces), which represent the physical quantities of the
network system graph (branches, node-sets, cut-sets, loop-sets) and algebraic
relations representing this mathematical quantities. In this chapter, the electrical
network graph is modeled in terms of linear algebra. The physical quantities and
relations are modeled using the b-dimensional Euclidean linear space A, span by the
orthonormal basis of vectors a1 , a2 … ab .

3.1 Cut-Set and Loop-Set Vectors

Consider the electrical network graph of b-directed branches. Let us assume that
branches are numbered from 1 to b. To each axis of the linear space A, the electrical
network branch j is assigned (see Fig. 3.3).
The branch vector, denoted by bj (index j is a number of branch), is modeled as a
b-dimensional vector with one nonzero element equal to plus one in place of branch
number, e.g., a branch vector number 3 is written as follows:

b3 ¼ ½01 02 13 . . .0b1 :0b T ð3:1Þ

where 13 denotes a real network branch number j = 3.

The cut-set vector Cs is defined generally as a b-dimensional vector, which is the
sum of branch vectors, which form the directed cut-set.
X
Cs ¼ bc ð3:2Þ

where bc are the branch vectors belonging to the cut-set vector.

© Springer International Publishing AG 2017 13

A. Kłos, Mathematical Models of Electrical Network Systems,
Lecture Notes in Electrical Engineering 412,
DOI 10.1007/978-3-319-52178-7_3
14 3 Algebraic Model of Network Graph

The node-set vector is a special kind of a cut-set vector, which is a sum of branch
vectors, which form the directed node-set.
The set of linearly independent node-set vectors is a set of all minus any one
node-set vector in the network. In what follows, the node-set vectors are identified
with the cut-set vectors. In what follows, any set of cut-sets may include the
node-sets and the cut-sets.
The loop-set is defined in Chap. 2 as a set of branches forming a closed path.
The loop-set vector Ls can be intuitively defined as a b-dimensional vector,
which is the sum of branch vectors forming the directed loop-set.
X
Ls ¼ bl ð3:3Þ

where bl are branch vectors belonging to the loop-set vector.

However, the loop-set vectors are in a certain sense a dual concept to that of the
cut-set vectors. Consequently, in terms of linear algebra, the loop-set vector is a
b-dimensional vector defined as follows:
– Its components are all equal to +1 or −1 or 0;
– It is orthogonal to every cut-set vector.
So the definition (3.3) is interpreted as follows: Any vector of components
1, −1, 0, whose inner product (scalar product) with every cut-set vector is zero, is
the loop-set vector.

3.2 Topology of Network

The network analysis and solution are based on finding the linearly independent
loop-sets and cut-sets. The loop-sets and cut-sets are fundamental notions in the
mathematical modeling of network systems, because they are used in formulations
of the Kirchhoff’s laws. In particular, a set of m linearly independent loop-set vectors
and a set of n linearly independent cut-set vectors are used in formulations of the
current and voltage relations. Taking into account the great number of possible
cut-sets and loop-sets in real networks, the finding of linearly independent sets may
be very difficult. Both sets can be easily found after defining network topology. The
term “topology” relates to the graph of network but may have various meanings. In
what follows, let us define the topology of electrical network as follows:
The topology of electrical network is defined as a subdivision of the b-branch
network graph into the tree (any set of n tree branches) and the complementary
cotree (a set of m complementary (not belonging to tree) branches.

n þ m ¼ b: ð3:4Þ

Generally, there are many topological tree–cotree subdivisions of a given net-

work graph. The choice of any subdivision means that the network is topologically
3.2 Topology of Network 15

oriented. The algebraic operations, using the b-dimensional vectors and matrices,
require the ordering of branches to be appropriate to the chosen topology.
Generally, the ordering is not limited. Taking into account the ordering which is
often used in practice, we assume that the first m rows and columns of vectors and
matrices are associated with the cotree branches and the last n rows and columns of
vectors and matrices are associated with the tree branches.
The vector of branch numbers, denoted by Bj , is subdivided into two subvectors:

Bj ¼ ½ Bm Bn  ð3:5Þ

where
Bm vector of units, of cotree branches.
Bn vector of units, of tree branches.
In the next sections, the topological models of loop-sets and cut-sets are derived.

3.3 Topological Model of Loop-Sets

Each cotree branch generates a unique loop-set. As it follows from the definition of
a tree/cotree topology, the tree is a subgraph, which includes all network nodes.
Consequently, every cotree branch connects two nodes, each of them belonging to a
tree. The tree is open subgraph; so for every cotree branch, there must exist a set of
tree branches which, together with this cotree branch, form a closed path in the
network.
The topological loop-set vector generated by a cotree branch (called in what
follows cotree loop-set vector or loop-set vector), denoted Lc , is defined as the
following sum of branch vectors: one cotree branch vector and the unique set of tree
branch vectors, which form together a closed path in the network:
X
Lc ¼ bc ð3:6Þ

where bc are branch vectors belonging to the cotree loop-set vector.

Direction (arrow) of a cotree branch, which generates the loop-set vector, defines
the direction of this loop-set vector. So all the tree branches belonging to loop-set,
if they have the same direction as the direction of co-tree branch which generate the
loop-set, they have +1 in loop-set and if they have the opposite direction they have
−1 in the loop-set (see Fig. 3.1).
Every cotree branch generates a loop-set vector, so in any topological structure
of network, there are m topological loop-set vectors.
The topological loop-set matrix (called in what follows loop-set matrix), denoted
L, represents a set of m linearly independent topological loop-set vectors. Each row
of matrix L is the topological loop-set vector.
16 3 Algebraic Model of Network Graph

5
4 3 2 7 1 6 5
1 -1 -1 1 1 4
6 3 4 L= 1 -1 1 3
2
1 -1 1 1 2

1 7 Lm Ln
Tree Cotree Loop-set matrix L

Fig. 3.1 Example of loop-sets and loop-set matrix L

L ¼ ½ Lc;1 Lc;2 Lc;3 : Lc;k : Lc;m T ð3:7Þ

where Lc;k are the loop-set vectors (rows of matrix L).

The linear independency of m loop-set vectors follows from the definition of
loop-set vector. A loop-set vector cannot be any sum of remaining loop-set vectors
because each sum must include two or more cotree branches what contradicts with
the definition of loop-set vector. Loop-set matrix is of order m  b and of rank
m. Generally, any loop-set incidence matrix can be subdivided after appropriate
rearrangement of columns, into two submatrices:

L ¼ ½ Lm Ln  ð3:8Þ

where Lm is a non-singular matrix of order m  m and rank m and Ln is of order

m  n.
Taking into account the topological ordering of branches (see Eq. 3.5) and that
the loop-sets are generated by cotree branches, the particularly useful form of
loop-set matrix L is as follows (see Fig. 3.1):

L ¼ ½ 1m Ln  ð3:9Þ

where
Ln is the tree–cotree incidence matrix of order m  n.
1m is a unit matrix of order m  m of m cotree branches.
The matrix Ln is a well-known incidence matrix relating tree branches to cotree
branches and is used frequently in the network analysis and solution.

3.4 Topological Model of Cut-Sets

The tree is any open set of branches in network. Each tree branches generate a
cut-set generate a cut-set. The cut-set is a set of branches including one set of cotree
branches which together fulfill the current law. Visually, the cut-set can be illus-
trated on a graph of network by dotted line (not full line) which is cutting the cur-set
branches.
3.4 Topological Model of Cut-Sets 17

5 4 3 2 7 1 6 5
1 1 7
3 4 1 1 1 1 1
6 2 C = -1 -1 -1 1 6
-1 -1 1 5

1 7 Cm Cn
Tree Cotree Cut-sets Cut-set matrix C

Fig. 3.2 Example of the cut-sets and cut-set matrix C

E.g., in Fig. 3.2 the cut-set generated by tree-branch 1 is cut (include) the
branches 1, 2, 3, and 4. Algebraically such set of branches is used in form of a
cut-set vector, and a set of all cut-sets in network is used, in form of a cut-set matrix
(see Fig. 3.2).
The cut-set vector generated by a tree branch (in what follows called tree cut-set
vector or cut-set vector), denoted Ct , is defined as a sum of the following branch
vectors: one tree branch vector and a set of cotree branch vectors subdividing the
network into two subnetworks.
X
Ct ¼ bt ð3:10Þ

where bt are the branch vectors belonging to the tree cut-set vector.
Direction (arrow) of a tree branch that generate the cut-set defines the direction
of cut-set vector. So the all cotree branches belonging into cut-set if they have the
same direction as direction of the tree branches then have number +1 in cut-sector;
if they have opposite direction then have number −1 in cut-set vector (see Fig. 3.2)
Each tree branch generates the cut-set vector, so in any network there are an
independent cut-set vectors.
The cut-set matrix, denoted C, represents a set of n linearly independent cut-set
vectors.

C ¼ ½ Ct;1 Ct;2 Ct;3 ... Ct;k ... Ct;n T ð3:11Þ

where each Ct;k is a cut-set vector (row of matrix C).

The linear independency of n cut-set vectors follows from the definition of
cut-set. A cut-set vector cannot be any sum of remaining cut-set vectors because
each sum must include two or more tree branches what contradicts with the defi-
nition of cut-set. Cut-set matrix is of order n  b and of rank n. Generally, any
cut-set incidence matrix can be partitioned, after appropriate reordering of columns,
into two submatrices:
C ¼ ½Cm jCn  ð3:12Þ

where Cn is a non-singular matrix of order n  n and rank n; Cm is of order n  m.

18 3 Algebraic Model of Network Graph

Taking into account the topological ordering of branches has (see Eq. 3.5), the
branches associated with columns of cut-sets matrix C are ordered as follows: first
m branches are the cotree branches and last n are the tree branches. So the matrix
C is partitioned as follows:

C ¼ ½Cm j1n  ð3:13Þ

where
Cm the tree incidence matrix of order n  m and
1n is a unit matrix of order n  n representing tree branches.
Matrix C is illustrated in Fig. 3.2. The matrix Cm is a well-known incidence
matrix relating cotree branches to the tree branches and is used frequently in the
network analysis and solution.

3.5 Orthogonality of Cut-Sets and Loop-Sets

From the definition of loop-set vectors (3.6) and cut-set vectors (3.10), it follows
that they are orthogonal to each other.

Ct LTc ¼ Lc CtT ¼ 0 ð3:14aÞ

It means that the loop-set matrices (3.8 and 3.9) and cut-set matrices (3.12 and
3.13) are orthogonal to each other.

CLT ¼ LC T ¼ 0 ð3:14bÞ

The orthogonality of L and C reflects what is called the duality of network

topology, which is an important feature of electrical networks. The various prac-
tically useful interrelationships can be derived using orthogonality of cut-sets and
loop-set matrices. Equations (3.14a, b) make possible finding matrix C if matrix
L is known or finding matrix L if matrix C is known. Such relations can be derived
as follows:
Using orthogonal relation of matrix C in the form C ¼ ½Cm Cn  and matrix L in
the form L ¼ ½ 1m Ln .
 
1m
½Cm Cn  ¼0 ð3:15Þ
LTn
3.5 Orthogonality of Cut-Sets and Loop-Sets 19

Cm 1m þ Cn LTn ¼ 0

LTn ¼ Cn1 Cm

Hence,
 
L ¼ 1m ðCn1 Cm ÞT ð3:16Þ

what means that matrix L can be found using matrix C

Using orthogonal relation of matrix C in the form C ¼ ½Cm 1n  and
L ¼ ½ Lm Ln .
 
CmT
½Lm Ln  ¼0 ð3:17Þ
1n

Lm CmT þ Ln 1n ¼ 0

CmT ¼ L1
m Ln

Hence,
 
C ¼ ðL1
m Ln Þ
T
1n ð3:18Þ

What means that matrix C can be found using matrix L

Using orthogonal relation of matrices C ¼ ½ Cm 1n  and L ¼ ½ 1m Ln .
 
1m
½Cm 1n  ¼0 ð3:19Þ
LTn

Cm 1n þ 1m LTn ¼ 0

Hence,

Cm ¼ LTn ð3:20Þ

Ln ¼ CmT ð3:21Þ

The relations (3.20 and 3.21) are very useful in practical applications.

3.6 Linear Space Model of Network Topology

From the engineering point of view, it would be of some advantage to visualize the
topological models, described above, in a linear space A. Visual illustration of linear
20 3 Algebraic Model of Network Graph

2
a2
C1.2 C 3,12
L2

L
1 3

b2 Tree Cotree
L2
C3,2 Branches 2 1 3
Branch vector b2 = [1 0 0]
b1 T

Branch vector b1 = [0 1 0]
T
C a1
b3 C1,2 Branch vector b3 = [0 0 1]T
Loop-set vector LT2 = [1 1 − 1]T
Cut-set vector C1T, 2 = [− 1 1 0]T
Cut-set vector C 3T, 2 = [1 0 1]T
a3
Loop-set subspace L = Line L3
Cut-set sub-space C = Plane C1,3 C 2,3

Fig. 3.3 The three-dimensional linear space illustration of subspaces L and C

space model of network topology can be done for very simple, 3-branch network
only (see Fig. 3.3). In order to visualize the network topology, in terms of linear
algebra, the loop-set and cut-set vectors (the rows of the matrices L and C) are
transposed into column matrices LT and C T . Generally, the loop-set and cut-set
vectors and matrices can be modeled as a linear subspaces of space A (see Fig. 3.3).
The set of m loop-set vectors L (in Fig. 3.3 one vector L3 ) form the loop-set
subspace L of space A. Analogically, the set of n cut-set vectors (in Fig. 3.3 vector
Cv1 and Cv2 ) form the linear cut-set subspace C of space A. The subspaces L and C
are orthogonal to each other. They are graphically illustrated in Fig. 3.3 for a simple
example of 3-branch network.

3.7 Topological Transformation

The above-derived linear space model leads to a new algebraic structure. The
cut-set subspace and the loop-set subspace are orthogonal to each other. The
m loop-set vectors (matrix LT ) and the n cut-set vectors (matrix CT ) are, so far,
vectors in space A (let us call it the old space). Note that the loop-set vectors and the
cut-set vectors span a new b-dimensional space (let us call it the new space (see
Fig. 3.3). Consequently, in the new space, the loop-set and cut-set vectors are the
basis of new space, so they are unit vectors in this space. It means that in the old
3.7 Topological Transformation 21
   
1m CmT
space, the loop-set matrix L ¼ T
and the cut-set matrix C ¼ T
are
LTn 1n
modeled by a joint matrix:
" #
  1m CmT
LT C T
old
¼ : ð3:22Þ
LTn 1n

In the new space, the loop-set and cut-set matrices are modeled by the joint unit
matrix:
 
  1m 0
LT CT ¼ ð3:23Þ
new 0 1n

Taking into account Eqs. (3.22 and 3.23), the transformation equation of
loop-sets and cut-sets from new to old space is as follows:
 " # " #
1m 0 1m CmT 1m CmT
¼
0 1n LTn 1n LTn 1n

It means that the

 transformation
 matrix of loop-sets and cut-sets from new to old
space is equal to LT C T .
Note that interpreting physically, the above transformation means transformation
of m cotree branches into m loop-sets, which are defined by this m cotree branches;
and means transformation of n tree branches into n cut-sets, which are defined by
this n tree branches (see Sects. 3.3 and 3.4).
The following definitions are to be formulated:
The new space is the topological space. It is the b-dimensional linear space,
denoted T, in which unit basis vectors indicate the cotree and tree topological
structure of network system.
 
If transformed to space, A (old space) is equal to matrix LT C T .
The topological transformation matrix denoted T (in what follows called topo-
logical matrix T or matrix T) is transformation matrix of vectors in space T into
vectors in space A.
" #
  1 CmT
T ¼ LT C T ¼ ð3:24Þ
LTn 1

The topological transformation matrix T is an important quantity, which makes

possible widening the analysis of network systems. It will be often used in the next
chapters.
22 3 Algebraic Model of Network Graph

3.8 Topological Matrix T

The topological transformation matrix T takes the simplest and practically useful
forms, based on the well-known incidence matrices CmT and LTn .
Substituting Eqs. (3.20–3.24)
 
1 CmT
T¼ ð3:25Þ
Cm 1

and substituting Eqs. (3.21–3.24)

 
1 Ln
T¼ ð3:26Þ
Ltn 1

The topological matrix T has some interesting properties:

– For a given network, there are as many matrices T as there are pairs of: n
independent cut-sets (nodes) and m independent loop-sets (meshes).
– All matrices T are similar and have the same nonzero integer determinant. Note
that it comes from the linear independency and orthogonality of incidence
matrices C and L.
– The matrix T is non-singular because the columns of LT and columns of C T are
linearly independent and orthogonal to each other.
– The matrix T 1 is not a (0, 1, −1) incidence matrix. The values of its entries are
greater than −1 and are less than 1.
– The matrix T T T describes the structure of loop-sets and cut-sets, e.g., the lower
right submatrix of a matrix T T T shows the structure of cut-sets as follows: Each
row shows the relation of a cut-set generated by tree branch j to the remaining
cut-sets. The diagonal element of row is the numbers of branches belonging to
the cut-set generated by tree branch j. The off-diagonal elements are the numbers
of branches, belonging to the remaining cut-sets, which are common with
branches of cut-set generated by branch j. The plus/minus sign of the
off-diagonal elements denotes the same (plus) or opposite (minus) direction of
branch in cut-set j and the directions of branch in remaining cut-sets. The
direction of a cut-set is defined by the direction of tree branch generating this
cut-set.
– Analogically the upper left submatrix of T T T describes the structure of a set of
loop-sets.
Figure 3.4 illustrates the matrices T, T 1 , and T T T for a simple example of
network.
Matrix T is a powerful tool enabling the widening of electrical network analysis
and methodology of network solution. The algebraic model of network developed
in the next chapters is based on the application of matrix T. Note that in the
3.8 Topological Matrix T 23

1 2 3 5 9 4 6 7 8
6 7 1 0 0 0 0 1 1 0 0 1
5
9 0 1 0 0 0 1 1 1 1 2
0 0 1 0 0 1 1 1 0 3
8 0 0 0 1 0 0 1 1 0 5
T= 0 0 0 0 1 0 0-1-1 9
4 3 2 1
-1-1-1 0 0 1 0 0 0 4
Tree -1-1-1-1 0 0 1 0 0 6
0-1-1-1 1 0 0 1 0 7
Cotree 0-1 0 0 1 0 0 0 1 8
DeterminantT=75

1 2 3 5 9 4 6 7 8 1 2 3 5 9 4 6 7 9
45 -15 -15 0 -15 -15 -15 15 0 1 3 2 2 1 0 0 0 0 0 1
-15 40 -15 -5 20 -10 -5 0 –20 2 2 5 3 2–2 0 0 0 0 2
-15 -15 45 -15 0 -15 0 -15 15 3 2 3 4 2–1 0 0 0 0 3
0 -5 -15 40 5 20 -20 -15 10 5 1 2 2 3-1 0 0 0 0 5
T –1 = 1/75 -15 20 0 5 40 -5 -10 15 20 9 T TT = 0 2–1-1 3 0 0 0 0 9
15 10 15 -20 5 35 -20 0 -5 4 0 0 0 0 0 4 3 2 1 4
I5 5 0 20 10 -20 35 -15 5 6 0 0 0 0 0 3 5 3 1 6
-15 0 15 15 -15 0 -15 30 –15 7 0 0 0 0 0 2 3 5 2 7
0 20 -15 -10 -20 -5 5 -15 35 8 0 0 0 0 0 1 1 2 3 8

Fig. 3.4 Topological matrices T, T 1 , and T T T for a simple example of network

above-presented algebraic model of network, the primary notion (not to be math-

ematically defined) is the network branch only. Different than that in graph theory,
the node is defined as a set of branches. The application of the above formulas does
not need introducing the notion of slack node.
Chapter 4
Algebraic Model of Network Currents

4.1 Matrix Model of Network Currents

The linear algebra model of network currents can be derived starting from the
model of branch current. Consider the network formed from b electrical branches,
numbered from 1 to b. To every branch, the physical quantity, called the branch
current, denoted ij (j is the branch number) is associated, which may have a real or
complex value. The branch current ij has positive sign if its direction of flaw is the
same as the direction of a network branch and has negative sign otherwise.
The current vector of network denoted I is defined as a column vector of branch
currents.

I ¼ ½ i1 i2 i3 . . . ib T ð4:1Þ

The elements of current vector satisfy fundamental Kirchhoff’s current law

which ensures that for every cut-set the sum of currents, over all branches belonging
to this cut-set, is equal to zero. If the ordering of branches is not restricted, then the
current law is expressed in matrix form as follows:

Cs I ¼ 0 ð4:2Þ

where Cs is any possible row cut-set vector, which has the same branch ordering
then vector I.
If the network topology (tree/cotree) is defined, not loosing generality, then we
assume that the set of n cut-sets, generated by a tree, are the rows of a cut-set
matrix, denoted in what follows by C . The ordering of columns in this matrix is as
follows: The first m columns are associated with cotree branches and the last n with
tree branches. Using such topology, the current law is expressed as follows (see
Chap. 3, Sect. 3.4):

© Springer International Publishing AG 2017 25

A. Kłos, Mathematical Models of Electrical Network Systems,
Lecture Notes in Electrical Engineering 412,
DOI 10.1007/978-3-319-52178-7_4
26 4 Algebraic Model of Network Currents

CI¼0 ð4:3Þ

where C ¼ ½ Cm 1n  is cut-set matrix generated by a tree.

The ordering of branches in vector I is as follows:


I
I¼ m ð4:4Þ
In

where
Im is the current state vector of order m. It is the current vector of cotree branches
In is the current vector of tree branches
Vector In can be found using current state vector from Eq. (4.3) written as
follows.


I
½ Cm 1n  m ¼ Cm Im þ 1n In ¼ 0 ð4:5Þ
In

where Cm is the tree–cotree incidence matrix of order n  m

Solving Eq. (4.5) for In gives
In ¼ Cm Im ð4:6Þ

The current of tree branches In can be also found, using loop-set matrix L.
Substituting Cm from Eq. (3.20) into (4.6), we have:

In ¼ LTn Im ð4:7Þ

Finally substituting Eq. (4.7) into (4.4), we have:

I ¼ LT Im ð4:8Þ

The current vector of cotree branches Im is the current state vector. Knowing
vector Im , the current vector of tree branches can be found from Eqs. (4.6) and
(4.7), called the current state equations. Equation (4.8) is the particular expression
of current law Equation (4.3). Note the following difference between the formu-
lations of the above equations: the current law (4.3) and current state Equation (4.6)
are expressed using the tree cut-sets, but the same law in Eq. (4.8) and current state
Equation (4.7) is expressed using the cotree loop-sets.

4.2 Linear Space Model of Network Currents

In this section, the currents of network system are expressed using the linear algebra
model. The algebraic model of the current relationships is derived using the linear
space model of network topology described in Sect. 3.6.
4.2 Linear Space Model of Network Currents 27

Let us model the currents of a b-branch electrical network using the b-dimen-
sional linear space A.  The set of all b-dimensional branch current vectors
 and
I satisfying Eq. (4.3) for any cut-set matrix C is a linear subspace of the space A
is called the current space I. From Eq. (4.3), it follows that the current subspace I is
orthogonal to the cut-set subspace C, and the subspace C is orthogonal to the
loop-set subspace L. It means that subspace I coincides with the loop-set subspace
L. Taking into account this coincidence, the subspace I is spanned by the basis of
m loop-set vectors L (see Sect. 3.6).
The linear space model of network currents can be illustrated by a very simple
example of 3-branch network in Fig. 4.1. Following topology of network is chosen:
Branches 1 and 3 are the tree and branch 2 is a cotree. The cut-set vectors defined
by m = 2 tree branches are cut-set vector C1 (generated by tree branch 1) and
cut-set vector C3 (generated by tree branch 3), which span the two-dimensional
cut-set subspace C (plane). The loop-set vector defined by n = 1 cotree branch is
loop-set vector L2 (generated by cotree branch 2) which spans the one-dimensional
loop-set subspace L Assuming the values of branch currents as i1 ¼ i2 ¼ 1 and
i3 ¼ 1, the current vector I ¼ ½ 1 1 1  spans the current subspace I (line).
The current subspace I is orthogonal (perpendicular) to the cut-set subspace C.

2 i2
a2
C1 L,2 C3

i1 i3
I 1 3
L2

i2
-i3 I Tree Cotree
C3
i1
Branch 2 1 3
C1 = [− 1 1 0]T
C a1
C1 C 3 = [1 0 1]T
⎡i2 ⎤ ⎡ 1 ⎤
I I = ⎢⎢ i1 ⎥⎥ = ⎢⎢ 1 ⎥⎥
⎢⎣i3 ⎥⎦ ⎢⎣− 1⎥⎦
a3 ⎡1⎤
⎡ − 1 1 0 ⎤ ⎢ ⎥ ⎡0 ⎤
CI =⎢ ⎥⎢1⎥=⎢ ⎥
⎣ 1 0 1⎦ ⎢− 1⎥ ⎣0⎦
⎣ ⎦

Fig. 4.1 Linear space model of currents for a simple example of 3-branch network
28 4 Algebraic Model of Network Currents

4.3 Topological Transformation of Current Space

So far the currents of network system are modeled in space A.  Taking into account
that the cut-set vectors and loop-set vectors span the b-dimensional topological
space T (see Chap. 3, Sect. 3.7), the network current vector can also be expressed in
a new basis, namely in basis of topological space T. It means that there exists linear
 As it was shown in Sect. 3.7, the
transformation of vectors from space T to space A.
topological matrix T is the transformation matrix of vectors between space A  and
topological space T. In order to find the current vector I (defined in space A) 
transformed into space T, let us write vector I as follows:




Im 0 I
I ¼ Im;o þ Io;n ¼ þ ¼ m ð4:9Þ
0 In In

where Im;o and Io;n are the cotree and tree branch current vectors of order b.
It can be easily proved that the current vector Im;o after topological transfor-
mation from space T to space A  is the current vector I. Starting from Eq. (4.8)—
I ¼ L Im and substituting into this equation the cotree current vector Im of order
T

m by the cotree current vector Im;o of order b, and by substituting the loop-set matrix
 
LT into the matrix of topological transformation T ¼ LT C T , we have:
 
I ¼ LT C T Im;o ð4:10Þ

Note that:
 
LT C T Im;o ¼ LT Im ¼ I : ð4:11Þ

It means that current vector Im;o is vector of topological space T, and if trans-

formed using matrix T is the current vector I of space A.

I ¼ T Im;o ð4:12Þ

All vectors Im;o form the subspace I m;o of space T. Both vectors I and Im;o are the
same current vector in two spaces: in subspace I of space A  and in subspace I m;o of
space T. Note that vector Im;o is b-dimensional, and because of the orthogonality of
the spaces C and I, vector Im;o cannot involve any linear combinations of the
columns of matrix L, so Im;o has only n nonzero components.
Both the vectors Im;o and Im determine the current state of network, and each
of them are called the current state vectors. They differ by dimension only; Im;o is
b-dimensional and Im is m-dimensional vector.
4.3 Topological Transformation of Current Space 29

Illustrating the relation (4.12), in case

2 of3three-dimensional2 current
3 space in
1 1
Fig. 4.1, let us substitute vector I ¼ 4 1 5, vector Im;o ¼ 4 0 5, and matrix
1 0
2 3
1 1 1
T¼4 1 1 0 5.
1 0 1
Equation (4.12) is as follows:

2 1 3
2 3 2 32 3
1 1 1 1 1
6 7 6 76 7 ð4:13Þ
4 1 5¼4 1 1 0 54 0 5
1 1 0 1 0

I ¼ T Im;o

4.4 Current Relations Using Topological Matrix T

Using the topological transformation matrix T, various interrelations between the

b-dimensional current vectors can be derived. The starting points are Eqs. (4.10)
and (4.12).
Matrix T is non-singular, so the current state vector Im;o can be found using
Eq. (4.12) as follows:

Im;o ¼ T 1 I ð4:14Þ

Interpreting physically the above equation, it means that the set of network
currents, which is dependent on each other (must fulfill the current law), can be
substituted and considered using the subset of these network currents, which are
independent of each other.
A relation between Im;o and Io;n comes as a result of substituting Eq. (4.9) into
(4.12).

Io;n ¼ ðT  1ÞIm;o ð4:15Þ

This is just a b-dimensional version of relation (4.6) or (4.7). Matrix ðT  1Þ is

singular, so if the tree currents are known one cannot find the cotree currents. The
substitution of Eq. (4.10) into (4.14) gives the relation between Io;n and I:

Io;n ¼ ð1  T 1 ÞI ð4:16Þ
30 4 Algebraic Model of Network Currents

If the transposed topological matrix T T is used, then it leads to a interesting

relation.

T T I ¼ ImL ð4:17Þ

where ImL can be called the loop current vector. In order to show that vector ImL has
physical meaning, note that


L
ImL ¼T I¼
T
I ¼ L I; because C I ¼ 0: ð4:18Þ
C

Hence, each entry of ImL is the algebraic sum of the branch currents along the
loop-set defined by a cotree branch. It means that having m numbers, each of them
being the algebraic sum of currents along one of the m loop-sets, one can fined all
branch currents in the network. Inverting Eq. (4.17), we have:

I ¼ ðT T Þ1 ImL ð4:19Þ

Substituting Eq. (4.11) into (4.19) gives the relation between Im;o and ImL .

Im;o ¼ ðT T TÞ1 ImL ð4:20Þ

From Eqs. (4.19) and (4.20), it follows that the current vector ImL uniquely
defines the current state of network. Substituting Eq. (4.19) into (4.16) gives the
relation between ImL and Io;n .

Io;n ¼ ð1  T 1 ÞðT T Þ1 ImL ð4:21Þ

The above-given relation does not exhaust all possible one. Reducing the rela-
tions to the order m and n enables the derivation of various practically useful
equations. As it comes from the above relations using the algebraic model of
electrical network the non conventional relations can be derived, which may be
useful in the network analysis and solution.
Chapter 5
Algebraic Model of Network Voltages

5.1 Matrix Model of Network Voltages

Analogically as in the case of currents, the linear algebra model of network voltages
can be derived starting from the model of branch voltage. The algebraic model of
network voltage relationships is derived using the linear space model of network
topology described in Sect. 3.6. Consider the network formed from b branches. To
every electrical branch, the physical quantity, called the branch voltage vj (j is the
branch number), is associated, which may have a real or complex value. The branch
voltage vj has positive sign if the direction of the positive voltage drop along the
branch is opposite to the direction of a network branch and has negative sign
otherwise.
The voltage vector of the network denoted V is defined as a column vector of
branch voltages.

V ¼ ½v1 v2 v3 . . . vb  T ð5:1Þ

The elements of voltage vectors satisfy the fundamental Kirchhoff’s voltage law,
which ensures that for every loop-set, the sum of voltages, over all branches
belonging to the loop-set, is equal to zero. The voltage law is expressed in the
matrix form as follows:

Ls V ¼ 0 ð5:2Þ

where Ls is any possible row loop-set vector which has the same branch ordering
than vector V.
If the network topology is defined, then, not loosing generality, we assume that
the set of m loop-sets, generated by a cotree, are the rows of a loop-set matrix,
denoted in what follows by L . The ordering of columns in this matrix is as follows:

© Springer International Publishing AG 2017 31

A. Kłos, Mathematical Models of Electrical Network Systems,
Lecture Notes in Electrical Engineering 412,
DOI 10.1007/978-3-319-52178-7_5
32 5 Algebraic Model of Network Voltages

The first m columns are associated with cotree branches and the last n with tree
branches. Using such topology, the current law is expressed as follows:

L V ¼0 ð5:3Þ

where L ¼ ½ 1m Ln  is the loop-set matrix [see Chap. 3, Eq. (3.9)].

The ordering of branches in vector V is as follows:


Vm
V¼ ð5:4Þ
Vn

where
Vn is the voltage state vector of order n. It is the voltage vector of tree branches.
Vm is the voltage vector of cotree branches.
Vector Vm can be found using voltage law Eq. (5.3) written as follows:


V
½ 1m Ln  m ¼ 1m Vm þ Ln Vn ¼ 0 ð5:5Þ
Vn

Solving for Vm gives

Vm ¼ Ln Vn ð5:6Þ

Substituting Ln from Eq. (3.21), Ln ¼ CmT

Vm ¼ CmT Vn ð5:7Þ

Finally, taking into account Eq. (5.4),

V ¼ C T Vn ð5:8Þ

The voltage vector of tree branches Vn is the voltage state vector. Knowing
vector Vn , the voltage vector of cotree branches can be found from Eqs. (5.6) and
(5.7), called the voltage state equations. Equation (5.8) is the particular expression
of voltage law in Eq. (5.3).
Physically it means that every element of voltage vector V is the product of
transposed cut-set matrix and voltage state vector. Note the following difference
between above equations: the voltage law Eq. (5.3) and voltage state Eq. (5.6) are
expressed using tree loop-set matrix but the same kind voltages; voltage law
Eq. (5.8) and voltage state Eq. (5.7) are expressed using tree cut-set matrix.
5.2 Linear Space Model of Network Voltages 33

5.2 Linear Space Model of Network Voltages


In this section, the voltage vectors are modeled in the b-dimensions linear space A.
The set of all b-dimensional branches voltage vectors V satisfying Eq. (5.3) for any
matrix L is a linear subspace V of the space A, and is called the voltage space V.
From Eq. (5.3), it follows that the voltage subspace V is orthogonal to the cut-set
subspace C. It means that subspace V coincides with the subspace C. Taking into
account this coincidence, the subspace V is spanned by the basis of n cut-set vectors
C [see Eq. (3.11)].
The linear space model of network voltages can be illustrated by a very simple
example of 3-branch network in Fig. 5.1. The following topology of network is
chosen. Branches 1 and 3 are a tree, and branch 2 is a cotree. The cut-set vectors
defined by m = 2 tree branches are as follows: cut-set vector C1 (generated by tree
branch 1) and cut-set vector C3 (generated by tree branch 3). Vectors C1 and C3
span the two-dimensional cut-set subspace C (plane). The loop-set vector defined
by n = 1 cotree branch is loop-set vector L2 (generated by cotree branch 2) which
spans the one-dimensional loop-set subspace L (line).
Assuming the values of branch voltages, v1 ¼ v2 ¼ 1 and v3 ¼ 2, the voltage
vector V ¼ ½ 1 1 2  and spans the voltage subspace V (plane). The voltage
subspace V is orthogonal (perpendicular) to subspace L.

2 v2
a2
L2

v1 v3
1 3
L
v2
L ,2 Tree Cotree
C3
v1
V Branch 1 2 3
V a1 L1 = [1 1 − 1] = L
C1 ⎡ v1 ⎤ ⎡1 ⎤
V = ⎢⎢v 2 ⎥⎥ = ⎢⎢1 ⎥⎥
v3

⎢⎣v3 ⎥⎦ ⎢⎣2⎥⎦
⎡1 ⎤
L V = [1 1 − 1] ⎢⎢1 ⎥⎥ = [0]
a3

⎢⎣2⎥⎦

Fig. 5.1 Algebraic model of voltage space V for a simple example of 3-branch network
34 5 Algebraic Model of Network Voltages

5.3 Topological Transformation of Voltage Space

So far, the voltages of network system are modeled in space A. Taking into account
that the cut-set vectors and loop-set vectors span a new b-dimensional linear space
T (see Chap. 3, Sect. 3.5), the network voltage vector can also be expressed in a
new basis, namely basis of topological space T. It means that there exists linear
transformation of vectors from space T to space A.  The transformation matrix is
 As
matrix T (see Sect. 3.5). Voltage vector V is defined in the old basis of space A.
it was shown in Sect. 3.5, the topological matrix T is the transformation matrix of
vectors between spaces A  and T. In order to find the voltage vector V (defined in

space A), transformed into space T, let us write vector V as follows:
In space T, let us write vector V as follows:




Vm 0 Vm
V ¼ Vm;o þ Vo;n ¼ þ ¼ ð5:9Þ
0 Vn Vn

where Vm;o and Vo;n are cotree and tree branch voltage vectors of order b (see
Eq. (5.4))
It can be easily proved that the voltage vector Vo;n after topological transfor-
mation from space T into space A  is the voltage vector V. Starting from Eq. (5.8)
—V ¼ CT Vn and substituting in this equation the tree voltage vector Vn of order
n by the tree voltage vector Vo;n of order b and substituting the cut-set matrix C T by
 
the matrix of topological transformation T ¼ LT C T we have:
 
V¼ LT C T Vo;n ð5:10Þ

Note that
 
LT C T Vo;n ¼ C T Vn : ð5:11Þ

It means that voltage vector Vo;n is the vector of topological space T and if

transformed using matrix T is the voltage vector V of space A.

V ¼ T Vo;n ð5:12Þ

All vectors Vo;n form the subspace Vo;n in topological space T Both vectors V
and Vo;n are the same current vector in two spaces: Vo;n in subspace V of space A 
and Vo;n in subspace Vo;n of space T. Note that vector Vo;n is b-dimensional, and
because of the orthogonality of the spaces L and V, vector Vo;n cannot involve any
linear combinations of the columns of matrix C, so Vo;n has only m nonzero
components. Both the vectors Vo;n and Vn determine the current state of network,
and each of them is called the voltage state vectors. They differ by dimension only,
Vo;n is b-dimensional vector, and Vn is n-dimensional vector.
5.3 Topological Transformation of Voltage Space 35

Illustrating the relation (5.12), in case of three-dimensional current space in

Fig. 5.1, let us substitute into (5.12):
2 3 2 3 2 3
1 0 1 1 1
V ¼ 4 1 5; Vo;n ¼ 4 1 5; T ¼4 1 1 05 ð5:13Þ
2 2 1 0 1

Equation (5.12) is as follows:

2 3 2 2 1 3 32 3
1 1 1 1 0
415 ¼ 4 1 1 0 54 1 5
2 1 0 1 2

V ¼ T Vo;n

5.4 Voltage Relations Using Topological Matrix T

Using the topological transformation matrix T, various interrelations between the

b-dimensional current vectors can be derived. The starting point is Eqs. (4.10) and
(4.12).
Using the topological transformation matrix T, various interrelations between the
b-dimensional voltage vectors can be derived. The starting point is Eqs. (5.10) and
(5.12).
Matrix T is non-singular, so the voltage state vector Vo;n can be found using
Eq. (4.12)

Vo;n ¼ T 1 V ð5:14Þ

Interpreting physically the above equation, it means that the set of network
voltages, which are dependent on each other (must fulfill the voltage law), can be
substituted and considered using the subset of these network voltages, which are
independent of each other.
The relation between Vo;n and Vm;o comes as a result of substituting V of
Eq. (5.9) into (5.12).

Vm;o ¼ ðT  1Þ Vo;n ð5:15Þ

This is just b-dimensional version of relation (5.7). Matrix ðT  1Þ is usually

singular, so if the cotree voltages are known, one cannot usually find the tree
voltages. The substitution of Eq. (5.10) into (5.14) gives the relation between Vm;o
and V:
36 5 Algebraic Model of Network Voltages

Vm;o ¼ ð1  T 1 Þ V ð5:16Þ

An interesting voltage vector can be derived using the transposed topological

matrix T T .

T T V ¼ VnC ð5:17Þ

where VnC can be called the cut-set voltage vector. In order to show that vector VnC
has physical meaning, note that


L
VnC ¼ T T V¼ V ¼ CI since LV ¼0 ð5:18Þ
C

Hence, each entry of VnC is the algebraic sum of the branch voltages of the
cut-set, defined by a tree branch. Physically, it means that having n numbers, each
of them being the algebraic sum of branch voltages of one of n tree cut-sets, one can
find all branch voltages in the network. Inverting Eq. (5.16), we have

V ¼ ðT T Þ1 VnC ð5:19Þ

0
Substituting Eq. (5.10) into (5.17) gives the relation between VnC and Vn

VnC ¼ ðT T TÞ1 Vo;n ð5:20Þ

From Eqs. (5.17) and (5.18), it follows that the voltage vector VnC uniquely
defines the current state of network. Substituting Eq. (5.17) into (5.14) gives the
relation between Vm;o and VnC

Vm;o ¼ ð1  T 1 Þ ðT T Þ1 VnC ð5:21Þ

The above-derived formulas do not exhaust all possible. Reducing the relations
to the order m and n enables derivation of various practically useful equations.
Chapter 6
Algebraic Model of Current–Voltage
Vectors

6.1 Orthogonality of Current and Voltage Vectors

The algebraic models of currents and voltages have been derived in the previous
chapters. In this chapter, the important physical notion—the combined current–
voltage vector—is defined and described. Analogically as the current and the voltage
models, the current–voltage model is derived in terms of linear algebra. In order to
show the background of current–voltage quantity, let us start from the power law.
The branch power sj in the jth branch of network can be defined as follows:

sj ¼ ij vj or sj ¼ vj ij ð6:1Þ

where ij and vj are the branch current and branch voltage in the jth branch, and ij
and vj are the complex conjugates of ij and vj . The power law of network states that
the sum of all branch powers at any instant is equal to zero. In other words, for any
feasible current and voltage vectors I and V

X
j¼b
I V ¼ sj ¼ 0 ð6:2Þ
j1

where I  is the conjugate transpose of the branch current I.

This means that the current subspace I and the voltage subspace V are orthog-
onal if the same branch ordering is used in both spaces.
The Eq. (6.2) is sometimes taken as axiomatic, but it can be proved that it is a
consequence of orthogonality of the cut-set and loop-set matrices as follows:
Suppose I and V are given in the form of Eqs. (4.8) and (5.8) as follows:

I ¼ LT Im and V ¼ C T Vn

© Springer International Publishing AG 2017 37

A. Kłos, Mathematical Models of Electrical Network Systems,
Lecture Notes in Electrical Engineering 412,
DOI 10.1007/978-3-319-52178-7_6
38 6 Algebraic Model of Current–Voltage Vectors

Substituting these equations to (6.2) we have:

I  V ¼ ðLT Im Þ C T Vn ¼ Im ðLT Þ C T Vn ð6:3Þ

Since L is a real matrix, ðLT Þ ¼ ðLT ÞT ¼ L, then

I  V ¼ Im L C T Vn ð6:4Þ

Taking into account the orthogonality of the cut-set and loop-set matrices
L CT ¼ 0 [see Eq. (3.14)], the above equation is just the power law

I V ¼ 0 ð6:5Þ

So it was proved that the current subspace I and the voltage subspace V are
orthogonal to each other.

6.2 Current–Voltage Vector

In this section, summarizing the results of current and voltage algebraic models
given in Chaps. 4 and 5, the current–voltage vector is defined.
Taking into account that the current state vectors Im;o and voltage state vectors
Vo;n [see Eqs. (4.9) and (5.9)] are vectors in the topological space T and are
orthogonal to each other, they can be added together in the current–voltage state
vector.
The current–voltage state vector is defined as a sum of vectors Im;o and Vo;n ,
denoted as Ki;v
     
Im 0 I
Ki;v ¼ Im;o þ Von ¼ þ ¼ m ð6:6Þ
0 Vn Vn

where Ki;v is a vector of space T of order b.

The current–voltage space K i;v is the set of all current–voltage vectors Ki;v .
Vector Ki;v may be unexpected from the physical point of view but is natural
from the mathematical viewpoint. It is not widely known but it is linked to a
familiar vector, which includes the current and voltage state vectors and which is
used in classical network analysis.
In order to justify the physical meaning of the summation of currents and
voltages, let us use the topological transformations of current vectors and of voltage
vectors described in Chaps. 4 and 5. Using topological transformation matrix T, the
current and voltage state vectors Im;o and Vo;n are transformed from topological
space T into space A [see Eqs. (4.14) and (5.14)] and are the current and voltage
vectors I and V. Substituting Eqs. (4.14) and (5.14) into Eq. (6.6) we have:
6.2 Current–Voltage Vector 39

Ki;v ¼ T 1 I þ T 1 V ¼ T 1 ðI þ VÞ ð6:7Þ

where ð I þ VÞ is a new, non-conventional, and interesting vector (as it will be

shown in what follows). Denoting this vector K:

K ¼ I þ V; ð6:8Þ

the relation between the current state vector Ki;v and network current vector I is as
follows:

Ki;v ¼ T 1 K ð6:9Þ

where
The current–voltage vector K is defined as the sum of the current vector I and
the voltage vector V in the network.
Relation (6.9) means that in terms of linear algebra, Ki;v is the topological
transformations of current–voltage vector K.
Every element of vector K is the sum of branch current ij and branch voltage vj

kj ¼ ij þ vj : ð6:10Þ

The current–voltage subspace K of space A is the set of all current–voltage

vectors K. Subspace K includes the spaces I and V, and is the directed sum of
subspaces I and V. The vectors I and V may be recovered as the projections of K
into I and V.
The current–voltage vector K as a sum of vectors I and V is unexpected from the
physical point of view but is natural from the mathematical viewpoint.

6.3 Linear Space Model of Current–Voltage Vectors

The current–voltage state space of a simple example of 3-branch network is

illustrated in Fig. 6.1. In Fig. 6.1a, the spaces I, V, and K are subspaces of space A.
Vectors I, V and K are numerically defined in orthonormal basis (a1, a2, a3) of
space A.
In Fig. 6.1b, the spaces I m;o , V o;n , and K i;v are subspaces of space T. Vectors
Im;o , Von , and K are numerically defined in cut-set and loop-set basis (C1, C3, L2) of
space T.
Let us illustrate the numerical values of current–voltage vectors using the branch
currents and branch voltages of simple examples of 3-branch network in Chaps. 4
and 5.
40 6 Algebraic Model of Current–Voltage Vectors

a2

L L
I I ,o m

I L2 Im,o
C3
C3 Ki,v Ki,v
1
Vo,n
V
Vo,n
V a1 C1
C1

a3

a) Linear space A b) Linear space T

Fig. 6.1 The current–voltage vectors in linear spaces A and T. a Linear space A. b Linear space T

2 3 2 3 2 3 2 3 2 3 2 3
1 1 1 0 1 1 1 1=3 1=3 1=3
6 7 6 7 6 7 6 7 6 7 6 7
I ¼ 4 1 5; Im;o ¼ 4 0 5; V ¼ 4 1 5; Vo;n ¼ 4 1 5; T ¼ 4 1 1 0 5; T ¼ 4 1=3 2=3 1=3 5;
1 0 2 2 1 0 1 1=3 1=3 2=3
2 3 2 3 2 3 2 3 2 3 2 3
1 0 1 1 1 2
6 7 6 7 6 7 6 7 6 7 6 7
Ki;v ¼ Im;o þ Vo;n ¼ 4 0 5 þ 4 1 5 ¼ 4 1 5; K ¼ I þ V ¼ 4 1 5 þ 4 1 5 ¼ 4 2 5
0 2 2 1 2 1
2 3 2 3 2 3 2 3 2 3 2 3
1 1 1 1 2 1=3 1=3 1=3 2 1
6 7 6 7 6 7 6 7 6 7 6 7
K ¼ T Ki;v ¼ 4 1 1 0 5 4 1 5 ¼ 4 2 5; Ki:v ¼ T 1 K 4 1=3 2=3 1=3 5 4 2 5 ¼ 4 1 5
1 0 1 2 1 1=3 1=3 2=3 1 2

6.4 Current–Voltage Vectors’ Relations Using Matrix T

The equations for the recovery of current and voltage state vectors from either K or
Kiv are using the order m x b and n x b matrices ½ 1m 0  and ½ 0 1n  to pick out
the first m and the last n components of matrix Kiv

I m ¼ ½ 1m 0 Ki v Vn ¼ ½0 1n Ki v ð6:11Þ

Using this and equations ( ), ( ) and assuming all matrices have matching row
and column orderings
6.4 Current–Voltage Vectors’ Relations Using Matrix T 41

I ¼ LT ½1m 0Ki;v ¼ LT ½1m 0T 1 K ð6:12Þ

V ¼ C T ½0 1n Ki;v ¼ C T ½0 1n T 1 K ð6:13Þ

Let us define a new vector opposite to the current–voltage vector denoted as Kv i ,

which is the sum of cotree voltages and tree currents vectors.
The voltage–current vector Kv i is the sum of cotree voltages and tree currents
vectors:
     
Vm 0 Vm
Kv i ¼ þ ¼ ð6:14Þ
0 In In

Using Eqs. (4.7) and (5.7), the relation between Kv i and Ki;v is as follows:

Kvi ¼ ðT  1ÞKiv ð6:15Þ

Analogically using Eqs. (4.16) and (5.16), the relation between Kv i and K is as
follows:

Kvi ¼ ð1  T 1 ÞK ð6:16Þ

Using matrix T T and summing the vectors ImL and VnC [see Eqs. (4.17) and
(5.17)], a new current–voltage vector K L;C is as follows:

K L;C ¼ T T K ¼ ImL þ VnC ¼ ImL þ VnC ð6:17Þ

where each of n entries of K L;C is the algebraic sum of the branch currents along the
loop-set, and each of m entries of K L;C is the algebraic sum of the branch voltages
belonging to the cut-set. Inverting Eq. (6.17), we have:

K ¼ ðT T Þ1 K L;C ð6:18Þ

Note that vector K L;C is the topological transformation of vector K.

6.5 Comment

Up to this chapter, the modeling process has produced formulae using more general
approach to the mathematical modeling of electrical networks. In this chapter, as
happens frequently, the model has produced unexpected results concerning the
current–voltage vector K, which do not arise naturally from physics. The current–
voltage state vector K appears as natural in the linear space model formulated here;
yet physically, this vector is new and unexpected in network theory. However, it
42 6 Algebraic Model of Current–Voltage Vectors

means that mathematical summation of physical quantities may have physical

meaning. The addition of currents and voltages is not an obvious process to carry
out, especially as the units of measurements do not matter as long as they are
consistent for all currents and for all voltages. For example, currents might be in
amperes and voltages in volts. The vector Ki;v , on the other hand, is often used in
physics and should present no difficulties to circuit theorists because it uniquely
defines the current and voltage state of network. However, the vector K does the
same thing and so has an equal right to be called a state vector. In terms of linear
algebra, Ki;v and K are just two different column matrices representing the same
current–voltage vector with respect to two different bases as was remarked previ-
ously in Chaps. 4 and 5.
Chapter 7
Kirchhoff’s Laws Using Matrix T

One of the main results of the algebraic models of current and voltage states,
derived in previous chapters, is the introduction of two b-dimensional linear spaces
(b—number of network branches). It makes possible derivation of various current,
voltage, current–voltage relations and formulae in two different and isomorphic
linear spaces: space A and space T. This was done in Chaps. 4, 5, and 6 using
classical formulation of the Kirchhoff’s laws and topological matrix T. However, it
appears that modeling process is leading to the non-conventional formulations of
Kirchhoff’s laws. In this chapter, three of such formulations are presented.

7.1 Basic b-Dimensional Formulation

In Chaps. 4 and 5, the current and voltage vectors of network are expressed in two
linear spaces: the current and voltage vectors I and V in the space A, and the current
and voltage state vectors Im;o and Vo;n in the topological basis of space T. The
relations between these expressions are as follows (see Eqs. 4.14 and 5.14):
Im;o ¼ T 1 I ð7:1Þ

Vo;n ¼ T 1 V ð7:2Þ

where



Im Im
Im;o ¼ and I ¼
0 In



0 Vm
Vo;n ¼ and V¼ :
Vn Vn

T is the topological transformation matrix.

© Springer International Publishing AG 2017 43

A. Kłos, Mathematical Models of Electrical Network Systems,
Lecture Notes in Electrical Engineering 412,
DOI 10.1007/978-3-319-52178-7_7
44 7 Kirchhoff’s Laws Using Matrix T

Note that vector Im is on both sides of Eq. (7.1) and vector Vn is on both sides of
Eq. (7.2). Let us transpose in Eq. (7.1) a vector Im from left to right hand side, what
means adding number −1 to the m upper diagonal elements of matrix T 1 , and
analogically; transpose in Eq. (7.2) a vector Vn from left- to right-hand side, what
means adding –1 to the n lower diagonal elements of matrix T 1 . As a result of such
transpositions, matrix T 1 in Eqs. (7.1 and 7.2) changes as follows:
In current equation (7.1), the new matrix denoted ðT 1 ÞI
" #
1 T11  1m T21
ðT ÞI ¼ ð7:3Þ
T31 T41

In voltage equation (7.2), the new matrix denoted ðT 1 ÞV

" #
T11 T21
ðT 1 ÞV ¼ ð7:4Þ
T31 T41  1n

where Tj1 are submatrices of matrix T 1 and 1m , 1n are unit matrices of order
m and n.
The current and voltage laws (7.1 and 7.2) take the homogenous form as
follows:

ðT 1 ÞI I ¼ 0 ð7:5Þ

ðT 1 ÞV V ¼ 0 ð7:6Þ

Let us now change matrices ðT 1 ÞI and ðT 1 ÞV as follows: In both matrices,

each entry of each row is divided by a diagonal element of this row. Such division
does not change the homogenous equations (the spaces of solutions remain
unchanged) because all diagonal entries of matrix T 1 are not equal to zero. As a
result, the new matrices are denoted as follows: AI called current law matrix and AV
called voltage law matrix. The current and voltage laws (7.1 and 7.2) are as follows:

AI I ¼ 0 ð7:7Þ

AV V ¼ 0 ð7:8Þ

Equations (7.7 and 7.8) are a general formulation of the current and voltage laws
and can be called the b-dimensional formulation of Kirchhoff’s laws. They are
analogy to the classical formulations C I ¼ 0 but AI and AV are not incidence
matrices. The matrices AI and AV are of order b  b and of rank m and n accord-
ingly. The values of elements of AI and AV depend on the structure of network
graph and the diagonal elements are equal to 1. Generally for a given network, the
matrices AI and AV are not unique and depend on the topology of network.
7.1 Basic b-Dimensional Formulation 45

However, independent of tree–cotree topology, certain elements in all matrices AI

and AV (the same elements in both matrices) are equal to zero.

7.2 Pseudo-Unit Formulation

Another formulation of Kirchhoff’s laws can be derived starting from the

b-dimensional formulation (Eqs. 7.7 and 7.8). Let us change matrix AI as follows:
All diagonal elements of matrix AI , which are equal to 1, are transferred from the
left to the right side of Eq. (7.7). So the diagonal elements of AI are equal to 0. Such
matrix, with minus sign, is denoted BI .

BI ¼ AI þ 1b ð7:9Þ

Doing the same operation with matrix AV , we have the following:

BV ¼ AV þ 1b ð7:10Þ

After this operations, matrices AI and AV are as follows:

AI ¼ BI þ 1b ð7:11Þ

AV ¼ BV þ 1b ð7:12Þ

Substituting Eqs. (7.11 and 7.12) into (7.7 and 7.8), we have final equations as
follows:

BI I ¼ I ð7:13Þ

BV V ¼ V ð7:14Þ

Matrices BI and BV are a sort of pseudo-unit matrices. So the Eqs. (7.13 and
7.14) can be called the pseudo-unit formulation of Kirchhoff’s laws.
Note that the diagonal elements of the matrices BI and BV are equal to zero. So
each row in Eq. (7.13) is a function relating one branch current to all remaining
branch currents and each row in Eq. (7.14) is a function relating one branch voltage
with all remaining branch voltages. Note that, while in the classical network
analysis, such relations are done using simple (0, 1, −1) incidence matrices C and L,
and then in pseudo-unit formulation, one current and one voltage are the sums of all
other currents and the sums of other voltages, with coefficients equal to the elements
of matrices BI and BV . Generally, the matrices BI and BV are not unique. The
investigation done show that its uniqueness depends on the structure of network
graph. However, for some networks, e.g., for full graph networks, there is only one
matrix BI and one BV . For majority of other network graphs, there is a number (it
46 7 Kirchhoff’s Laws Using Matrix T

depends on network graph) of different matrices BI and BV , but in all of them, some
rows and some entries are identical.

7.3 Current–Voltage Formulation

As it is shown in Chap. 6, the current–voltage state vector Ki;v ¼ Im;o þ Vo;n is the
topological transformations of current–voltage vector K ¼ I þ V.

Km;n ¼ T 1 K ð7:15Þ

It means that Eq. (7.15) contains both current and voltage Kirchhoff’s laws.
Practically, it means that knowing the sum of current and voltage values in each
network branch, and knowing in what units the summation is done, one can find,
using matrix T, all branch currents and branch voltages in the network. The
Eq. (7.15) can be called the combined current–voltage formulation of Kirchhoff’s
law.
Note that the summation of current and voltage may have applications in
measuring and information technology, e.g., measuring together (instead of sepa-
rate) a sum of branch current and voltage may change the measurement systems in
power system networks.
Chapter 8
Current–Voltage Functional Relations

8.1 Branch Parameters

Up to this point, the branch currents and branch voltages of a network system have
been assumed to be unrelated, except that the current space I and the voltage space
V are orthogonal and that individual current and voltage vectors obey the power law
I
V ¼ 0 (see Chap. 6). This means that current vector I and voltage vector V are
independent on each other, as long current I remains in current subspace I and
voltage V in voltage subspace V. In physical networks, this is not the case and the
branch currents and branch voltages are related each other. These relations, gen-
erally called the branch current–voltage functional relations (in what follows called
current–voltage functions) are considered in this chapter. In physical networks, they
take on many different forms of which Ohm’s law is a special type, which often
occurs in electrical networks. The most general form of the current–voltage func-
tional relations, in the b-branch network, would be represented by a set of math-
ematical functions of 2b variables

f ðI; VÞ ¼ f ði1 ; i2 ; . . .ib ; v1 ; v2 ; . . .vb Þ ð8:1Þ

where function f ðI; VÞ may be very different.

In electrical network systems, the f ðI; VÞ functions depend on the physical
nature of branches.
In this work, the considered electrical network systems are restricted to being
time constant and linear or nonlinear of the power-type relation ij vj ¼ pj . In this
chapter, the current–voltage functions of network system branches are described
using Ohm’s law. In all the network systems, the network branches initiate (gen-
erate) the current state and voltage state of network and determine the relations
between current values and voltage values. It means that the network system must
include branches which are: current or/and voltage sources and current–voltage
functions Mathematically, to each branch the linear function is associated, called

© Springer International Publishing AG 2017 47

A. Kłos, Mathematical Models of Electrical Network Systems,
Lecture Notes in Electrical Engineering 412,
DOI 10.1007/978-3-319-52178-7_8
48 8 Current–Voltage Functional Relations

the branch equation or branch parameter. There are three kinds of branch equa-
tions (branch parameters):
– First kind is the branch equations, which are branch current–voltage functions.
They transform branch current into branch voltage or vice versa, using constant
coefficients and are called branch admittance, denoted yj or branch impedance,
denoted zj .
– Second kind is the branch equations, which generate the currents or voltages in a
network. They are the constant current or constant voltage sources and are called
the ideal current sources and the ideal voltage sources.
– Third kind is the branch equations, which are both branch current–voltage
functions and current source and/or voltage source.
In the classical network analysis, if the inductive and capacitive couplings
between branches (self- and mutual couplings) are to be taken into account, then the
branch equation j of third kind is written, traditionally either in admittance or
impedance notation, as follows:

X
k¼b
ij ¼ ðyj;k vk Þ þ e ð8:2aÞ
k¼1

or

X
k¼b
vj ¼ ðzj;k ik Þ þ e ð8:2bÞ
k¼1

where
ij, vj current of branch j and voltage of branch j.
yj;k ; zj;k self- and mutual branch admittances and branch impedances, constant
branch parameters.
e current source or/and voltage source, constant branch parameters.
Each of the above two notations of branch equations can be used as the full
mathematical model of the branch equations of b-branch network; however, not
both notations can be used together for a whole or a part of network, because yj;k
and zj;k are the different notations only of the very same parameters of a branch.
However, both notations can be used if all branches are partitioned into two parts.
The model of branch equations in this chapter refers to the algebraic model of
network system presented in this book. If the network is topologically partitioned
into m cotree branches and n tree branches, then, in what follows, we assume that
the branch equations of a cotree branches are admittances (Eq. 8.2a) and the branch
equations of a tree branches are impedances (Eq. 8.2b), (m + n = b). Note, how-
ever, that in such notation, we may have the current sources and the voltage sources
in both equations. From algebraic viewpoint, the sources should be dislocated. The
branch equations of cotree branches are currents, so they should include the current
sources only and the branch equations of tree branches are voltages, so they should
8.1 Branch Parameters 49

include the voltage sources only. It means that the branches with current sources in
tree branches should be dislocated to the cotree and changed from impedance type
into admittance type, and branches with voltage sources in cotree branches should
be dislocated to the tree branches and changed from admittance type to impedance
type. After such dislocations, the branch equations are as follows:

X
k ¼m
ij ¼ ðyj;k vk Þ þ ei ð8:3Þ
k¼1

and
X
k¼n
vj ¼ ðzj;k ik Þ þ ev ð8:4Þ
k¼1

where
ei current sources in cotree branch equation.
ev voltage sources in tree branch equations.
Note that the numbers of current sources and voltage sources are restricted and
imposed by current and voltage laws. The consistency condition of unique solution
of a network problem is fulfilled if there are no more than m current sources and no
more than n voltage sources in the network.
From the algebraic point of view, it may be of advantage to separate each current
source and each voltage source in branch equations (8.3 and 8.4). Physically, this is
equivalent to replace the branch equation with current source by its equivalent of two
branch equations: first—active part of source in Eq. (8.3), now called the ideal cur-
rent source; second—passive part of source in Eq. (8.3). Analogically replacing the
branch equation with voltage source by its equivalent of two branch equations is the
voltage source separation: first—active part of source in Eq. (8.4), now called the
ideal voltage source; second—passive part of source in Eq. (8.4). It means additional
branches in the network system. After separation, the branch equations are as follows:
ij ¼ eij ð j ¼ 1; 2; 3; . . . f 0 Þ ð8:5Þ

X
m
ij ¼ ðyjk vk Þ ð8:6Þ
j¼1

and
vj ¼ evj ð j ¼ 1; 2; 3; . . . f 00 Þ ð8:7Þ

X
n
vj ¼ ðzj;k ik Þ ð8:8Þ
j¼1
50 8 Current–Voltage Functional Relations

where
f 0 is a number of ideal current sources.
f 00 is a number of ideal voltage sources.
Branches, which are ideal sources, are a point-like branches, which do not have
self- and mutual admittances or impedances. Generally in the network, the current
branches (8.5 and 8.6) may be connected in series and the voltage branches (8.7 and
8.8) are often connected in parallel. Equations (8.5–8.8) may serve to the formulation
of some special branch equations in the practical network solution problems. For
example, the power system generator can be modeled as follows: ideal current source
and ideal voltage source are connected in parallel and current–voltage functions are
connected in series. The branch equation of some branches is often a nonlinear,
complicated function; however, after linearization, the branch equation can be
modeled using three types of equations: current source, voltage source, and current–
voltage equations. In what follows, only the linear (Ohm’s law) branch equations are
used.

8.2 Ohm’s Law of Network System—Algebraic Model

Each real network system must include at least one current or voltage source.
Without any source, the network is empty (I = 0, V = 0) and the classical network
Ohm’s law in such network does not exist. The classical Ohms law can be used
only to the parts of the network system.
In this section, the algebraic model of the Ohm’s law in network system is
derived, which takes into account the current and/or voltage sources.
Consider the network system which is a set of b interconnected branches.
Suppose that any tree/cotree topological structure of network is chosen. Suppose
that the branch equations are of both types (passive and active) and the Ohm’s law
is in the form of Eqs. (8.5–8.8). The network branches (branch equations) are
ordered according to the order of m cotree and n tree branches. The ideal current
sources are branch equations of cotree branches, so they are parts of current state
vector, and the ideal voltage sources are branch equations of tree branches, so they
are parts of voltage state vector Vn . The current state vector Im which takes into
account ideal current sources, now called as the network-current state vector and
denoted as Im;F , is as follows:
 
Im;I
Im;F ¼ ð8:9Þ
Im;y

where
Im;I ¼ EI is subvector of the network-current state vector of order f 0 ′ (f 0 number of
sources); its elements are ideal current sources.
8.2 Ohm’s Law of Network System—Algebraic Model 51

Im;y is the subvector of network-current state vector, which elements are

cotree branch admittances, of order m  f 0 .
Analogically the voltage state vector Vn which takes into account ideal voltage
sources, now called as the network-voltage state vector and denoted as Vn;F , is as
follows:
 
Vn;V
Vn;F ¼ ð8:10Þ
Vn;z

where
Vn;V ¼ EV is subvector of the network-voltage state vector of order f″ (f″ number
of sources), its elements are ideal voltage sources.
Vn;z is the subvector of network-voltage state vector, which elements are
tree branch impedances, of order n  f 00 .
The Ohm’s law expressed in the form of Eqs. (8.5 and 8.8) can be written in the
matrix form as follows:
   
Im;I EI
Im;F ¼ ¼ ð8:11Þ
Im;y Ym;f 0 V

where
Ym;f 0 matrix of cotree branch admittances of order m−f′  b.
   
Vn;V EV
Vn;F ¼ ¼ ð8:12Þ
Vn;z Zn;f 00 I

where
Zn;f 00 matrix of tree branch impedances of order n−f″  b
The Kirchhoff’s current and voltage laws can be expressed in the form of
Eqs. (4.8 and 5.8), which are derived in Chaps. 4 and 5.

I ¼ LT Im ð8:13Þ

V ¼ C T Vn ð8:14Þ

Substituting Eq. (8.14) to (8.11) and Eq. (8.13) to (8.12), we have

   
I EI
Im;F ¼ m;I ¼ ð8:15Þ
Im;y Ym;f 0 CT Vn
52 8 Current–Voltage Functional Relations

   
Vn;V EV
Vn;F ¼ ¼ ð8:16Þ
Vn;z Zn:f 00 LT Im

where
Im;F The current state vector of network system, of order m.
Vn;F The voltage state vector of network system, of order n.
Equations (8.15 and 8.16) are the algebraic model of the Ohm’s law of network
system.
Note that both equations satisfy the Kirchhoff’s and the Ohm’s laws and are the
mathematical relations between the current and voltage vectors: Im;I ; Im;y ; Vn;V ; Vn;z
and branch parameters: EI ; EV ; Zn;f 00 ; Ym;f 0 in network system. If all branch equations
are the current–voltage functions only (f′ = 0 and f″ = 0), then Eqs. (8.15 and 8.16)
reduce to equations:

Im ¼ Ym C T Vn ð8:17Þ

Vn ¼ Zn LT Im ð8:18Þ

which are the classical Ohm’s law in the form of relations: current state vector as a
function of voltage state vector or vice versa (matrices Ym ; Zn are restricted: Ym is of
order m  b and Zn is of order n  b).
Note that in Eqs. (8.15 and 8.16), the network-current state vector Im;F is a
function of voltage state vector Vn and network-voltage state vector Vn;F is a
function of current state Im . It means that if the network topology and branch
admittances Ym are known, then choosing any voltage state vector Vn , the
network-current state vector Im;F can be found. Analogically, if network topology
and branch impedances Zn are known, then choosing any current state vector Im , the
network-voltage state vector Vn;F can be found.

8.3 Linear Space Model of Current–Voltage State

Vectors in Network System

In Chap. 6, the algebraic model of the current–voltage state vector Ki;v and the
linear state space Ki;v of all vectors Ki;v are defined. So far vector Ki;v and space Ki;v
are derived using Kirchhoff’s laws only and are illustrated in the form of a linear
space on a simple example of network in Fig. 6.1 (see Sect. 6.3). In this section, the
current–voltage vector is defined, which satisfy Kirchhoff’s and Ohm’s laws. Let us
start from definition of vector Ki;v given in Chap. 6.
8.3 Linear Space Model of Current–Voltage State Vectors … 53

     
Im 0 I
KI;V ¼ Im;o þ Vo;n ¼ þ ¼ m ð8:19Þ
0 Vn Vn

Substituting in Eq. (8.19) vector Im by its image Im;F from Eq. (8.15) and vector
V n by its image Vn;F from Eq. (8.16), we have
2 3 2 3 2 3
  Im;I 0 EI
Im;F 6 Im;y 7 6 0 7 6 Ym;F C T Vn 7
KI;V;F ¼ ¼6
4 0 5þ
7 6 7 6
4 Vn;V 5 ¼ 4
7
5 ð8:20Þ
Vn;F EV
T
0 Vn;z Zn;F L Im

where
KI;V;F is the current–voltage state vector of network system.
All vectors KI;V;F form the current–voltage space of network system KI;V;F . All
vectors Im;F and all vectors Vo;F form, accordingly the current space of network
system Im;F and the voltage space of network system Vn;F . Note that Eq. (8.20)
represents both the Kirchhoff’s and the Ohm’s network system laws.
In what follows, the current–voltage space of network system KI;V;F is illustrated
on a simple example of 3-branch network. It is shown that ideal current and voltage
sources are reducing the dimensions of spaces, Im;F ; Vo;F and space KI;V;F .
1. If in the network the ideal current sources do exist and there are no ideal voltage
sources, then in Eq. (8.15) matrix Ym;F is restricted (is of order m − f′  b) and
network-current state vector Im;F includes f′ constant ideal current sources (in-
cludes subvector Im;I ). It means that the network-current–voltage state vectors
KI;v;F include f′ constant elements (ideal current sources). So the space KI;V;F is
of dimension b − f′. It is illustrated in Fig. 8.1. Branch 2 (one cotree branch) is
ideal current source, so vector Im;I is equal constant value = EI2 , and space Im;F
is of dimension zero. Branches  1 and 3 (two branches of a tree) are impedance
Vn;F1
branches, so Vn;F ¼ and space Vo;F is of dimension 2 [plane (a1, a3)]. It
Vn;F3
means that the dimension of network space KI;V;F is equal b − f′ = 2 and is
shown as the two-dimensional lattice plane in Fig. 8.1.
2. Analogically, if the voltage sources do exist and there are no ideal current
sources, then in Eq. (8.15) matrix Zn;F is restricted (is of order m − f″  b) and
network-voltage state vector Vn;F includes f″ constant ideal voltage sources
(includes subvector Vn;V ). It means that the network-current–voltage state vec-
tors KI;v;F include f″ constant elements. So the space KI;V;F is of dimension b − f
′. It is illustrated in Fig. 8.2. Branch number 1 (one of two tree branches) is ideal
voltage source EV1 , so Vn;V1 ¼ EV and space Vn;F is of dimension 1 (line).
Branch 2 (cotree branch) is admittance branch, so Im;F ¼ Im;Y and space Im;F is
of dimension 1 [line (a2)]. It means that the dimension of network space KI;V;F is
equal b − f″ = 2 and is shown as the two-dimensional lattice plane in Fig. 8.2.
54 8 Current–Voltage Functional Relations

a2

Im,F = EI
E I2
2

1 3
K i,v,F Vn,F1 Vn, F3
K i,v,F

Vn,F3 Vn,F1
Vo,n,F Tree Cotree
a3 Vn,F a1

Fig. 8.1 Illustration of space KI;V;F in case if branch 2 is ideal current source

a2
Im,F
2
Im,F2 1 3
Vn,F,1 = EV Vn,F3

K i,v,F
Ki,v,F
Vn,F3 Tree Cotree
EV1
Vo,n,F Vn,F
a3 a1

Fig. 8.2 Illustration of space KI;V;F in case if branch 1 is ideal voltage source

3. If the current and voltage sources exist, then in Eq. (8.15) matrices Ym;F and Zn;F
are restricted. It means that the network-current–voltage state vectors KI;v;F
include f ′ + f ″ constant elements. The space Im;F is of dimension zero and space
Vn;F is of dimension 1 So the space KI;V;F is of dimension b − (f ′ + f ″) = 1. It
can be illustrated if lattice linear spaces in Figs. 8.1 and 8.2 are drawn together
in one linear space (not shown), in which intersection of lattice planes forms a
line. It means that space KI;V;F is one-dimensional.
8.4 Formulation of Ohm’s Law Using Matrix T 55

8.4 Formulation of Ohm’s Law Using Matrix T

The network system model developed in previous sections was derived in the form
of the current and voltage state vectors Im;F and Vn;F . In the practical point of view,
it is of advantage to derive this model in the form of current and voltage vectors
I and V. In this section, such model is derived using topological matrix T.
Consider an electrical, b-branch network system in which current sources are in
cotree and voltage sources are in tree and each branch which includes the source is
separated into two branches: first, branch which has ideal current or voltage source,
and second, branch which has current–voltage function. The Ohm’s law can be
formulated in matrix form as follows:

Fm;n KI;V;F ¼ Km;n ð8:21Þ

where
Fm;n The matrix of branch parameters called the network parameters matrix.
2 0 3
1f
6 Zm;y 7
Fm;n ¼ 6
4 00
7
5 ð8:22Þ
1f
Yn;z

Assuming that there are no mutual couplings in the network, the matrix Fm;n
consists of two types of diagonal submatrices: the unit submatrices indicating ideal
current and voltage sources EI and EV , respectively, and the current–voltage
function matrices of impedance and admittance types. Note that, different as in
previous sections, the passive branch parameters of cotree branches are impedances
and passive branch parameters of tree branches are admittances. Generally, the
ordering of submatrices in matrix Fm;n is optional. Let us assume that the ordering
of submatrices in matrix Fm;n is as follows:
1f
0
Unit matrix indicating the ideal current sources in cotree, of order f ′  f′ (f ′ is
a number of current sources).
Zm;y submatrix of impedance branches in cotree, of order m  f 0 ðm  f 0 is a
number of impedance branches).
1f
00
Unit matrix indicating the voltage sources, of order f 00 ðf 00 is a number of
voltage sources).
Yn;z Submatrix of tree admittance branches of order n  f 00 ðn  f 00 is a number of
admittance branches).
Using vector Ki;v;F in the form (see Eq. 8.20):
56 8 Current–Voltage Functional Relations

2 3
Im;I
6 Im;y 7
Ki;v;F ¼6 7
4 Vn;V 5 ð8:23Þ
Vn;z

where
Im;I ¼ EI vector of ideal current sources in cotree, of order f ′.
Im;y vector of branch impedances in cotree, of order m  f 0 .
Vn;V ¼ EV vector of ideal voltage sources in tree, of order f 00 .
Vn;z vector of branch admittances in tree, of order n  f 00 .
Vector Km;n in Eq. (8.21) is a current–voltage vector; arranged according to the
ordering of matrix Fm;n .
2 3
EI
6 Vm;y 7
Km;n ¼ Fm;n Ki;v;F ¼6
4 EV 5
7 ð8:24Þ
Im;z

where
Vm;y voltage vector of impedance branches in a cotree, of order mz.
In;z current vector of admittance branches in a tree, of order ny .
Eq. (8.24) is the Ohm’s law only. In order to take into account the Kirchhoff’s
laws, let us use topological transformation matrix T (see Chap. 6; Eq. 6.9):

Ki;v;F ¼ T 1 K ð8:25Þ

Substituting Eq. (8.25) into (8.26), we have

Km;n ¼ Fm;n T 1 K ð8:26Þ

where K ¼ I þ V
If the topological matrix T is known and if parameters of passive and active
branches are nonzero values, then matrix Fm;n is non-singular.
Denoting:

R ¼ Fm;n T 1 ð8:27Þ

The Ohm’s law using matrix T is as follows:

R K ¼ Km;n ð8:28Þ
8.5 Matrices of System Admittance and System Impedance 57

8.5 Matrices of System Admittance and System

Impedance

Substituting equation K ¼ I þ V into Eq. (8.28), we have

R I þ R V ¼ Km;n ð8:29Þ

Note that if writing the matrix equation (8.29) in the form of a set of b scalar
equations, then on the left- and right-hand sides of equations there are the elements
containing the same variables (branch currents and branch voltages). Transferring
this variables from the right- to the left-hand side of Eq. (8.29) and adding
accordingly, we have

RI I þ RV V ¼ 0 ð8:30Þ

where
RI matrix R after transferring branch currents.
RV matrix R after transferring branch voltages.
Equation (8.30) can be interpreted in two ways either the current vector I as a
function of voltage vector V or the voltage vector V as a function of current vector I.

I ¼ R1
1 R2 V ð8:31Þ

V ¼ R1
2 R1 I ð8:32Þ

Denoting:

YS ¼ R1
1 R2 ð8:33Þ

ZS ¼ R1
2 R1 V ð8:34Þ

where
YS can be called the matrix of network system admittance.
ZS can be called the matrix of network system impedance.
Finally Eqs. (8.31 and 8.32) are:

I ¼ YS V ð8:35Þ

V ¼ ZS I ð8:36Þ

The above equations are the network system equations relating b-

dimensional vectors: current vector I and voltage vector V of a network system.
Eqs. (8.35 and 8.36) are a non-conventional formulation of the Ohm’s laws of
58 8 Current–Voltage Functional Relations

network system, which include the branches which are current and/or voltage
sources. Note that each real network, called in this book the network system, must
include at least one current or voltage source. Without any source, the network is
empty (I = 0, V = 0) and the classical network Ohm’s law in such network does not
exist. The classical network Ohms law can be used to the parts of network. The
Ohms law of network system, derived above, refers to the whole network system
which includes at least one source.
Note the following special properties of network system matrices YS and ZS
1. The network system matrices YS ; ZS and equations I ¼ YS V; V ¼ ZS I fulfill the
fundamental equation:

YS ZS ¼ R1 1
I RV RV RI ¼ 1 ð8:37Þ

2. The matrices YS ; ZS , and equations I ¼ YS V, V ¼ ZS I are the analogy to the

classical admittance and impedance matrices and classical Ohm’s equations;
however, they cannot be identified with them. Note the important difference
between the classical matrices Y, Z, and the system matrices YS and ZS . The
elements of classical matrices are pure branch parameters only, independent on
network as a whole; the elements of system matrices are network system
parameters, dependent on network as a whole.
3. Matrices YS and ZS are not diagonal (even if matrix Fm,n is diagonal); so they are
analogy to the self- and mutual branch couplings of a network, but they cannot
be identified with them. Elements of matrices YS ; ZS are system self- and system
mutual admittances and impedances, which numerically differ from classical
branch admittances/impedances. They are the whole network system quantities.
4. Elements of matrices YS and ZS are network parameters of branches which
not only takes into account branch admittances/impedances but also takes
into account ideal current and voltage sources, which are not admittances/
impedances. They are system parameters characterizing actual state of network
system.
5. As it comes from Eqs. (8.33 and 8.34), the existence of system admittance YS
and impedance ZS depend on the non-singularity of matrices RI and RV .
Analysis of a simple network example, which are given in Chap. 14 shows that
these matrices may be non-singular or singular depending on the number and
configuration of ideal current and voltage sources in a network (see Chap. 14).
However, it is difficult to formulate the mathematical conditions of existence YS
and ZS because there may be some other reasons of singularity of matrices RI
and RV .
Chapter 9
General Comments to Part I

The book presents a modern and non-conventional approach to the theoretical

analysis and solution of electrical networks systems. The generally unknown
interrelations of network quantities and the formulations of the fundamental net-
work laws are derived. Some possibilities of its application in network analysis and
solution are given. Formulations are based on the original mathematical (linear
algebra) model of a network. It was shown that modelling network graph in the
form of a topological incidence matrix T makes possible derivation of various
formulations of network laws and is widening the network analysis. Various matrix
equations of voltage and current relations and general network solution equations
are derived. The algebraic model of a network topology leads to some unexpected
results. It was proved that summation of current and voltage values is mathemat-
ically sensible. The Ohm’s law of network system, which takes into account ideal
current and voltage sources are derived and illustrated.
Commenting generally, it seems to me that what was presented in this part of
book may lead to the following general remarks (conclusions), namely:
• First: that the network system is “the totality which is not a sum of its parts.”
The network system has its own character and features; has its own mathe-
matical quantities: vectors, matrices, linear spaces; and has, e.g., its own Ohm’s
law, etc.
• Second: It seems to us that the notions of electrical current and electrical voltage
are not entirely different physical quantities. Branch current and branch voltage
are of course the different physical quantities, but the network current state and
the network voltage state may have more in common. The facts that the current
state and voltage state vectors are orthogonal; that the notion of common,
current–voltage vector of network has physical meaning; that there exists the
topological interdependency between currents and voltages; that the vector of
sums of a branch current and a branch voltage have the mathematical meaning;

© Springer International Publishing AG 2017 59

A. Kłos, Mathematical Models of Electrical Network Systems,
Lecture Notes in Electrical Engineering 412,
DOI 10.1007/978-3-319-52178-7_9
60 9 General Comments to Part I

and that the Kirchhoff’s and Ohm’s laws can be united in a one network law all
these are interesting, worth of notice and need the further theoretical investi-
gations, which may lead to the development of various practical applications.
Generally, the theory of network systems needs further investigations, taking
into account nonlinear functions and dynamic states of networks.
An overall aim of the work described in the first part of book was to develop a
mathematical model of network topology and to derive the topological transfor-
mation matrix T which appears to be a useful tool of electrical network analysis and
solution.
The algebraic model of electrical network presented in the first part, independent
of any practical applications, is widening the theoretical background and may serve
to the further development of network knowledge.
Note that in the above presented algebraic network, the primary notion (not to be
mathematically defined) is the network branch only. Different than in graph theory,
the node is defined as a set of branches. Practical application of the above formulas
does not need introducing the notion of slack node.

9.1 Part II Application Examples

In the second part of the book, some applications of algebraic model of network
system are presented. Taking into account that the knowledge of electrical networks
is used in many fields of science and technology, the algebraic model may have
various applications. It is, of course, not possible even mention all of them in this
part of book. However, some applications concerning power system networks can
be presented in what follows. The recently used methodology of the electrical
network analysis and solution is based on rather simple mathematical model and
was developed mainly for needs in the field of power system networks. The aim of
this part of the book was widening the practical analysis and solution of such
networks, by using the algebraic model presented in the first part.
So far it was not necessary to distinguish the quantities used in model, into
known input data and the unknown to be found quantities. The numerical state of
variables is irrelevant, since the object of theory is only to find the mathematical
model, which reflects all the possible relationships between the quantities. In this
part, the subdivision into input data and unknown variables is necessary. The
applications are divided into the following:
• Applications of the non-conventional formulations of Kirchhoff’s laws given in
Chap. 7 to the practical network system analysis.
• Applications of the algebraic model of network system, derived in the first part
of book, to the widening of methodology of network solution methods.
 Part II
Application Examples

In the second part of book, some applications of algebraic model of network system
are presented. Taking into account that the knowledge of electrical networks is used
in many fields of science and technology, the algebraic model may have various
applications. It is, of course, not possible even mentioning all of them in this part of
book. However, some applications concerning power system networks can be
presented in what follows. The recently used methodology of the electrical network
analysis and solution is based on rather simple mathematical model and was
developed mainly for needs in the field of power system networks. An aim of this
part of book is widening the practical analysis and solution of such networks, by
using the algebraic model presented in first part.
So far, it was not necessary to distinguish the quantities used in model, into
known input data and the unknown to be found quantities. The numerical state of
variables is irrelevant, since the object of theory is only to find the mathematical
model, which reflects all the possible relationships between the quantities. In this
part, the subdivision into input data and unknown variables is necessary. The
applications are divided into:

• Applications of the non-conventional formulations of Kirchhoff’s laws given in

Chap. 7 to the practical network system analysis and
• Applications of the algebraic model of network system, derived in the first part of
book, to the widening of methodology of network solution methods.
Chapter 10
Applications of b-Dimensional
Formulation of Kirchhoff’s Laws

10.1 Basic b-Dimensional Formulation

The non-conventional formulations of Kirchhoff’s laws, given in the first part of

book (see Chap. 7), may have various applications; some of them are given in this
chapter.
In this section, the application of the basic b-dimensional formulation of
Kirchhoff’s law is given [see Eqs. (7.7) and (7.8)]. In Chaps. 4 and 5, various
current relations and voltage relations are derived in the form of m- and n-order
equations. The question arises what these equations have to do with b-dimensional
current and voltage equations.
Consider the current equation. It is well known that the tree currents can be found
from known cotree currents using Eq. (4.6) (In ¼ Cm Im ). In the classical network
analysis, the inverse relations of Eq. (4.6) do not exist because matrix Cm is singular.
However, using algebraic model of network, for some kinds of networks, such
relations do exist and can be derived in the matrix form using the b-dimensional
Kirchhoff’s current law in the form of Eq. (7.7) (AI I ¼ 0). In Eq. (7.7), matrix AI
can be subdivided into four submatrices:
  
AI;1 AI;2 Im
=0 ð10:1Þ
AI;3 AI;4 In

where
AI;1 is a non-singular matrix of order m  m,
AI;2 matrix of order m  n,
AI;3 matrix of order n  m,
AI;4 a non-singular matrix of order n  n.

© Springer International Publishing AG 2017 63

A. Kłos, Mathematical Models of Electrical Network Systems,
Lecture Notes in Electrical Engineering 412,
DOI 10.1007/978-3-319-52178-7_10
64 10 Applications of b-Dimensional Formulation of Kirchhoff’s Laws

Selecting the first row of Eq. (10.1), we have

AI;1 Im þ AI;2 In ¼ 0 ð10:2Þ

For any structure of network topology, if the number of cotree branches is less than
a number of tree branches, m\n, then the matrix AI;1 is non-singular and of order
m, and cotree current vector can be found as a function of tree current vector from
Eq. (10.2).

Im ¼ A1
I;1 AI;2 In ð10:3Þ

where matrix A1I;1 AI;2 is not incidence (1, 0, −1) matrix.

It means that the relation Im as a function of In does exist, but for special kind of
networks only. Note that the second row of Eq. (10.3) is the classical relation
In ¼ Cm Im and matrix A1 I;4 AI;3 is equal to the incidence matrix Cm .
Consider the voltage equation. It is well known that the cotree voltages can be
found from known tree voltages using Eq. (5.6) (Vm ¼ Ln Vn ). In the classical
network analysis, the inverse relations of Eq. (5.6) do not exist because matrix Ln is
singular. However, using algebraic model of network, for some kinds of networks,
such relations do exist and can be derived in the matrix form using the b-dimen-
sional Kirchhoff’s voltage law in the form of Eq. (7.8) (AV V ¼ 0). In Eq. (7.9),
matrix AV can be subdivided into four submatrices:
  
AV;1 AV;2 Vm
ð10:4Þ
AV;3 AV;4 Vn

where
AV;1 is a non-singular matrix of order m  m,
AV;2 matrix of order m  n,
AV;3 matrix of order n  m, and
AV;4 a non-singular matrix of order n  n.
Selecting the second row of Eq. (10.4), we have

AV;3 Vm þ AV;4 Vn ¼ 0 ð10:5Þ

For any structure of network topology, if the number of cotree branches is greater
than a number of tree branches, m [ n, then the matrix AV;4 is non-singular and of
order n, and the tree voltage vector can be found as a function of cotree voltage
vector from Eq. (10.5).

Vn ¼ A1
V;4 AV;3 Vm ð10:6Þ
10.1 Basic b-Dimensional Formulation 65

It means that the relation Vn as a function of Vm does exist, but for special kind
of networks only. Note that the first row of Eq. (10.4) is the classical relation:
Vm ¼ Ln Vn and the matrix A1 V;4 AV;3 is equal to the incidence matrix Lm .
The equations derived above, and some others, non-common not known rela-
tions, which can be derived by subdividing the b-dimensional formulation of
Kirchhoff’s law, may be useful in the network analysis.

10.2 Pseudo-Unit Formulation

Consider the pseudo-unit formulation of Kirchhoff’s laws, derived in Chap. 7 [see

Eqs. (7.13) and (7.14)]. Starting from these equations, the pseudo-unit matrices BI
and BV are derived which relate every branch current with the remaining branch
currents and every branch voltage with the remaining branch voltages.

I ¼ BI I ð10:7Þ

V ¼ BV V ð10:8Þ

Note that the matrices BI and BV can be found knowing topological (tree/cotree)
structure of network only. The matrices BI and BV are non-singular and have
interesting properties, which may be useful in network analysis. In this section, an
example of practical application of the pseudo-unit current equation I ¼ BI I is
presented.
It is intuitively known that any changing of a current or a voltage in any branch
of network system will have certain influence on the currents and voltages of the
other network branches. The Eqs. (10.7) and (10.8) enable the quantitative and
qualitative analyses of this influence. Note that changing a steady state of a real
network, namely changing the branch current ij by constant value Dij , generate the
changing of all branch currents, among them changing current of branch ij . In order
to keep the value Dij constant, the branch j must be the regulated current source.
The problem of finding the numerical values of branch currents after any change in
one branch can be solved using the following iteration procedure:
Let us write the pseudo-unit current equation I = BI I as follows:

Ia ¼ B I Ib ð10:9Þ

For any network current vector I ¼ Ia = Ib , Eq. (10.8) must be fulfilled. If in the
current vector Ib the element ib; j (current of branch j) is changed by adding arbitrary
value Dib; j , then Eq. (10.8) is not fulfilled. However, using the simple iteration
process, one can find a new, unique vector Ia0 ¼ Ib0 ¼ I 0 which fulfills the Eq. (10.8).
The iteration process is as follows:
66 10 Applications of b-Dimensional Formulation of Kirchhoff’s Laws

1. Choose any starting point current vectors I ¼ Ia ¼ Ib ; chose the current of

branch j to be changed ij;a = ij;b = ij .
2. Choose the change Dij;b of branch current ij;b and find the new, constant current
of branch j ikj;b ¼ ij;b þ Dij;b .
3. Change the current vector Ib by substituting ij;b ¼ ikj;b .
4. Find new vectors Ib0 = Ia0 using Eq. (10.8).
5. If i0j;b ¼ ikj;b , then go to 7.
6. If i0j;b is not equal to ikj;b , then Ib ¼ Ib0 and go to 3.
7. End of iteration process. It means that the pseudo-unit current Eq. (10.8) is
fulfilled and the starting point vectors I ¼ Ia ¼ Ib are the new current vectors
Ia0 ¼ Ib0 ¼ I 0 .
Note that the reverse iteration process leads to the starting point current vector
I ¼ Ia ¼ Ib .
Comparing the new branch currents with the starting point branch currents
makes possible finding of how changing of branch current ib; j influences on the
other branch currents in network. It is possible to do the accurate analysis of branch
currents interrelations.
In what follows such analysis is illustrated for a simple example of network in
Fig. 10.1.
Consider a network in which the number of cotree branches is greater than the
number of tree branches. An example of the network graph, the topology of

5 i5=-1

4 i4=- 6 i8 =3 8

6 i6 =5 3 i9=4 9
i3=4
7 i7=2 i1=-7 1

i2=3
2
tree Cotree

6 7 8 9 1 2 3 4 5
0.000 0.500 0.125 -0.250 -0.125 0375 0.000 -0.500 0.375 6
0.500 0.000 0.125 0.125 0.250 -0.375 -0.375 -0.500 0.000 7
0.125 0.125 0.000 0.500 -0.500 0.000 -0.375 0.250 0.375 8
-0.250 0.125. 0.500 0.000 -0.500 -0.375 0.000 -0.125 -0.375 9
BI = -0.125 0.250 -0.500 -0.500 0.000 -0.375 -0.375 0.125 0.000 1
0.333 -0.333 0.000 -0.333 -0.333 0.000 0.333 0.000 0.333 2
0.000 -0.333 -0.333 0.000 -0.333 0.333 0.000 -0.333 -0.333 3
-0.500 -0.500 0.250 -0.125 0.125 0.000 -0.375 0.000 0.375 4
0.333 0.000 0.333 -0.333 0.000 0.333 -0.333 0.333 0.000 5

Fig. 10.1 Example of the pseudo-unit matrix BI of a simple network

10.2 Pseudo-Unit Formulation 67

network, the starting point branch currents and matrix BI is shown in Fig. 10.1.
Suppose that branch currents can be regulated.
The analysis of elements of matrix leads to the following conclusions:
– If the non-diagonal elements bj;k are equal to zero (e.g., element bj;k = bj;5 ), then
it means that current of branch j does not depend on the current changes of
branch k and opposite.
– Greater/lower values of non-diagonal elements bj;k mean greater/lower changing
of current of branch j on the current of branch k and opposite.
Using the above matrix BI , the two cases of numerical calculations (using simple
computer program) can be solved:
In the first case, the current of branch 1 is changed by adding number 1,
Dib;1 = 1. The new, constant current of branch 1, Dikb;1 = −6.000.
In the second case, the current of branch 9 is changed by adding number −2,
ib;9 = −2. The new current of branch 9 ikb;9 = 2.000.
The results of case 1 are as follows:

Branch currents i6 i7 i8 i9 i1 i2 i3 i4 i5
Starting vector 5.000 2.000 3.000 4.000 −7.000 3.000 4.000 −6.000 −1.000
I ¼ Ia ¼ Ib
New vector 4.865 2.270 2.407 3.407 −6.000 2.593 3.593 −5.865 −1.000
0 0 0
I ¼ Ia ¼ Ib
0
Difference I  I −0.135 0.270 −0.593 −0.593 1.000 −0.407 −0.407 0.135 0.000
Difference % −2.7 11 −19.7 −14.8 14.3 −13.5 −10.2 2.0 0

The results of case 2 are as follows:

Branch currents i6 i7 i8 i9 i1 i2 i3 i4 i5
Starting vector 5.000 2.000 3.000 4.000 −7.000 3.000 4.000 −6.000 −1.000
I ¼ Ia ¼ Ib
New vector 5.544 1.728 1.812 2.000 −5.819 3.813 4.000 −5.728 −0.187
0 0 0
I ¼ Ia ¼ Ib
Difference 0.544 −0.272 1.188 −2.000 1.181 0.813 4.000 0.272 0.813
Difference % 10.9 13.6 39.6 50.0 16.8 27.0 0 4.5 18.7

In the first case, 14.3% change of current in branch 1 leads to the highest, that is,
equals to 19.7%, reduction of current in branch 8; the reductions of current from 2
to 14.8% in other branches, and not any change of current in branch 5.
In the second case, 50% change of current in branch 9 leads to the highest, that
is, equals 39.6%, reduction of current in branch 8; reductions from 4.5 to 27% in
other branches, and not any change of current in branch 3.
68 10 Applications of b-Dimensional Formulation of Kirchhoff’s Laws

Generally, using the pseudo-unit formulation of Kirchhoff’s laws, the qualitative

analysis and control of interdependences of branch current flow and branch voltage
values in electrical network branches can be conducted. Additionally, if the branch
voltages and branch impedances in the starting point are known, then one can
analyze the influence of any changes of branch currents on the branch voltages (and
powers) can be done.
Chapter 11
Network Solution Method Using
Algebraic Network Model

11.1 Derivation of Method Using Matrix T

Using the algebraic model of network derived in Part I, the general method of linear
network solution is derived in this section. Suppose that the graph of b-branch
network is known, and the topology is unrestricted, except that a tree includes
voltage sources and a cotree includes current sources. The network branches are
numbered from 1 to b (the first m are cotree branches, and the last n are tree
branches). Generally, the ordering of branches in the cotree and tree is optional. Not
loosing generality, the ordering in vectors and matrices of order b = m + n is
assumed as follows:
– First mi branches are cotree current sources, mi
m.
– Next m − mi branches are cotree admittances.
– Next nv branches are tree voltage sources, nv
n.
– Last n − nv branches are tree impedances.
In what follows, all vectors and matrices are of order b.
Starting from the b-dimensional formulation of the Kirchhoff’s laws [see
Chap. 4, Eq. (4.11) and Chap. 5, Eq. (5.10)].

I ¼ TIm;o ð11:1Þ

V ¼ TVo;n ð11:2Þ

where
I and V current and voltage vectors.
Im,o and Vo,n current state vector and voltage state vector [see (4.12), (5.11)],
T topological matrix.
The vectors Im,o and Vo,n include the current and voltage sources.

© Springer International Publishing AG 2017 69

A. Kłos, Mathematical Models of Electrical Network Systems,
Lecture Notes in Electrical Engineering 412,
DOI 10.1007/978-3-319-52178-7_11
70 11 Network Solution Method Using Algebraic Network Model

Im;o ¼ EI þ IY ð11:3Þ

Vo;n ¼ EV þ VZ ð11:4Þ

where
EI current vector of cotree current sources
IY current vector of cotree admittance branches
EV voltage vector of tree voltage sources
Vz voltage vector of tree impedance branches.
Introducing the elementary Ohm’s law:

IY ¼ Ym V ð11:5Þ

VZ ¼ Zn I ð11:6Þ

where
Ym matrix of cotree branch admittances.
Zn matrix of tree branch impedances.
Substituting Eq. (11.1) into (11.6) and Eq. (11.2) into (11.5), we have:

IY ¼ Ym TVo;n ð11:7Þ

VZ ¼ Zn TIm;o ð11:8Þ

Substituting Eq. (11.9) into (11.8) and Eq. (11.10) to (11.7), we have:

Im;o ¼ EI þ Ym TVo;n ð11:9Þ

Vo;n ¼ EV þ Zn TIm;o ð11:10Þ

Substituting Eq. (11.9) to (11.8) and Eq. (11.10) to (11.7), we have:

Im;o ¼ EI þ Ym TðEV þ Zn TIm;o Þ ð11:11Þ

Vo;n ¼ EV þ Zn TðEI þ Ym TVo;n Þ ð11:12Þ

After doing simple algebraic operations, the current and voltage state vectors Im,o
and Vo,n are as follows:

Im;o ¼ ð1  Ym TZn TÞ1 ðEI þ Ym TEV Þ ð11:13Þ

Vo;n ¼ ð1  Zn TYm TÞ1 ðEV þ Zn TEI Þ ð11:14Þ

11.1 Derivation of Method Using Matrix T 71

Substituting Eq. (11.13) to Eq. (4.12) and Eq. (11.14) to Eq. (5.12) we have the
final equations of linear network solution using matrix T.

I ¼ Tð1  Ym TZn TÞ1 ðEI þ Ym TEV Þ ð11:15Þ

V ¼ Tð1  Zn TYm TÞ1 ðEV þ Zn TEI Þ ð11:16Þ

Equations (11.13)–(11.16) relate all network quantities, and they can be used to
the solution of these quantities and to the solvability analysis depending on the
given data.

11.2 Classical Network Solution Methods

Equations (11.15) and (11.16) are generalization of classical nodal and mesh
methods of power system network solution. After assuming some simplifications
(done in classical analysis), these equations can be reduced to the well-known node
method and mesh method equations, as follows.
The node method is based on the following assumptions. The topology in
classical network analysis is usually assumed as follows: All node-to-earth branches
are tree branches, and all node-to-node branches are cotree branches. The voltage
vector of tree branches Vn and the admittance vector Ym of cotree branches are
known constant values. The unknown and to be found are the vector of tree currents
In. It means that in Eq. (11.15), Vn is known voltage source vector EV = Vn and the
vector of current sources EI = 0. The vector of tree impedances Zn = 0 and all
cotree branches are known admittances Ym. Substituting these data to Eq. (11.15)
makes this equation much simpler. If EI = 0 and Zn = 0, then (1 − YmTZnT)−1 is
equal to unit matrix. So the Eq. (11.15) reduces to the following equation:
      
Im 1 CT Ym 0 1 CT 0 0
¼
In C 1 0 Zn C 1 0 EV

Selecting the equation of order n, we have:

In ¼ CYm C T Vn ð11:17Þ

which is classical node method of network solution.

Analogically, the classical mesh method is based on the following assumptions.
The current vector of cotree branches Im and the vector of tree branch impedances
Zn are known constant values. The unknown and to be found are the vector of
cotree voltages Vm. It means that in Eq. (11.16), the current vector of cotree
branches Im is equal to EI and the voltage source vector EV = 0. The cotree
admittances Ym = 0. Substituting these data to Eq. (11.16) makes this equation
72 11 Network Solution Method Using Algebraic Network Model

much simpler. If EV = 0 and Ym = 0, then (1 − Zn T Ym T)−1 is equal to unit matrix.

Equation (11.16) reduces to the following equation:
      
Vm 1 CT 0 0 1 CT EI 0
¼
Vn C 1 0 Zn C 1 0 0

Selecting the equation of order m, we have:

Vm ¼ C T Ym CT Im ð11:18Þ

which is the classical mesh method of network solution.

The general method of network solution and using various kinds of network
topology make possible the derivation of methods and solution problems in cases of
various data given. The examples of such methods are given in the next sections.
Chapter 12
Method of Current and Voltage
Sensitivity Analysis

12.1 Derivation of Method

It is intuitively known that the current or voltage values of a network branch depend
on the current or voltage values of all other branches in the network (practically
may depend on certain part of network only). In particular applications, it depends
on some control variables which can be identified with current and voltage sources.
Generally, any branch current or branch voltage, if it is controllable, can be con-
ventionally treated as a current or voltage source. The method derived in this
chapter provided the information on how the current or voltage state of network
depends on the current and voltage sources.
Let us start from generalized network solution Eqs. (11.15) and (11.16).

I ¼ Tð1  Ym TZn TÞ1 ðEI þ Ym TEV Þ

V ¼ Tð1  Zn TYm TÞ1 ðEV þ Zn TEI Þ

Denoting:

Fm ¼ Tð1  Ym TZn TÞ1 ð12:1Þ

Fn ¼ Tð1  Zn TYm TÞ1 ð12:2Þ

Fm;y ¼ Tð1  Ym TZn TÞ1 Ym T ð12:3Þ

Fn;z ¼ Tð1  Zn TYm TÞ1 Zn T ð12:4Þ

Using the above notation, the relation between (11.15) and (11.16) can be
written as a function of current and voltage sources.

© Springer International Publishing AG 2017 73

A. Kłos, Mathematical Models of Electrical Network Systems,
Lecture Notes in Electrical Engineering 412,
DOI 10.1007/978-3-319-52178-7_12
74 12 Method of Current and Voltage Sensitivity Analysis

I ¼ Fm EI þ Fm;y EV ð12:5Þ

V ¼ Fn EV þ Fn;z EI ð12:6Þ

where
EI vector of ideal current sources, of order b, which has mi nonzero elements,
which are ideal current sources and
EV vector of ideal voltage sources, of order b, which has mv nonzero elements,
which are ideal voltage sources.
After selecting in Eq. (12.5) the mi columns of matrix Fm , which relate to the
current source vector EI , and selecting in Eq. (12.6) the mv columns of matrix Fm ,
which relate to the voltage source vector EI , the Eqs. (12.5) and (12.6) can be
written as follows:



 " 1 2
#

 " 2 #
Im Fm1 Fm2 EI Fm;y Fm;y 0 Fm1 EI Fm;y EV
¼ þ ¼ þ
In Fm3 Fm4 0 3
Fm;y 4
Fm;y EV Fm3 EI 4
Fm;y EV
ð12:7Þ



 " #

 " #
1 2 1
Vm Fn1 Fn2 0 Fn;z Fn;z EI Fn2 EV Fn:z EI
¼ þ ¼ þ 3
Vn Fn3 Fn4 EV 3
Fn;z 4
Fn;z 0 Fn4 EV Fn;z EI
ð12:8Þ

where matrices with upper index 1, 2, 3, and 4 are submatrices selected from
matrices Fm ; Fm;y and from matrices Fn ; Fn;z . These equations are of order
b. Reducing the order of Eqs. (12.7) and (12.8) and equations EI and EV , the
solution equations are as follows:



Fm1 2
Fm;y EI
I¼ ð12:9Þ
Fm3 4
Fm;y EV



Fn2 1
Fn;z EI
V¼ ð12:10Þ
Fn4 3
Fn;z Ev

Finally, the Eqs. (12.9) and (12.10), written in the simplest form, are as follows:

I ¼ HI E ð12:11Þ

V ¼ HV E ð12:12Þ

where
E vector of current and voltage sources of order mi + nv,
12.1 Derivation of Method 75

HI matrix of order b  (mi + nv) constructed from columns of matrices Fm ; Fm;y ,

which belong to the vector E, and
HV matrix of order b  (mi + nv) constructed from columns of matrices Fm;z ; Fn ,
which belong to the vector E.
Each element j; k of matrix HI is the sensitivity coefficient of current of branch j
versus one of the sources (voltage source EV;k or current source EI;k ) of branch k. It
describes how much branch voltage ij depends on the value of current or voltage
source. In other words, the multiplication of sensitivity coefficient j; k of matrix HI
by element k of matrix E (EI;k or EV;k ) is a part of branch current ij . The numerical
values of sensitivity coefficients in jth row of matrix HI designate the numerical
influence of current and voltage sources on the value of branch voltage ij .
Analogically, each element j; k of matrix HV is the sensitivity coefficient of
voltage of branch j versus one of the sources (voltage source EV;k or current source
EI;k ) of branch k. It describes how much branch voltage j depends on the value of
current or voltage source. In other words, the multiplication of sensitivity coefficient
j; k of matrix HV by element k of matrix E (EI;k or EV;k ) is a part of branch voltage
vj . The numerical values of sensitivity coefficients in jth row of matrix HV designate
the numerical influence of current and voltage sources on the value of voltage Vj .
Note that the elements of matrices HI and HV depend on the passive branch
parameters and that the sources are controlled ideal current and ideal voltage
sources.

12.2 Numerical Example

Equations (12.11) and (12.12) are illustrated on a simple example of power system
network in Fig. 12.1. Branch 4 is the current (power) input (power station); it is also
a slack-point branch and voltage source. Branches 1, 2, and 3 are current outputs
(loads) and also current sources. Branches 5, 6, 7, 8, and 9 are load transmission
lines.
Using passive branch parameters, the numerical results are given in Fig. 12.1 in
the form of equations I ¼ HI E and V ¼ HV E. Elements of matrices HI and HV are
the sensitivity coefficients. Analysis shows the well-known and characteristic fea-
tures of power system networks; e.g., that current flow does not depend on voltage
level determined by slack node voltage (branch 4). Moreover, analysis leads to
some practically useful conclusions.
– Current source E1;2 has the greatest influence on transmission current of line 5,
has lowest influence on current of transmission line 7, and has, of course, no any
influence on current sources EI;1 1 and EI;3 .
– Current sources have influence on all branch voltages, except on slack-point
branch voltage. Current source E1;2 has greatest influence on voltage of line 2
and lowest influence on voltage of branch 7.
76 12 Method of Current and Voltage Sensitivity Analysis

6
Branch 4 is voltage source.
EV,4 = V4 = 1
5 7 9
Branches 1, 2, 3
8 are current sources
4 11 EI,1 = I1 = 1.
EI,2 = I2 = 1
3 2 EI,3 = I3 = 1

Tree Cotree

I1 1.000 0 0 0 V1 0.771 -0.657 -0.486 1.000

I2 0 1.000 0 0 V2 -0.657 0.819 -0.562 1.000
I3 0 0 1.000 0 EI,1 V3 -0.486 -0.562 -0.676 1.000 EI,1
I5 -0.486 -0.562 -0.676 0 EI,2 V5 -0.486 -0.562 -0.676 0 EI,2
I9 = -0.343 0.486 0.229 0 EI,3 V9 = -0.114 0.162 0.076 0 EI,3
I4 -1.000 -1.000 -1.000 0 EV,4 V4 0 0 0 1.000 EV,4
I6 -0.514 -0.438 -0.324 0 V6 -0.771 -0.657 -0.486 0
I7 0.143 0.048 -0.141 0 V7 0.286 0.095 -0.190 0
I8 - 0.343 -0.514 0.229 0 V8 - 0.171 -0.257 0.114 0

I = HI E V = HV E

Fig. 12.1 Illustration of sensitivity coefficients of a simple power system network

– Voltage source EV;4 has no influence on branch currents but has greatest
(nominal) influence on voltages of current source branches 1, 2, and 3.
Generally, the above given method can be used if the value and location of
current and voltage sources are to be found, or in problems with needed current–
voltage interdependences.
In the power system operation, the method of sensitivity coefficients can be used
as a tool of power system security analysis and control. In order to avoid serious
disturbances of a whole or part of a power system, the possible full information is
needed concerning location and danger of disturbance. The method may serve to
provide the quick information of how the branch currents depend on the node
currents or transmission line currents (or as a rough estimate of how branch powers
depend on the chosen node or transmission line powers) and serve to prevent the
emergency situations (e.g., blackouts) in cases of overloads and under- or over-
voltages. The sensitivity method may be used in the power system planning
problems.
Chapter 13
Method of Arbitrary Input Data

13.1 Description of Arbitrary Input Data

The existing solution methods of linear networks, which have great number of
branches, are based on certain assumptions concerning the kind, number, and
distribution of given input data, which are strictly defined. They cover majority of
the practical network solution problems. However, there are problems in which the
form of input data can not be strictly defined. The method of arbitrary input data,
presented in this chapter, deals with problems in which the input data are arbitrary,
randomly distributed in the network. The method is based on the algebraic model of
network derived in the first part of book.
Consider the problem in which the graph of network is known and the input data
(values of branch currents, branch voltages, and branch parameters) are arbitrarily
distributed on the network branches. It is assumed that for a branch the two or one
or no any data may be given. Branches in series are treated as one branch. It is
assumed that there are branches with following kinds of data given:
– branch current only—(i-branch),
– branch voltage only—(v-branch),
– branch parameter (impedance/admittance) only—(f-branch),
– branch current and voltage—(i, v-branch),
– branch current and branch parameter (impedance/admittance)—(i, f-branch),
– branch voltage and branch parameter (impedance/admittance)—(v, f-branch),
– branch with not any data given (o-branch), and
– It is assumed that the input data can not be changed (e.g., branch with given
current and impedance (i, f-branch) can not be replaced by a v-branch). The
input data may be randomly distributed in the network.
It is obvious that the unique solution of such problem may not exist (problem
may be inconsistent), so the method of solution must include the solvability
analysis. Preliminary analysis of input data, done from the point of view of

© Springer International Publishing AG 2017 77

A. Kłos, Mathematical Models of Electrical Network Systems,
Lecture Notes in Electrical Engineering 412,
DOI 10.1007/978-3-319-52178-7_13
78 13 Method of Arbitrary Input Data

solvability, leads to the condition concerning the distribution of certain kind of data.
Problem can be solved if it is possible forming the cotree from some kinds of
branch data and forming the tree from some kinds of branch data. In order to use as
much as possible input data, it is necessary that the branches with given current
must belong to any cotree and the branches with given voltage must belong to any
tree (see Chaps. 4 and 5). It may be not possible if tree and cotree are comple-
mentary topological structure. Note that if there are branches with given current and
voltage (i, v-branches), then in case of complementary tree and cotree, they can not
be effectively used (current i or voltage v can not be used). Moreover, if there are
branches with no data given, then they are useless because they can not belong to a
tree and to a cotree. It means that in order to formulate the topology of network, it is
necessary to use the two different topologies: one for currents, called current
topology, represented by current topological matrix TI , and second one for volt-
ages, called voltage topology, represented by voltage topological matrix TV . It may
be necessary solvability condition. The cotree of current topology denoted by
i-cotree must include branches with given current input values (i-branches, i,
v-branches, and i, f-branches). The tree of voltage topology, denoted by v-tree, must
include branches with given voltage input values (v-branches, i, v-branches, and v,
f-branches). If there are more then m branches with given current or more then
n branches with given voltage, the input data are not consistent (may be omitted).

13.2 Derivation of Solution Equations

The solution method of the above-formulated problem can be derived using the
algebraic model of electrical network described in Part I. Consider the connected
and closed network, with branches numbered from 1 to b. Assuming the two
different topologies, i-cotree and v-tree should be chosen, the Kirchhoff’s laws can
be written as follows [see Chaps. 4 and 5, Eqs. (4.12) and (5.12)]:

I ¼ TI Im;o ð13:1Þ

V ¼ TV Vo;n ð13:2Þ

where
I and V the current and voltage vectors of a network,
Ti Current topological matrix (i-cotree and complementary tree),
TV Voltage topological matrix (v-tree and complementary cotree),
Im,o The current state vector of order b; the first m elements are cotree currents,
and last n elements are zeros, and
Vo,n The voltage state vector of order b; the first m elements are zeros, and last
n elements are tree voltages.
13.2 Derivation of Solution Equations 79

Using the input data, the current cotree vector Im;o can be formulated as the sum
of vectors:

Im;o ¼ Ii þ If ð13:3Þ

where
Ii current cotree vector (of order b), which elements include the known branch
currents (input data: i-branches, i, v-branches, and i, f-branches).
If current cotree vector which elements can be found using known i-cotree
admittances taken from f-branches and f, v-branches and using the unknown
voltage state vector Vo,n.

If ¼ Ym TV Vo;n ð13:4Þ

where Ym is the admittance matrix of i-cotree branches.

Analogically, the voltage state vector Im;o can be formulated as the sum of two
vectors:

Vo;n ¼ VV þ Vf ð13:5Þ

where
VV voltage tree vector (of order b) which elements include the known branch
voltages (input data: v-branches, i, v-branches, and v, f-branches),
Vf voltage tree vector which elements can be found using known v-tree
impedances taken from f-branches and f, i-branches and using the unknown
voltage state vector Im,o.

Vf ¼ Zn Ti Im;o ð13:6Þ

where Zn —impedance matrix of v-tree branches.

Substituting Eqs. (13.4) into (13.3) and (13.6) into (13.5), we have

Im;o ¼ Ii þ Ym Tv Vo;n ð13:7Þ

Vo;n ¼ Vv þ Zn TI Imo ð13:8Þ

Note that in the current state vector Im;o and the voltage state vector Vo;n , both are
in each of the above equations, so there are possible two substitutions. Substituting
(13.8) into (13.7) and substituting (13.7) into (13.8), we have

Im;o ¼ Ii þ Ym Tv ðVv þ Zn Ti Im;o Þ ð13:9Þ

Vo;n ¼ Vi þ Zn Ti ðIi þ Ym Tv Vo;n Þ ð13:10Þ

80 13 Method of Arbitrary Input Data

Solving Eq. (13.9) for Im;o and solving Eq. (13.10) for Vo;n , the current state
vector and the voltage state vectors can be found as follows:

1
Im;o ¼ 1b  Ym Tv Zn Ti ðIi þ Ym Tv Vv Þ ð13:11Þ

1
Vo;n ¼ 1b  Zn Ti Ym Tv ðVv þ Zn Ti Ii Þ ð13:12Þ

Denoting:


Fy ¼ 1b  Ym Tv Zn Ti ð13:13Þ


Fz ¼ 1b  Zn Ti Ym Tv ð13:14Þ

The final current state vector and voltage state vector are as follows:

Im;o ¼ Fy1 ðIi þ Ym Tv Vv Þ ð13:15Þ

Vo;n ¼ Fz1 ðVv þ Zn Ti Ii Þ ð13:16Þ

13.3 Solvability Analysis

Let us now consider the solvability conditions. In the above equations, vectors of
the current and voltage states Im;o and Vo;n are to be found. Vectors Ii ; Vv and
matrices Yc Zt are formed from the given input data.
In order to find the current state vector, the number of nonzero entries in the
current cotree vector Im;o must be equal to m. In other words, the current topological
structure, represented by topological matrix Ti , must be chosen in such a way that
the i-cotree include i-branches, i, v-branches, and i, f-branches and if necessary
f-branches and f, v-branches, but can not include o-branches. The branches with
known current and admittance (i, f-branches) must be treated as branches with
known current, and f, v-branches must be treated as branches with known admit-
tance only.
In order to find the voltage state vector, the number of nonzero entries in the
voltage tree vector Vo;n must be equal to n. In other words, the voltage topologi-
cal structure, represented by topological matrix Tv , must be chosen in such a
way that the v-tree include v-branches, i, v-branches, and v, f-branches and if
necessary f, i-branches and f-branches, but can not include o-branches. The bran-
ches with known voltage and impedance (v, f-branches) must be treated as branches
with known voltage only and i, f-branches as known impedance only.
The additional necessary solvability condition is the non-singularity of matrices
Fy and Fz . Note that if the branch j, with given parameter (admittance/impedance),
belongs to the current i-cotree and also to the voltage v-tree (current topology and
13.3 Solvability Analysis 81

voltage topology are non-complementary), what means that the entry zj in matrix Zt
is in the same time the entry yj ¼ 1=zj in matrix Yc , then as a result of multiplication
(taking into account matrices Tv and Tv ) the diagonal entries j of matrices Fz and Fy
are equal to zero. It may mean singularity of matrices. It leads to the following
solvability condition: If branch parameter is used as impedance in matrix Fz , then it
can not be used as admittance in matrix Fy and opposite. If such condition is
fulfilled, then the diagonal entries of matrices Fz and Fy are equal to 1 and

Fy1 Ii ¼ Ii ð13:17Þ

Fz1 Vv ¼ Vv ð13:18Þ

The final current state and voltage state equations are as follows

Im;o ¼ Ii þ Fy1 Ym Tv Vv ð13:19Þ

Vo;n ¼ Vv þ Fz1 ZTi Ii ð13:20Þ

and the final current and voltage equations of a network are as follows
 
I ¼ Ti Ii þ Fy1 Ym Tv Vv ð13:21Þ


V ¼ Tv Vv þ Fz1 Zn Ti Ii ð13:22Þ

In practical applications, the independent solution of currents and voltages is not

necessary. The vector I can be found from Eq. (13.21) and vector V from simple
equation avoiding calculation of matrix Fz1 as follows

V ¼ Tv ðVv þ Zn IÞ ð13:23Þ

or opposite; if vector V is found from Eq. (13.22), then vector I (avoiding calcu-
lation of matrix Fy1 ) can be found from simple formula

I ¼ Ti ðIi þ Ym VÞ ð13:24Þ

13.4 Numerical Example

The input data of the arbitrary method of network solution are illustrated on a
simple example of network shown in Fig. 13.1. There is given network graph and
following branch data: one i, v-branch, one i-branch, one v-branch, one f, i-branch,
and one v, f-branch; for three branches, there are not any input data.
82 13 Method of Arbitrary Input Data

6
Branch Data given
1 v1 = 10 i1 = 3
5 7 2 i2 = 3
9 3 data not given
8 4 datanotgiven
5 v5 = 1
6 z6 = 2
4 3 2 1 7 v7 = 12 y7 = 1/3
8 i8 = 2 z8 = 1
9 data not given

Tree Cotree

Fig. 13.1 Input data of arbitrary method of network solution

As it follows from the solvability conditions, there are two topological structures
necessary. The tree of voltage topology must include branches 1 and 7 because of
given voltages v1 and v7. The cotree of current topology must include branches 1, 2,
and 8 because of given currents i1, i2, and i8. The branches with not given data can
not belong to voltage tree and current cotree. The voltage topology and the current
topology are shown in Fig. 13.2.
The numerical calculations must be done using only one topology of network.
Any one is possible but one of two in Fig. 13.2 is recommended. Assuming current
tree/cotree topology, the ordering of branches in vectors and matrices is as follows:
First, m = 5 branches are cotree branches numbered 1, 2, 5, 7, and 8, and the last
n = 4 branches are tree branches numbered 3, 4, 6, and 9. The matrices
TI ; TV ; Ym ; Zn (see Eqs. 13.13 and 13.14) are given below. Note that the ordering
of branches in matrices must be the same

6 6
z6
5 7 5 7
v7 y5 y7 9
9
8 8
v1 z8 i1 i8 i2
1 4 3 2 1 4 3 2

Voltage tree Current cotree

Fig. 13.2 Chosen voltage and current topologies

13.4 Numerical Example 83

1 2 5 7 8 3 4 6 9
2 3
1 0 0 0 0 1 0 1 1 1
6 0 1 7
6 1 0 0 0 1 0 0 72
6 7
6 0 0 1 0 0 1 1 1 1 7
6 75
6 7
6 0 0 0 1 0 1 1 0 1 77
6 7
TI ¼ 6
6 0 0 0 0 1 1 1 0 0 7
78
6 7
6 1 1 1 1 1 1 0 0 0 73
6 7
6 0 0 1 1 1 0 1 0 0 7
6 74
6 7
4 1 0 1 0 0 0 0 1 0 56
1 1 1 1 1 0 0 0 1 9

1 2 5 7 8 3 4 6 9
2 3
1 0 1 0 0 0 0 1 0 1
6 0 1 7
0 7 2
6 1 0 0 0 1 0
6 7
6 1 0 1 0 0 0 1 0 0 7
6 75
6 7
6 0 1 0 1 0 0 0 1 0 7 7
6 7
TV ¼  6
6 0 0 0 0 1 1 0 0 1 778
6 7
6 0 0 0 0 1 1 1 0 0 73
6 7
6 0 1 1 0 0 1 1 1 0 7
6 74
6 7
4 1 0 0 1 0 0 1 1 0 56
0 0 0 1 1 0 0 0 1 9

1 2 5 7 8 3 4 6 9
2 3
0 0 0 0 0 0 0 0 0 1
60 0 0 7
6 0 0 0 0 0 0 72
6 7
60 0 1 0 0 0 0 0 0 7
6 75
6 7
60 0 0 1=2 0 0 0 0 0 77
6 7
Ym ¼ 6
60 0 0 0 0 0 0 0 0 7
78
6 7
60 0 0 0 0 0 0 0 0 73
6 7
60 0 0 0 0 0 0 0 0 7
6 74
6 7
40 0 0 0 0 0 0 0 0 56
0 0 0 0 0 0 0 0 0 9
84 13 Method of Arbitrary Input Data
1 2 5 7 8 3 4 6 9
2 3
0 0 0 0 0 0 0 0 0 1
60 7
0 7 2
6 0 0 0 0 0 0 0
6 7
60 0 0 0 0 0 0 0 0 7
6 75
6 7
60 0 0 0 0 0 0 0 0 77
6 7
Zn ¼ 6
60 0 0 1 0 0 0 0 0 778
6 7
60 0 0 0 0 0 0 0 0 73
6 7
60 0 0 0 0 0 0 0 0 7
6 74
6 7
40 0 0 0 0 0 0 2 0 56
0 0 0 0 0 0 0 0 0 9

The voltage and current input data are as follows:

1 2 5 7 8 3 4 6 9
Ii ¼ ½ 10 0 0 2 0 0 0 0 0

1 2 5 7 8 3 4 6 9
VV ¼ ½ 6 1 0 0 0 0 0 0 0 

After substituting the input data into the solution equations

I ¼ Ti ðIi þ Fy1 Ym Tv Vv Þ

V ¼ Tv ðVv þ Fz1 Zn Ti Ii Þ

The final results in the form of current and voltage vectors are as follows:

1 2 5 7 8 3 4 6 9
I ¼ ½ 6 1 1 1 2 2 3 2 3

1 2 5 7 8 3 4 6 9
V ¼ ½ 10 9 1 2 1 8 7 3 1 

Note that all unknown branch parameters (impedances/admittances) can be

easily found from the above results.
The main part of computer program is the procedure of finding the needed
current and voltage topologies. Algorithm of finding simple tree and complemen-
tary cotree of a graph is not a problem. Finding the cotree separate, not as a
complement to the tree, generally is very difficult (it is possible in special graphs
only). Finding the current cotree and the voltage tree, according to the solvability
conditions, can be done using various algorithms. Note that both topologies can be
found using the procedure of finding a tree of graph only (voltage tree from
13.4 Numerical Example 85

v-branches, i, v-branches, v, f-branches, and f-branches and a current cotree which is

a complement to a tree from v-branches, v, f-branches, o-branches, and f-branches).
Generally, the arbitrary input data method can be applied to the solution of
problems with various distributions of input data. For example, in power system
operation problems, the method can be used in the problem of power system state
estimation.
Chapter 14
Logical Optimization Method
of Network System

14.1 Example of Network System Matrices

In this chapter, the logical method of finding the quasi-optimal state of network
system is derived, using the system admittance and system impedance matrices YS
and ZS of a network [see Chap. 8, Eqs. (8.37) and (8.38)]. The logical method is
illustrated on a simple example of nine-branch power system network. The input
data are the system admittance YS and system impedance ZS , which are found using
the relations derived in Sect. 8.3 as follows:
1. The starting point is the chosen topology of network system and the current and
voltage vectors I and V in Fig. 14.1.
The topology of network may have essential influence on the results of com-
putations. The star-like form of tree used in power system networks (tree
branches are node-to-earth branches) may lead to negative results. In example of
network in Fig. 14.1, we assume that the tree consists of branches 4, 6, 7, and 8
and the cotree consists of branches 1, 2, 3, 5, and 9.
2. The active and passive branch parameters should be chosen, particularly the
ideal current sources in cotree and ideal voltage sources in tree. The remaining
branches are impedances in cotree and admittances in tree. In our example, we
assume that currents of branches 1 and 2 are ideal current sources and voltage of
branch 4 is an ideal voltage source.
3. Using the elements of current vector I and elements of voltage vector V (see
Fig. 14.1), the values of branch impedances and branch admittances are cal-
culated. It makes possible the formulation of following network parameters:
matrix Fm;n [see Chap. 8, Eq. (8.24)] and network system state vector KI;V;F [see
Chap. 8, Eq. (8.24)]. The ideal sources in matrix Fm;n are indicated by number
1, and numerical values of sources are elements of vector KI;V;F .
4. Using inverted topological matrix T
1 , the equation Fm;n T
1 I þ Fm;n T
1 V ¼ Km;n
can be formulated [see Eqs. (8.25)–(8.29)].

© Springer International Publishing AG 2017 87

A. Kłos, Mathematical Models of Electrical Network Systems,
Lecture Notes in Electrical Engineering 412,
DOI 10.1007/978-3-319-52178-7_14
88 14 Logical Optimization Method of Network System

Fig. 14.1 Network graph and 6

current voltage state of 5 1 8 1
5 7 -2 2 12 2
network 9
1 3 9 3
8
2 5 3 5
I = 3 9 V = -4 9
4 3 2 2 1 -4 4 6 4
-6 6 2 6
2 7 1 7
Tree Cotree
5 8 3 8
Earth node

5. Reducing vector Km;n to zero by transferring currents and voltages from

this vector to the vectors I and V, we have RI I þ RV V ¼ 0 [see Eqs. (8.30) and
(8.31)].
6. Finally, the system admittance matrix YS and the system impedance matrix ZS
are found from equations YS ¼
R
1
1
I RV , ZS ¼
RV RI , and the system equa-
tion I ¼ YS V and V ¼ ZS I are as follows:

2 3 2 1 2 3 5 9 4 6 7 8 32 3
5
4:000
0:167 0:111 0:667
0:000 4:000 1:000
0:000 0:000 8 1
6
2 7 6
1:000 1:500 7 6 12 7
7 6
6 7 6 0:750 0:111 0:667 1:875 0:000 0:000 1:000 7 2
6 7 6 76 7
61 7 6
1:000
0:167 1:000 0:667 0:750 0:000 0:000 1:000 0:000 7 6 7
6 7 6 7 69 7 3
6 7 6 76 7
62 7 6
1:000
0:167 0:111 1:000 0:750 0:000 0:000 1:000 0:000 7 6 3 7 5
6 7 ¼ 6 76 7
63 7 6 0:000 0:147
0:111
0:667
0:125 0:000 0:000
1:000
1:500 7 6
4 7 9
6 7 6 76 7
6 7 6 76 7
6
4 7 6 4:000 0:000 0:000
0:667 0:000
4:000
1:000 0:000
0:000 7 6 6 7 4
6 7 6 76 7
6
6 7 6 3:000 0:000 0:000 0:000 0:000
3:000
0:000
0:000 0:000 7 62 7 6
6 7 6 76 7
6 7 6 76 7
42 5 4
0:000
0:333 0:222 1:133 1:500 0:000 0:000 2:000 0:000 5 4 1 5 7
5
0:000
0:147 0:000
0:000 1:875 0:000 0:000
0:000 2:500 3 8

I ¼ YS V

2 3 2 1 2 3 4 5 6 7 8 9 32 3
8
56:000
8:000 7:000 0:000 0:000
56:000
0:000 0:500 4:800 5 1
6 12 7 6
48:000
6:000 6:000 0:000 0:000
48:000 0:000 0:000 3:600 76
2 7
7 6 2
6 7 6 7
6 9 7 6
63:000
9:000 9:000 0:000 0:000
63:000 0:000 0:000 5:400 7 6 7
6 7 6 76 1 7 3
6 3 7 6
0:000 0:000 0:000 3:000 0:000
0:000 1:000
1:500 0:000 7 6 2 7 5
6 7 6 76 7
6
4 7 ¼ 6
0:000 0:000
0:000 0:000 0:571
0:000
0:000
0:286 0:343 7 6 7
6 7 6 76 3 7 9
6 6 7 6
56:000
8:000 7:000 0:000 0:000
56:000
0:333 0:500 4:000 7 6
4 7 4
6 7 6 76 7
6 2 7 6 0:000
0:000
2:000 0:000
1:000
0:000 7 6 7
6 7 6 0:000 0:667 1:000 76
6 7 6
4 1 5 4
1:000 0:000
0:000
2:000
0:429
1:000 0:667 1:286 0:257 54 2 5 7
3
8:000
1:000 1:000 0:000
0:429
8:000
0:000
1:214 0:743 5 8
I¼ Ys V

Equations I ¼ YS V and V ¼ ZS I make possible the analytical investigation of

current voltage functions in a network system. Elements of matrix YS are system self
and mutual admittances of network branches, and elements of matrix ZS are system
self and mutual impedances of network ranches. They are the coefficients of branch
currents as a function of branch voltages and coefficients of branch voltages as a
function of branch currents. Mathematically, they are derivatives di/dv and dv/di of
current voltage functions. Generally, the elements of matrices YS and ZS charac-
terize the quantitative influence of voltages on currents and currents on voltages in
the network system.
14.2 Logical Analysis of Network System Equations 89

14.2 Logical Analysis of Network System Equations

In this section, the analysis of system equations is illustrated on a simple example of

the state of network as shown in Fig. 14.1. Various analytical methods may be
used, but it is of advantage using numerical method of how much currents change
as a function of unit voltage values and of how much voltages change as a function
of unit current values. It means that equations I ¼ YS V and V ¼ ZS I should be
substituted by suitable coefficients. Consider equation I ¼ YS V. Each element yj;k
of matrix YS is a coefficient of equation:

ij;k ¼ yj;k vk

Current ij;k is a part of current ij of branch j, and vk is an element k of voltage

vector V. Reducing this dependence ij;k on vk to the unit value of voltage vk , we
have the current-per-unit-voltage coefficient sij;k .

sij;k ¼ yj;k =vk

Analogically, in equation V ¼ ZS I, each element zj;k of matrix YS is a coefficient

of equation:

vj;k ¼ zj;k ik

Voltage vj;k is a part of voltage vj of branch j and ik is an element k of current

vector I. Reducing this dependence vj;k on ik to the unit value of current ik , we have
the voltage-per-unit-current coefficient svj;k .

svj;k ¼ zj;k =ik

Using coefficients sij;k svj;k , one can solve various practical problems. Let us
analyze current voltage interrelations in some branches of a network system in
Fig. 14.1.
1. The first row of equation I ¼ YS V concerns current of branch number 1 as a
function of voltage vector V. Branch 1 is an ideal current source; if it is not a
part of network, then it produces constant current and its admittance is equal to
zero. However, branch 1, if it is an ideal current source in the network system,
has system self and system mutual admittances, which are “thrown upon” on
branch 1 by a whole network system. It means that in network system, the
branch number 1 is as well the ideal current source as the admittance branch.
The nominal value of the ideal current source of branch 1 remains unchanged,
but it is a function of network voltages (elements of vector V) and the system
self and system mutual admittances. Numerically, the analysis of how the
90 14 Logical Optimization Method of Network System

current of branch 1 depend on network voltages can be done using the coeffi-
cients si1;k which are as follows:

Branches 1 2 3 5 9 4 6 7 8
si1;k −0.500 −0.014 0.014 0.222 0.000 0.667 0.500 0.000 0.000

It means that the greater positive influence on the current of branch 1 has the
voltage of branch 4, and the greater negative influence has the voltage of branch 1.
Not any influence has voltages of branches 9, 7, and 8. If the current of branch 1 is
to large then in order to decrease this current it is of advantage to decrease the value
of ideal voltage source of branch 4 and decrease voltage of branch 6, and if it is not
possible then increase voltage of branch 1 and decrease voltage of branch 6. The
qualitative estimate of how much voltage value should be increase can be done
using coefficients si1;k .
2. The first row of equation V ¼ ZS I concerns voltage of branch 1 as a function of
voltage vector V. Branch 1, if it is an ideal current source in the network system,
has system self and system mutual impedances, so branch 1 is as well the ideal
voltage source as the impedance branch. The nominal value of ideal current
source of branch 1 is an element of current vector I. The voltage of branch 1 is a
function of network currents (elements of vector I) and the system self and
system mutual impedances of branch 1. Numerically, the analysis of how the
voltage of branch 1 depend on network current vector 1 can be done using the
coefficients sv1;k . Such coefficients for the first row of equation V ¼ ZS I are as
follows:
Branches 1 2 3 5 9 4 6 7 8
sv1;k −11.200 4.000 7.000 0.000 0.000 14.000 0.000 0.250 0.960

It means that the greater positive influence over the voltage of branch number 1 has
the current of branch 4, and the greater negative influence has the current of branch 1.
Not any influence has currents of branches 5, 9, and 6. If the power of branch number 1
must be increase, taking into account the coefficients si1;k and sv1;k , then the best
operation is to decrease the value of ideal voltage source of branch 4 and to decrease
the current of branch 4, and if it is not possible then decrease the current of branch 3
and increase the current of branch 1. The qualitative estimate of how much power
value should be increase can be done using coefficients si1;k and sv1;k .
3. The third row of equation I ¼ YS V concerns the current of branch 3 as a
function of voltage vector V. Branch 3, if it is not a part of network, has its own
self-admittance equal to 1/9 (no mutual admittances), but if it is a part of
network in Fig. 14.1 then the system self-admittance of branch 3 has also the
14.2 Logical Analysis of Network System Equations 91

same value 1/9. It is not a rule, because the same kind branch number 5 has the
self-admittance equal to the system self-admittance. Additionally, branch 3 has
system mutual admittances, which characterize the difference between the
classical self and mutual branch parameters and the system self and system
mutual branch parameters. The current of branch 3 is a sum of products of
network voltages (vector V) and system self and system mutual admittances of
branch 3. Numerically, the analysis of how the current of branch 3 depend on
network voltage vector V can be done using the current-per-unit-voltage coef-
ficients si3;k which are as follows:

Branches 1 2 3 5 9 4 6 7 8
si3;k −0.125 −0.014 0.111 0.222 −0.188 0.000 0.000 1.000 0.000

The greater positive influence over the current of branch 3 has the voltage of
branch 7, and not any influence has the voltages of branches 4, 6, and 8. In case if
current of branch 3 is too small, then the optimal operation is to increase the voltage
of branch 7. Note that the ideal voltage source of branch 4 cannot be used because
si3;4 ¼ 0.
4. The third row of equation V ¼ ZS I concerns the voltage of branch 3 as a
function of current vector I. Branch 3, if it is not a part of network, has its own
self-impedance equal to 9 (no mutual impedances), and if it is a part of network
then its system self-impedance has the same value, but additionally branch 3 has
system mutual impedances. The current of branch 3 is a sum of products of
network currents and system self and system mutual impedances. Numerically,
the analysis of how the voltage of branch 3 depends on network currents can be
done using the voltage-per-unit-current coefficients sv3;k .

Branch 1 2 3 5 9 4 6 7 8
sv3;k −12.600 4.500 0.9.000 0.000 0.000 15.750 0.000 0.000 1.080

It means that the greater positive influence over the voltage of branch 3 has the
current of branch 4, and not any influence has voltages of branches 5, 9, 6, and 7.
If the power of branch 3 should be increase, instead of diminishing current of
branch 3 and voltage of branch 3, then it is better to decrease the current of branch 4
and decrease the voltage of branch 7 (see coefficients sv3;k and si3;k ).
5. The sixth row of equation I ¼ YS V concerns the current of branch 4 (which is
ideal voltage source) as a function of voltage vector V. If branch 4 is not a part
of network, then it is constant ideal voltage, and its admittance is equal to zero.
Branch 4, as an ideal voltage source in the network system, has system self and
system mutual admittances, which are “thrown upon” on branch by a whole
network system, so it is both voltage source and admittance branch. The
92 14 Logical Optimization Method of Network System

nominal value of an ideal voltage source is an element of voltage vector V.

Numerically, the analysis of how the current of branch 4 depends on network
voltages can be done using the si4;k .

Branch 1 2 3 5 9 4 6 7 8
si4;k −0.500 −0.000 0.000 0.222 0.000 −0.667 −0.500 0.000 0.000

It means that the greater negative influence on the current of branch 4 has the
ideal voltage source of branch 4, and the greater positive influence has the voltage
of branch 5. Not any influence has voltages of branches 2, 3, 9, 7, and 8. In case if
the current of branch 4 should be increase, then the best operation is increasing the
nominal value of ideal voltage source of branch 4: If it is not possible, then voltages
of branch 1 and branch 6 should be increased.
6. The sixth row of equation V ¼ ZS I concerns voltage of branch 4 (which is ideal
voltage source) as a function of current vector I. Branch 4, if it is not a part of
network, produces constant voltage and its impedance is equal to zero. Branch
4, as a part of network system, has system self and system mutual impedances,
which are thrown upon on branch by whole network system, so it is both the
ideal voltage source and system impedance branch. The constant value of ideal
voltage source of branch 4 is an element of voltage vector V and is a sum of
products of network currents and system self and system mutual impedances.
The analysis of how the voltages of branch 4 depend on network currents can be
done using svj;k :

Branch 1 2 3 5 9 4 6 7 8
sv4;k −11.200 4.000 7.000 0.000 0.000 14.000 0.000 0.250 0.960

The greater negative influence on the voltage of branch 4 has the current of
branch 4, and the greater negative influence has the current of branch 1. Not any
influence has currents of branches 5, 9, and 6.
If the power of branch 3 should be increase, then the current of branch 4 should
be increase and the ideal voltage source of branch 4 should be increased.
Note, e.g., that the coefficient sv4;1 = −11.200 has much higher negative value
then the coefficient si4;1 = 0.5, and the coefficient sv4;4 = 14.000 has much higher
positive value then the coefficient si4;4 = 0.667. Generally, it means that the currents
have greater influence on voltages and voltages have littler influence on currents.
7. The row number 9 of equation and V ¼ ZS I concerns the voltage of branch 8 as
a function of current vector I. Branch 8, if is not a not a part of network, has its
own self-impedance and no mutual impedances. If it is a part of our network
then has system self-impedance and system mutual impedances thrown upon by
14.2 Logical Analysis of Network System Equations 93

a whole network system. Numerically, the analysis of how the voltage of branch
8 depends on network currents can be done using the sv8;k coefficients.
Branch 1 2 3 5 9 4 6 7 8
sv8;k −1.600 −0.500 1.000 0.000 −0.143 2.000 0000 −0.107 0.143

It means that the greater positive influence on the voltage of branch 8 has the
current of branch 4; and the greater negative influence has the current of branch 1.
Not any influence has the voltages of branches 5 and 6. In case if the voltage of
branch 4 should be increase, then the best operation is diminishing the current of
branch 4 and increasing the nominal value of ideal current source of branch 1. If it
is not possible, then currents of branch 1 should be increased and of branch 3
should be increase.
8. The row number 9 of equation I ¼ YS V concerns current of branch 8 as a
function of voltage vector V. Branch 8, if it is a part of our network, has system
self-admittance and system mutual admittances thrown upon on branch 8 by
whole network system. The current of branch 8 is a sum of products of network
voltages and system self and system mutual impedances. The analysis of how
the current of branch 8 depends on network voltages can be done using the si8;k
coefficients which are as follows:
Branch 1 2 3 5 9 4 6 7 8
si8;k 0.000 0.035 0.000 0.000 −0.469 0.000 0000 0.000 0.833

It means that branch 8 has system self-admittance and only two system mutual
admittances. The greater positive influence on the current of branch 8 has the
voltage of branch 8, and the greater negative influence has the voltage of branch 9.
If the power of branch 8 should be increase, then the current of branch 4 and the
voltage of branch 8 should be increase.
Each of the above-presented examples of logical analysis and application
equations I ¼ YS V and V ¼ ZS I is the first step only in the logical method of
finding the quasi-optimal state of network system. Note that if in any of the above
examples the branch current or branch voltage is changed, then current and voltages
in all branches are changed. If the vectors and matrices, before and after change, are
not in prescribed limits, then calculation must be repeated. The new current vectors
I and V, and new matrices YS and ZS are to be found, and the logical analysis as
above should be repeated in iteration process.
94 14 Logical Optimization Method of Network System

14.3 Application Possibilities

The applications of a method illustrated in previous section are restricted, because

in some cases the matrices YS and ZS are singular, and the equations I ¼ YS V,
V ¼ ZS I does not exist. It depends mainly on the number and configuration of ideal
current and voltage sources in a network system. In order to illustrate the appli-
cation possibilities, the various cases of number and distribution of ideal sources are
done using a simple example of network system as shown in Fig. 14.1. In Sects. 8.1
and 8.2, it was shown that using the input data and assuming the ideal sources in
branches in branches 1, 2, and 4, the matrices YS and ZS are non-singular, and the
logical analysis is possible. The same results are obtained using the same input data
and assuming up to three ideal sources located differently in network branches.
Following computations were done.
– 1 source located in branch 3 or in branch 6 or in branch 8.
– 2 sources located in branches 3, 5, in branches 2, 6, or in branches 5, 8.
– 3 sources located in branches 1, 2, 3 or in branches 4, 6, 9.
The same (positive) results are obtained if 4 sources are assumed located in
branches 1, 2, 3, 5 or branches 4, 6, 7, 8 only, but for other locations of 4 sources
results are negative.
If 4 sources are located in branches 1, 2, 3, 6, then matrix YS is non-singular and
matrix is singular. It means that equation I ¼ YS V exists and equation V ¼ ZS I
does not. The logical analysis is restricted.
If 4 sources are located in branches 4, 6, 7, 8, then matrix ZS is non-singular and
matrix YS YS is singular. It means that equation V ¼ ZS I exists and equation I ¼
YS V does not. The logical analysis is restricted.
If 5 and more sources are assumed, then the relation between the equations
I ¼ YS V and V ¼ ZS I does not exist and the logical analysis cannot be done.
Interesting is the case of network system in Fig. 14.1, in which there are no
sources and all branches are current–voltage functions only. In this case, both
matrices YS and ZS are non-singular, and the equations I ¼ YS V and V ¼ ZS I does
exist, so the logical analysis can be done. Note the important difference between the
classical matrices Y, Z, and the system matrices YS and ZS in this case. The elements
of classical matrices are pure branch parameters independent of network, and the
elements of system matrices are system parameters dependent of network system.
In our case, the elements of the first one are the self parameters of branches,
independent of network, and the elements of the last one are the self and mutual
system parameters, which are “thrown upon” on the elements by network system. It
illustrates the difference between a network as a sum of branches and a network
system. Network system is a unique physical quantity, which has its own features
and laws.
As it comes out from the above, the logical method of finding the quasi-optimal
state of network system can be computerized and practically implemented in
electrical power systems operation and development. Such method may help, e.g.,
14.3 Application Possibilities 95

to assess the reliability of power system in the case of over and under-loads, or over
and under-voltages of network branches (lines, transformers). The power system
operator, having the current and voltage state of power system (e.g., from the state
estimation) and using computerized method presented above, can be nearly
immediately provided with necessary information. Such a tool may help to restore
the power system security in cases of faults including blackouts.
Chapter 15
Current-Based Method of Load
Flow Solution

Using classical input data, the process of finding the load flow in power system
networks is done in two steps. First, the voltage state vector is calculated; second,
the needed output quantities (branch power loads) are found. In recently used
methods, the voltage state vector, which is a vector of all nodal (node-to-earth)
voltages, is calculated using iteration process. However, taking into account the
algebraic model of network system described in Part I of book and the general
network solution method derived in Part II Chap. 11, the load flow can be found
using current state vector. The current-based load flow method for any electrical
network can be derived using general network solution method (Sect. 11.1). The
derivation may be complicated because the b-dimensional Eqs. (11.15) and (11.16)
must be reduced to n and m dimensional. In case of power system load flow
method, the input data are known (classical input data), so it simplifies the
derivation of method. However, in order to find the current state vector, the suitable,
current state-oriented topology of network (different than the classical one) should
be chosen. In this chapter, the current-based load flow method (CBLF) is derived
and verified.

15.1 Current State-Oriented Topology of Network

Consider any power system network. Suppose that there are given the classical load
flow input data. If there are known load values of all node-to-earth branches sn;j and
nominal network voltage Vn , then the nodal current values in;j ¼ sn;j =Vn are very
good estimate of real current values. Such nodal currents vector could not be used
as the current state vector because it is not a current vector of any set of cotree
branches. However, such vector can help to find the current state-oriented topology
of network. In order to use the nodal current vector as a part of current state vector,
the elements of nodal current vector must belong to the currents of cotree branches.

© Springer International Publishing AG 2017 97

A. Kłos, Mathematical Models of Electrical Network Systems,
Lecture Notes in Electrical Engineering 412,
DOI 10.1007/978-3-319-52178-7_15
98 15 Current-Based Method of Load Flow Solution

Fig. 15.1 Illustration of 6

current state-oriented
5 7
topology of network 9 2 3 4 5 9
1 1 1 1 0 6
8
C= 1 1 0 1 -1 7
1 2 3 4 0 1 0 0 -1 8
Cn Cc
Tree Cotree

Earth node

Let us subdivide a network graph into a tree and a cotree as it is shown in Fig. 15.1.
The set of tree branches include one node-to-earth branch [a slack node branch (in
Fig. 15.1 branch 1)] and d freely chosen node-to-node branches (in Fig. 15.1
branches 6, 7, 8). The cotree include the remaining branches, among them:
n node-to-earth branches (excluding slack-node-to-earth branch), and
c node-to-node branches which doesn’t belong to a tree (in Fig. 15.1 branches 2, 3,
4, 5, 9).
A tree defined as above determines d independent cut-sets of a network, and
each of them is defined by a tree branch. Each network cut-set includes one tree
branch and a set of cotree branches parting this cut-set from the rest of network. The
cut-set incidence matrix C, of order d  ðn þ cÞ, relates cotree branches (columns)
to tree branches (rows). Each row of C represents the cut-set defined by a tree
branch. Note that the submatrix Cm transposed is a loop-set incidence matrix in the
classical node-to-node network. Matrix C can be written as follows:

C ¼ ½ Cn Cm  ð15:1Þ

where
Cn submatrix relating tree branches to node-to-earth cotree branches of order
d  n and
Cc submatrix relating tree branches to node-to-node cotree branches of order
d  c.
The matrix C for an example of network is shown in Fig. 15.1.
A cotree defined as above determines n þ c independent loop-sets of network,
and each of them is defined by a cotree branch. Each network loop-set includes one
cotree branch and a set of tree branches forming a closed loop.

15.2 Current State Vector and Current Flow Method

The current state vector Im in the above described topological structure includes two
subvectors:
15.2 Current State Vector and Current Flow Method 99

 
I
Im ¼ n ð15:2Þ
Ic

where
In current vector of node-to-earth cotree branches of order n and
Ic current vector of node-to-node cotree branches of order c.
The current state vector Im can be derived using general network solution
method (see Chap. 11). However, the derivation will need laborious decomposition
of b-dimensional equations, which can be substituted by the following simpler
derivation.
Let us start from the Kirchhoff’s current law for d independent cut-sets:

Id þ Cn In þ Cc Ic ¼ 0 ð15:3Þ

where Id —current vector of tree branches excluding slack node branch of order d.
Using Kirchhoff’s voltage law for m loop-sets defined by a node-to-node cotree
branches,

Vc  CcT Vd ¼ 0 ð15:4Þ

where
Vc voltage vector of node-to-node cotree branches of order c and
Vd voltage vector of tree branches excluding slack node branch of order d.
Ohm’s law for a node-to-node tree branches and node-to-node cotree branches:

V d ¼ Z d Id ð15:5Þ

Vc ¼ Zc Ic ð15:6Þ

where
Zd matrix of node-to node tree branch impedances of order d and
Zc matrix of node-to-node cotree branch impedances of order c.
Substituting Vd and Vc from Eqs. (15.5) and (15.6) to the Eq. (15.4), we have the
following:

Zc Ic  CcT Zd Id ¼ 0 ð15:7Þ

Substituting Id from Eq. (15.3) to Eq. (15.7) we have the following:

Zc Ic þ CcT Zd Cn In þ CcT Zd Cc Ic ¼ 0 ð15:8Þ

100 15 Current-Based Method of Load Flow Solution

Finally, doing simple algebraic operation, the current vector of node-to-node

cotree branches Ic can be found as follows:

Ic ¼ ðZc þ CcT Zd Cc Þ1 CcT Zd Cn In ð15:9Þ

In Eq. (15.9), the matrices Cn ; Cc and Zc ; Zd are known from input data. It means
that the current vector Ic is a linear function of nodal currents In .
Denoting

F ¼ ðZc þ CcT Zd Cc Þ1 CcT Zd Cn ; ð15:10Þ

the current state vector Im is as follows:

   
In 1
Im ¼ ¼ ½I  ð15:11Þ
Ic F n

where vector In is a current vector of nodal (node-to-earth) branches, which can be

found from nodal loads and nominal voltage of network system.
However, using the physical interpretation of Eq. (15.9) and taking into account
the topology of network, the current state vector can be partially found “by
inspection” as follows:
Denoting the parts of Eq. (15.9) as follows:

ZF ¼ ðZc þ CcT Zd Cc Þ ð15:12Þ

VF ¼ CcT Zd Cn In ; ð15:13Þ

the current vector Ic is as follows:

Ic ¼ ðZF Þ1 VF In ð15:14Þ

where VF is the loop-set voltage vector of order c, and is the analogy to the node-set
current vector. Each element of vector VF is the sum of voltages along the loop-set.
ZF is the loop-set impedance matrix of order c, which is the number of loop-sets
in the classical node-to-node network, and is the analogy to the node admittance
matrix. Each diagonal element of a matrix ZF is a sum of branch impedances
forming a loop-set, and each off-diagonal element is a sum of the impedances of the
branches common to the ith and jth loop-sets. In real networks, the impedance
matrices Zm and Zd are non-singular, so the matrix ZF is non-singular.
It means that the vector ZF can be found by inspection as follows:
   
In 1
Im ¼ ¼ ½I  ð15:15Þ
Ic ZF1 VF n
15.2 Current State Vector and Current Flow Method 101

Finally, the current state vector Im can be found using either the mathematical
(15.11) or physical (15.15) form. Having the current state vector, the current and
voltage vectors I and V can be easily found.
The algorithm of current flow method is based on the above Eqs. (15.3)—
(15.11). Using the numerical sparse technique, one can substantially reduce the
scope of numerical calculations. There is also a possibility of finding the matrix ZF
and the vector VF by inspection from network graph, subdivided as in Fig. 15.1 into
a tree and a cotree.

15.3 Current-Based Load Flow Method

In this section, the current-based load flow method (CBLF) in power system net-
work system is derived, using the above derived current flow method. The starting
point is the classical load flow input data:
– Graph of network in the form of a set of b interconnected, closed branches
– Constant voltage value of any one slack node branch
– Active and reactive powers (P, Q loads), and active powers and voltages (P, V
loads) of node-to-earth branches except a slack node branch
– Passive parameters of node-to-node branches
The CBLF method uses the iteration process in which the calculation of current
state vector is repeated, until initially assumed node-to-earth voltages converge to
the right values and reactive power in P, V branches are calculated.
The algorithm of the current-oriented load flow method includes steps before
and in iteration process.
1. Choose a tree of a network—including slack node-to-earth branch and any set
of node-to-node branches. The remaining branches are a set of complementary
cotree.
2. Find incidence matrix C ¼ ½ Cn Cn  of the tree–cotree structure.
3. Assume a reasonable starting values of reactive power Q in P, V branches.
4. Find current vector In of node-to-earth cotree branches using nodal powers and
slack node voltage.
5. Find impedance matrices Zc , Zd from input data.
6. Find current vector Ic from the Eq. (15.9).
7. Find current state vector Im from Eq. (15.11).
8. Find current vector of network system I using current flow method.
9. Find voltages of node-to-earth branches.
10. Find the active and reactive power values (P, Q loads) or active powers and
voltage values (P, V loads) of node-to-earth branches.
11. Check the voltage mismatches in PV nodes and, if all they are not in the
prescribed limits, correct the reactive powers Q in PV nodes.
102 15 Current-Based Method of Load Flow Solution

12. Check the power mismatches and if they are not in the prescribed limits then go
to the iteration number 3 else go to the end of iteration process.
The method of obtaining the needed voltage values in PV nodes is based on the
evaluation of correction coefficients DV/DQ for each PV node and in each iteration
(starting from a second one). According to this coefficient and voltage in P, V
nodes, the reactive power Q is changed using accelerating coefficient.

15.4 Verification of Method

In power systems, the load flow calculations are done very often using various
methods mainly based on finding voltage state vector. Since publication in 1973 by
B. Stott and O. Alsac [ ], the fast decoupled load flow method (based on voltage
state vector) and its improved versions are practically used worldwide to this day.
The load flow method based on current state vector is a new method and as such
needs practical verification on example of real power system network. The
current-based load flow method (CBLF) was compared with the classical fast
decoupled method (FDLF). Case studies were done on the three test networks as
follows:
– Test network I—400/220 kV transmission network, 101 nodes (node-to-earth
branches), 6 of them are PV nodes (node-to-earth branches), and 140
node-to-node branches (the real high-voltage power system network)
– Test network II—110 kV network, 77 nodes, 2 of them are PV nodes 92
node-to-node branches—The part of real 110 kV power system network sup-
plied from one 400 kV substation
– Test network III—real 15 kV network, 519 nodes, and 524 node-to-node
branches
For each of the test network, the load flow calculations were done using CBLF
and FDLF computer programs written in Pascal language. Following results are
compared:
– Ite—Number of iterations
– T—Total computation time
– Ti—Time of one iteration
– Tp—Time of preliminary (prior to the iterations) computations
The times (T, Ti, and Tp) in the FDLF method are given in relation to the times
in the BOLF method, which are taken equal to 1. The computations were done
assuming in both programs the same power accuracy and flat start voltage values.
Case study 1
The load flow calculations for the test network I were done for the following
loading states:
15.4 Verification of Method 103

– Lmin—Minimal loading state

– Ln—Normal (average) loading state
– Lmax—Maximal loading state
– Lf—After failure loading state—the outage of two generating units and four
400 kV lines after short circuit on 400 kV bus bars
The both CBLF and FDLF programs were used, assuming power accuracy
DP < 1 MW, and in CBLF method, voltage accuracy in PV nodes DV < 1 kV. The
computation results are shown in the Table 15.1.
The greater differences of a number of iterations and computation times in FDLF
method, compared with CBLF method, in loading states Lmax and Lf, come from
the additional iterations needed in FDLF method to correct the Q values in PV
nodes (if Q > Qmax). Such additional iterations are not needed in the CBLF method
because to keep the given voltage values in the PV nodes, the corrections of Q
values are done in each iteration.
In order to show the convergence ability of both methods, the numbers of
iterations were compared for the test network I, by normal loading state Ln,
increased by coefficient g to the divergence of iteration process. The load flows in
both methods were done assuming all nodes as PQ nodes (in order to avoid the
additional iterations due to the necessary changes in Q values in FDLF method).
The number of iterations is shown in Table 15.2.
Case study II
The load flow calculations for the test II were done for the following loading states:
– Ln—Normal (average) loading state
– Lmax—Maximal loading state

Table 15.1 The computation results for CB and FD

Lmin Ln Lmax Lf
CB FD CB FD CB FD CB FD
Ite 8 9 11 13 13 18 14 27
T 1 1.1 1 1.3 1 1.35 1 1.9
Ti 1 0.96 1 0.96 1 0.97 1 0.97
Tp 1 1.34 1 1.35 1 1.70 1 1.75
CB—CBLF method, FD—FDLF method

Table 15.2 The number of g 1 1.02 1.04 1.05 1.06 1.07 1.076 1.08
iterations
Ite 12 13 15 17 20 31 DIV
FDLF
Ite 10 11 12 13 14 22 33 DIV
CBLF
DIV—Iteration process diverges
104 15 Current-Based Method of Load Flow Solution

Table 15.3 The results of Ln Lmax Lf

computations for CB and FD
CB FD CB FD CB FD
Ite 3 9 5 12 10 15
T 1 2.52 1 2.97 1 1.98
Ti 1 1.18 1 1.20 1 1.21
Tp 1 2.14 1 1.86 1 1.51
CB—CBLF method FD—FDLF method

Table 15.4 Characteristic Normal load High load

feature of the CBLF and
CBLF NDLF CBLF NDLF
NDLF method
T 1 3.53 1 5.7
Ti 1 1.22 1 1.24
Ite 2 5 3 7

– Lf—After failure loading state—normal load state after the outage of one
generating station (40 MW) and four 110 kV lines due to the short circuit on
110 kV bus bars
The results of computations, shown in the Table 15.3, enable the comparison of
both methods.
The greater numbers of iterations and computation times in FDLF method come
mainly from the difference of an order of matrices to be inverted. In the CBLF
method, the order of a mesh impedance matrix ZF (see Eq. (15.12) is equal to
c = 14, while in FDLF method, the order of Jacobian matrix is equal to n = 76.
Case study III
The load flow calculations for 15 kV test network are done using COLF method
and the Newton’s decoupled load flow method (NDLF) because FDLF method
diverges. The results are given in Table 15.4.
Table 15.4 shows the characteristic feature of the CBLF method. Its effective-
ness rises with a lower ratio of the number of node-to node branches to the number
of nodes, which in this case is 5/519 = 0.01. In CBLF method, the total time of
computations and the number of iterations are above two times than in the NDLF
method. So, the CBLF method is more effective.

15.5 Conclusions

From the theoretical viewpoint, the CBLF method is based on a non-conventional

approach to the load flow solution. As such, it differs substantially from the other
methods. Compared with recently used methods, the main differences of the CBLF
method are as follows:
15.5 Conclusions 105

– From the mathematical viewpoint, CBLF is a linear method because the main
computation problem is finding the current state vector from a set of linear
equations, while in recently used methods, the main problem is finding a voltage
state vector from a set of nonlinear equations.
– Generally, CBLF method is numerically more stable, compared with classical
methods. Finding load flows in more loaded networks needs less iterations.
– The topological structure (tree and cotree) of a network graph in CBLF method
is different than in classical methods. The tree of network, instead of
node-to-earth branches, consists of some node-to-node branches and slack
node-to-earth branch. Remaining branches including node-to-earth branches are
cotree branches.
– The final results of computations, apart of voltages and load flaws, include
additionally current flaws in a network.
From the practical viewpoint, the main differences of CBLF method are as
follows:
– Instead of solving, in each iteration, a set of n (number of nodes) equations, the
set of c equations is solved. It makes the effectiveness of COLF method
depending very much on a coefficient k

k ¼ c=n ð15:16Þ

where c is the number of node-to-node branches in a cotree and n is the number of

node-to-earth branches mines one in the network.
– In power system networks, the effectiveness of CBLF method depends on the
voltage level. For a high-voltage networks (400 and 220 kV) in normal load
conditions, the effectiveness of CBLF is nearly the same as the classical
methods; however, in abnormal conditions, it may be much higher
(see Tables 15.1 and 15.2). In cases of voltage levels less than 400 kV, CBLF
method is faster then the classical methods (see Tables 15.3 and 15.4).
– As to the disadvantages, the PV nodes in the CBLF method must be specially
treated. In the iteration process, they are substituted by PQ nodes, and in order to
keep the voltages on the prescribed values, the corrections of reactive power Q
in PV nodes are needed in each iteration. The algorithm of correction (used in
the case studies 1 and 2 shown above) is based on the evaluation of V=Q
sensitivities and the application of an accelerating coefficient. The correction
algorithm used is optimal (minimum number of iterations) for the case studies
above but it may be not optimal for other network structures.
– The CBLF method is universal. It can be effectively used for all kinds (voltage
levels) of a meshed and open power system networks.
– From the computational viewpoint, the CBLF computer program needs less
computer space than the recently used load flow programs.