You are on page 1of 77

ELEKTRO-MFE 20008

Examensarbete 30 hp
October 2020

Investigation of Three-Phase
Balancing Techniques
A Comparative Study of Different Solutions
With Respect to Telecom Industry Needs

Mahir Mahdee

Masterprogram i förnybar elgenerering


Master Programme in Renewable Electricity Production
Abstract
Investigation of Three-Phase Balancing Techniques

Mahir Mahdee

Teknisk- naturvetenskaplig fakultet


UTH-enheten In the telecom industry, one-phase rectifiers are commonly used in
base stations and site power systems. The abundant of this one phase
Besöksadress: loads connected to the three- phase network where three-phase
Ångströmlaboratoriet
Lägerhyddsvägen 1 installations are required can enhance the amount of unbalances in
Hus 4, Plan 0 the three-phase systems. Unbalanced systems can cause the three-phase
connections to be out of the norm, i.e. it can be out of the
Postadress: requirements of the distribution company or installation standards
Box 536
751 21 Uppsala and thus shall not be allowed to be connected to the network.
Additionally, unbalance can cause problems in terms of energy loss in
Telefon: kWh, greenhouse emissions, which can also impose specific standards
018 – 471 30 03 and guidelines for three-phase electricity users and manufacturers.
Telefax: Those problems are addressed in the literature.
018 – 471 30 00
This thesis investigates the requirements and standards for phase
Hemsida: balancing in three- phase installations and uncovers solutions to
http://www.teknat.uu.se/student
mitigate these unbalances in the perspective of the telecom industry.
Through market investigation, the knowledge and solutions to this day
have been found and summarized, which is then used to further dig
into the technical understanding of phase balancing techniques. The
investigation shows that there is only a small market that requests
and has standards for phase balancing of loads. The different
solutions evaluated and modelled shows they can mitigate the load
balancing issue in three-phase installation likey to a specific limit
according to required standards.

Handledare: Ali Can Arslan


Ämnesgranskare: Juan de Santiago
Examinator: Irina Temiz
ELEKTRO-MFE 20008
I

Popular Scientific Summary

With the advancement in renewable energy generation and new technologies, the pres-
ence of unknown sources and the continues changes in the energy consumption trends
have led to new challenges in modern days electric power systems. In practical power
systems, it is usually impossible to maintain perfect balance or symmetry in phase cur-
rents and voltages. The abundant of these loads, such as single-phase loads, connected
to a three-phase network increases the amount of imbalances in the three-phase systems.

In the telecom industry, single-phase rectifiers are mostly used. These rectifiers are
used to power most telecom products within the telcom sites. Some markets in telecom
require the use of three-phase installations; single-phase converters connected to the
three-phase system may require phase balancing of loads.

This research aims to investigate the requirements and standards for phase balancing
in three-phase installations and to find solutions to mitigate these imbalances. Market
research has been conducted to understand the requirement of phase balancing. Different
ways to solve the issue of phase balancing from the perspective of the telecom industry
have been researched.

The study shows that there is only a small market that requires phase balancing of loads
where Japan is one of them with a phase unbalance standard of 30% maximum. Various
evaluated and proposed solutions – active and passive – have shown great potential
in minimizing the unbalance according to standards. The solutions were analyzed and
implemented in PSIM and Matlab/SIMULINK tools, respectively.

© Uppsala University Mahir Mahdee


II

Acknowledgements

This thesis becomes a reality with the warm-hearted support and help of many indi-
viduals. I would like to extend my sincere gratitude and appreciation towards all of
them.

First and foremost, I would like to express my heartfelt gratitude for this endeavour to
GOD, the almighty, for the wisdom he bestowed upon me in terms of strength, good
health and peace of mind in order to complete this research.

I am very thankful to everyone at Ericsson AB, namely my supervisor Ali Can Arslan
for mentoring me throughout the whole project. I would like to specially acknowledge
Stefan Signori, Fredrik Arvidsson X, Patrick Johansson CK, Lei Zhan, Qiang Liu X for
insightfully answering all my questions, providing valuable advice and guiding me
towards the right direction; I really appreciate their input.

I am grateful to my manager Martin Olsson PG for his continuous support with whatever
I asked for to complete this project. To everyone from the embedded power team, I am
humbled by all of their encouragement and motivation.

My deepest gratitude towards Dr Juan de Santiago, who was my supervisor at Upp-


sala University; he not only helped me with this research but also with many other
educational decisions throughout my Master’s study.

I would also want to extend my humble appreciation towards Uppsala University for
providing me with the full scholarship for perusing my Master’s study here in Sweden.

Finally, yet importantly, my ardent gratitude towards my family and friends all around
the world for their love, inspiration and moral support in overcoming the difficult times
I had in successfully completing this research work.

© Uppsala University Mahir Mahdee


Table of Contents III

Table of Contents

Abstract I

Popular Scientific Summary I

Acknowledgements II

List of Tables V

List of Figures VI

List of Acronyms VIII

1 Introduction 1
1.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.4 Review of Power Demand in Telecom Sites . . . . . . . . . . . . . . . . . 3
1.5 Structure of the Thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

2 Theory 5
2.1 Three Phase System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.2 Symmetrical Components . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.3 Definitions and Quantifications . . . . . . . . . . . . . . . . . . . . . . . . 8
2.4 Standards . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.4.1 Special Characteristics for Japan: . . . . . . . . . . . . . . . . . . . 10
2.5 Sources of Unbalance and its Consequence . . . . . . . . . . . . . . . . . 11
2.6 Environmental Aspects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

3 Market Overview 15
3.1 Merus Power . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
3.2 CME transformateur . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
3.3 Ferroamp . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

4 Solutions 19
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
4.2 Controlling the DC side of power supply units through load sharing . . 20
4.2.1 Redundancy: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

© Uppsala University Mahir Mahdee


Table of Contents IV

4.3 Three Phase Rectifier . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29


4.3.1 Three Phase PFC . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
4.3.2 DC-DC Conversion Stage . . . . . . . . . . . . . . . . . . . . . . . 34
4.3.3 Cost . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
4.4 Automatic Phase Balancing Technique - Solution Based on the Primary
Side of Load . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
4.4.1 Advantages of Automatic System . . . . . . . . . . . . . . . . . . 37
4.4.2 Mathematical Study . . . . . . . . . . . . . . . . . . . . . . . . . . 37
4.4.3 Phase Balancing Technique . . . . . . . . . . . . . . . . . . . . . . 38
4.4.4 Zero crossing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
4.4.5 Component Study . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
4.5 Active Load Balancer Based on Multilevel Converter-based Topology - A
Solution Based on the Primary Side of Load . . . . . . . . . . . . . . . . . 43
4.5.1 Three-level NPC inverter . . . . . . . . . . . . . . . . . . . . . . . 43
4.5.2 Control Strategy overview . . . . . . . . . . . . . . . . . . . . . . . 45
4.5.3 Component Study . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
4.6 Transformer Based Passive Solution - A Solution Based on the Primary
Side of Load . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
4.6.1 Basic Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
4.6.2 Compensation Circuit . . . . . . . . . . . . . . . . . . . . . . . . . 47
4.7 Summary of Brief Comparison of Solutions Presented . . . . . . . . . . . 48

5 Simulations 49
5.1 Three-Phase Four-wire . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
5.2 Three-Phase Three-wire delta . . . . . . . . . . . . . . . . . . . . . . . . . 53

6 Conclusion 56
6.1 Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57

Literature 58

Appendices 62

A Environmental analysis 62

B Open delta 63

C Delta modelling 65

D Cost estimation tool 67

© Uppsala University Mahir Mahdee


List of Tables V

List of Tables

Table 2.1: International standards values . . . . . . . . . . . . . . . . . . . . . . . 10

Table 4.1: Power load calculation on the DC side with different loading conditions
and the respective AC side current and Unbalances . . . . . . . . . . . 22
Table 4.2: Unbalance calculation of passive load share . . . . . . . . . . . . . . . . 24
Table 4.3: Unbalance calculation using DC side load currents . . . . . . . . . . . . 25
Table 4.4: List of common components required . . . . . . . . . . . . . . . . . . . 42
Table 4.5: List of common components required . . . . . . . . . . . . . . . . . . . 45
Table 4.6: Brief comparison of the solutions presented in this Chapter . . . . . . . 48

Table 5.1: Major parameters used in the simulation . . . . . . . . . . . . . . . . . 49


Table 5.2: Current and unbalance ratio of the five cases . . . . . . . . . . . . . . . 55

© Uppsala University Mahir Mahdee


List of Figures VI

List of Figures

Figure 2.1: wye-delta wiring structure . . . . . . . . . . . . . . . . . . . . . . . . 5


Figure 2.2: Phasor representation of a balanced and unbalanced system . . . . . 6
Figure 2.3: Imbalance current with its symmetrical components . . . . . . . . . 7
Figure 2.4: Example of uneven distribution of one-phase loads . . . . . . . . . . 11

Figure 3.1: Product from the two companies . . . . . . . . . . . . . . . . . . . . . 15


Figure 3.2: Working Principle of ALB . . . . . . . . . . . . . . . . . . . . . . . . . 16
Figure 3.3: Product EQUI8 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
Figure 3.4: EQUI8 installation near telecom base station . . . . . . . . . . . . . . 17
Figure 3.5: Ferroamp phase balancing concept . . . . . . . . . . . . . . . . . . . 18

Figure 4.1: Different solution concept . . . . . . . . . . . . . . . . . . . . . . . . . 19


Figure 4.2: Efficiency vs Load curve of the example systemt . . . . . . . . . . . . 21
Figure 4.3: A redundant system with 2 PSU´s per phase . . . . . . . . . . . . . . 26
Figure 4.4: A system with only one PSU for redundancy . . . . . . . . . . . . . . 27
Figure 4.5: Flowchart for switching algorithm . . . . . . . . . . . . . . . . . . . . 28
Figure 4.6: A typical composition of three-phase rectifier . . . . . . . . . . . . . 29
Figure 4.7: Vienna PFC rectifier topology . . . . . . . . . . . . . . . . . . . . . . . 30
Figure 4.8: Control of Vienna PFC showing the reference generation . . . . . . . 31
Figure 4.9: Detailed control layout of Vienna PFC rectifier . . . . . . . . . . . . . 31
Figure 4.10: Schematic of three phase full bridge PFC . . . . . . . . . . . . . . . . 32
Figure 4.11: Control principle of full bridge PFC . . . . . . . . . . . . . . . . . . . 33
Figure 4.12: Typical structure of LLC resonant converter . . . . . . . . . . . . . . 34
Figure 4.13: Interleaved LLC full-bridge resonant converter . . . . . . . . . . . . 35
Figure 4.14: Cost of a 6kW system with one and three phase rectifier . . . . . . . 36
Figure 4.15: Switch selector circuit . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
Figure 4.16: Flowchart of the balancing algorithm . . . . . . . . . . . . . . . . . . 41
Figure 4.17: Switching mechanism . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
Figure 4.18: A typical schematic of an active load balancer . . . . . . . . . . . . . 44
Figure 4.19: Zig-Zag Transformer . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
Figure 4.20: Zig-zag transformer compensation system . . . . . . . . . . . . . . . 47
Figure 4.21: Zig-zag transformer compensation system . . . . . . . . . . . . . . . 47

Figure 5.1: Simulation with balanced load . . . . . . . . . . . . . . . . . . . . . . 50


Figure 5.2: Simulation result of three phase under balanced loads . . . . . . . . 50
Figure 5.3: Simulation with unbalanced load . . . . . . . . . . . . . . . . . . . . 51

© Uppsala University Mahir Mahdee


List of Figures VII

Figure 5.4: Simulation result of three-phase under the unbalanced loading . . . 51


Figure 5.5: Simulation with a load failure from phase A . . . . . . . . . . . . . . 52
Figure 5.6: Simulation result of three phase under the unbalanced loading . . . 52
Figure 5.7: Simulation with unbalanced load . . . . . . . . . . . . . . . . . . . . 53
Figure 5.8: Simulation result of three-phase under the unbalanced loading . . . 53
Figure 5.9: Simulation with a load disconnected from phase 1-2 . . . . . . . . . 54
Figure 5.10: Simulation result of three-phase under a load failure . . . . . . . . . 54

© Uppsala University Mahir Mahdee


List of Acronyms VIII

List of Acronyms

Symbol Description

AC Alternating Current
AF Active Filter
DC Direct Current
ICT Information and Communications Technology
NPC Neutral Point Clamped
PCC Point of Common Coupling
PLC Programmable Logic Controller
PSU Power Supply Unit
PFC Power Factor Correction
PWM Pulse Width Modulation
THD Total Harmonic Distortion
SiC Silicon Carbide
VAC Voltage Alternating Current

© Uppsala University Mahir Mahdee


Introduction 1

1 Introduction

The utmost fact in modern electricity is the reliability and the quality of the supplied
electricity. With the gradual shift toward emission-free generation from renewable
sources and the advancement of other new technologies have led to new challenges in
the modern days’ electric power system where power quality is a concern [1]. The power
quality in terms of voltage and current unbalance is much severe at the utilization level,
which makes it essential to study at distribution level at the terminals of end-users.

There are several reasons for the power quality issues which includes several natural
ways but most important is due to customer equipment’s as they often draw non-
sinusoidal current and behave as nonlinear loads. The abundant of these loads, such
as single-phase loads connected to the three-phase network intensifies the amount of
unbalances in the three-phase system. Some countries often have specific standards and
requirements set for three-phase installation, and this will eventually lead to direct or
indirect penalties for three-phase electricity users if not followed and direct towards the
need for three-phase balancing [2].

1.1 Motivation
In the telecommunication industry, single-phase (one-phase) rectifiers (AC/DC power
supply units) are mostly used for base station loads or site power systems. These
rectifiers are generally connected line to neutral or line to line (phase to phase) depending
on if the distribution system is of delta type (three-phase three-wire) or wye type (three-
phase four-wire). In a three-phase installation, the uneven distribution of these single-
phase rectifiers or their irregular load current consumption from each phase(s) results in
having over/under-loaded phase(s). This may trigger in three-phase unbalance which
can be problematic for some grids in terms of transformer derating or additional losses.

At Ericsson AB, the Embedded power section at Lindholmen is one of the section that
develops power products intended for the use of Ericsson’s information and communi-
cations technology (ICT) product portfolio including 5G and provides power solutions
globally. These power solutions (having one-phase rectifiers in most cases) are the engine
to drive the telecom infrastructure where some market requests such installation to use a
three-phase AC system with sufficient phase balancing. To understand the initiation and
the importance of such requests, Ericsson is in the quest of finding requirements and

© Uppsala University Mahir Mahdee


Introduction 2

standards related to three-phase balancing and uncovering implementational solutions


according to the standards.

1.2 Objectives
The goal of the study is to understand the requirements and implications of three-phase
balancing from Ericsson’s perspective and find solutions to mitigate such issues of
three-phase unbalances. The goals are divided into several parts as follows:

1. To find different solutions for three-phase balancing with regards to telecom


inverters.

2. To investigate and find requirements for three-phase balancing concerning different


national/industry needs or standards.

3. With finding different solutions and requirements, the solutions shall then be
evaluated, categorized and judged in terms of:
• A independent balancing equipment in the primary side of loads.
• A fully integrated three-phase converter topology(s).
• and the control of three separate single-phase converters with a common
system DC bus.

The different solutions shall be flexible enough for Ericsson to choose between different
alternatives, so it will have the option of whether a solution will be good enough for a
specific market or to a specific customer.

1.3 Method
Firstly, the work begins with extensive research to build knowledge based on inter-
views with market participants for probable existing solutions for three-phase balancing.
For the market research, semi-structured interviews are conducted primarily through
skype interviews and in some case, communication took place via email as first-hand
communication.

Next, a study has been done on the requirements for the three-phase balancing in the per-
spective of different market and national needs. These requirements have been verified
by people from within the company, i.e. Ericsson AB. Finally with that, investigation for
different solutions concerning current existing systems has been conducted with detailed
analysis and propositions. A simulation has been performed in Matlab/SIMULINK
platform to check the validity for a passive solution.

© Uppsala University Mahir Mahdee


Introduction 3

1.4 Review of Power Demand in Telecom Sites


In order to cope with the increasing growth rate of internet users, the requirements in the
telecommunication industry shall increase consistently. The growing interest towards
modern and reliable service in the context of mobile telecommunication sector has led to
an expansion in the number of radio base stations installations globally. Besides, these
massive deployments of radio base station and towers secure a continuous operation to
continually guarantee the quality of the network service in any locations

Telecommunication sites have cell tower where radios and other telecom devices are
installed. According to [3], nearly 3% (equivalent to 600 TWh) of world’s electricity
is consumed by the ICT’s devices provoking about 2% of the global CO2 emissions;
However, 9% of this ICTs is due to the communication network radios. Among this
power consumption of radio communication networks, 10% of this energy is due to the
users terminal while 90% is consumed by the base stations [4].

At a typical cell tower in telecommunication site, the power demand is determined by


the number of base transceiver stations accommodated. The power demand ranges from
1KW to 8.5KW and can go as high as 12KW in case of the growing power demand for
5G applications. The individual power requirements by the tower devices are discussed
in the literature [5] [6]. Most of these sites with cell towers use three-phase electricity
where efficient current consumption from each phase is preferred.

Meanwhile, Ericsson has invested in a concept of an energy-optimized radio base station


site where a wind-powered tower tube has been developed with the help of Vertical
Wind AB and Uppsala University Sweden [7]. Ericsson’s tower tube power by a vertical
axis wind generator has antenna and base station integrated into the mast. The four-
blade turbine with a five-meter blade attached vertically towards the tower is a sign
of Ericsson’s commitment to sustainability and the environment. This solution shall
reduce the power consumption up to 40% while reducing the overall carbon footprint to
a minimum.

Most of the base station devices requires DC power to operate. This purpose is fulfilled
by the telecom rectifiers where they convert the AC input to a suitable DC output
needed by most of these devices. Besides, these rectifiers are not always distributed
evenly within the three-phase of the AC system. The consequence is an increased power
consumption from the phases followed by three-phase unbalance in current among the
three-phases. Thus the trade-off in the power consumption and phase unbalance needs
to be addressed.

© Uppsala University Mahir Mahdee


Introduction 4

1.5 Structure of the Thesis


• In chapter 2, theoretical background on three-phase system is presented followed
by the quantification and standards for three-phase balancing.

• Chapter 3 presents the overview of the existing products on the market for three-
phase balancing.

• Chapter 4 introduces the solutions for three-phase balancing with their working
principle, pros and cons, analysis etc., for each of the solutions.

• Chapter 5 shows the simulations and their results for the passive solution referred
to chapter 4.

• Chapter 6 is a wrap up with an outline of the findings, and future work that could
be done on the line of this work.

© Uppsala University Mahir Mahdee


Theory 5

2 Theory

This chapter provides a comprehensive review on the background of three-phase unbalance system
related concepts such as symmetrical components, quantification, etc. The chapter ends with
the discussion of unbalance assessments through certain standards followed by environmental
aspects .

2.1 Three Phase System


A three-phase system can be classified into a three-phase three-wire or three-phase
four-wire structure. The former can either be delta or star connected with star point
commonly grounded and the later usually uses star connected phase winding where the
fourth wire i.e. the neutral wire taken from the star point.

Figure 2.1: Wiring structure (a) Wye and (b) Delta

In a perfectly balanced system, the voltage and the current phases (a, b, c) are equal in
magnitude and displaced by 120◦ .

Va (t) = V”cos(wt) (2.1)

Vb (t) = V”cos(wt − 120) (2.2)

Vc (t) = V”cos(wt − 240) (2.3)

© Uppsala University Mahir Mahdee


Theory 6

Ia (t) = I”cos(wt − ∅) (2.4)

Ib (t) = I”cos(wt − 120 − ∅) (2.5)

Ic (t) = I”cos(wt − 240 − ∅) (2.6)

Where V" and I" are the peak value of voltage and current, ω = 2π50 is the angular fre-
quency and ∅ is the power factor angle. Any deviation from the two basic characteristic
stated above will result in an unbalance condition.

Figure 2.2: Phasor representation of a (a) balanced and (b) unbalanced system

However, system analysis using the above equations are not valid in an unbalance
condition and would be very complex in (abc) domain [8]. Symmetrical components
are therefore essential in such case to represent unbalance electrical quantities which is
discussed in the next section.

2.2 Symmetrical Components


According to Fortescies’s theorem, any unbalance three-phase system of vectors (voltage
or current representation) can be resolved into three balanced systems of vectors which
are termed symmetrical components. The key concept with symmetrical components is
to transform the phase components into a new set of symmetrical components. Hence,
with the symmetrical component, it is feasible to analyze unbalances in a three-phase
system in a simplified way. Once the sequence networks are obtained, it is possible to
transpose it to the three-phase network again. This technique can be applied for both
current and voltage in the network.

© Uppsala University Mahir Mahdee


Theory 7

The transformation from phase currents to sequence currents are shown in the equations
below.

     0
Ia 1 1 1 Ia
 Ib  = 1 a2 a   Ia1  (2.7)
     

Ic 1 a a2 Ia2

1
Ia0 = ( Ia + Ib + Ic ) (2.8)
3
1
Ia1 = ( Ia + a2 Ib + aIc ) (2.9)
3
1
Ia2 = ( Ia + aIb + a2 Ic ) (2.10)
3

Sequence components are defined as positive sequence component, negative sequence


component and zero sequence component. Figure 2.3 shows the resolving of unbalanced
phase currents into three set of balanced sequence currents.

Figure 2.3: Imbalance current with its symmetrical components

In a balanced three-phase system, the only sequence component that exists is the positive
sequence component. Lets consider equation 2.8, 2.9 and 2.10 to check if the statement is
true or not.

For a balanced system in the positive phase sequence Ib = a2 Ia , Ic = aIa where a = 1∠120.
Substituting the values of Ib , Ic and a in equation 2.8, 2.9 and 2.10 we get:

I0a = 31 ( Ia + a2 Ia + aIa ) = 13 Ia (1 + a2 + a) = 0

Ia1 = 13 ( Ia + a3 Ia + a3 Ia ) = 31 Ia (1 + a2 + a) = Ia

Ia2 = 31 ( Ia + a4 Ia + a2 Ia ) = 31 Ia (1 + a2 + a) = 0

© Uppsala University Mahir Mahdee


Theory 8

Thus, the only non-zero component of a balanced sets of currents or voltages in the posi-
tive phase sequence (i.e balanced 3-phase system) is the positive sequence component.
Negative and zero sequence components exists only in unbalance condition. However,
in a three-phase system without a neutral conductor there is no zero-sequence current.

2.3 Definitions and Quantifications


It is significant to define and quantify the meaning of asymmetry or unbalance before
moving into unbalanced system investigation. There are several definitions and quantifi-
cations presented in literatures depending on applications based on unbalanced studies.
From a network viewpoint, the IEEE defines and quantifies unbalance as a ratio between
the maximum deviation of phase voltage from average voltage to the average voltage [9].
Another slightly different definition from IEEE dictionary states " the difference between
the highest and the lowest RMS voltage, referred to the average voltage" [9][10]. The
National equipment manufacturer´s association NEMA instead uses the line voltage as
the unbalance measure parameter [11].The true definition of voltage unbalance by IEC
use symmetrical components as the ratio between negative sequence components (I2 , V2 )
to positive sequence components (I1 , V1 ) [11]. From a AC customer/user perspective, it
is not possible to influence the voltage deviation but is possible to adjust the phase/line
current. The equations 2.11 to 2.17 illustrates the definitions and quantifications for
individual organization.

© Uppsala University Mahir Mahdee


Theory 9

• IEEE definition 1 Std. 112 (1991):

Max |Vpa − Vpavg |, |Vpb − Vpavg |, |Vpc − Vpavg |


IEEE1 = ∗ 100 (2.11)
Vpavg

where Vpa , Vpb , Vpc are the phase voltages and Vpavg is the average of phase voltage
[(Vpa + Vpb + Vpc )/3].

• IEEE definition 2 Std. 936 (1987):

Max |VpaRMS , VpbRMS , VpcRMS | − Min|VpaRMS , VpbRMS , VpcRMS |


IEEE2 = (2.12)
Vpavg

• NEMA (National Electric Manufacturers Associations of the USA) Std. (1993):

Max |VLa − VLavg |, |VLb − VLavg |, |VLc − VLavg |


MEMA = ∗ 100 (2.13)
VLavg

where VLa , VLb , VLc are the line voltages and Vpavg is the average of Line voltage.

• CIGRE definition:
s p
1− 3 − 6β
%UnbalanceFactor = p (2.14)
1+ 3 − 6β

|VAB |4 + |VBC |4 + |VCA |4


β= (2.15)
(|VAB |2 + |VBC |2 + |VCA |2 )2
where VAB , VBC , VCA are the line voltages.

• IEC (International Electrotechnical Commission) definition:

V2
%VoltageUnbalanceFactor = ∗ 100 (2.16)
V1

I2
%CurrentUnbalanceFactor = ∗ 100 (2.17)
I1
where V2 , I2 are the negative sequence voltage and current components and V1 , I1
are the positive sequence voltage and current components.

Although, these definitions of unbalance could produce different values for a single case
scenario; it shall be noted that current unbalance formulations are simply derived by
replacing the voltage components with current.

© Uppsala University Mahir Mahdee


Theory 10

2.4 Standards
Unbalance is typically valuated based on a numerical concept called 95% percentile. It
implies that the allowed value should not exceed for 95% for a certain period of time
which is 10 minutes accumulated values for a week in most cases [12]. The European
standard EN 50160 gives the limits of voltage unbalance based on equation 2.16 as 2%
for low voltage and medium voltage systems [13]. In additions, it also states that in
certain areas with 1φ and 2φ users the unbalance could reach up-to 3% where mostly
these standards are from network perspective.

The American National Standard C84.1, developed by NEMA recommends that electrical
supply voltage shall be designed and operated within the maximum voltage limit of 3%
both from network and customer perspective [14]. This standard is expressed based on
equation 2.13. Table 2.1 illustrates the limitation values for different standards.

Table 2.1: International standards values.

Standards Voltage Unbalance Limits

ANSI 3.0%

ER P29 2.0%

IEC 6100-2-13 2.0-3.0%

EN 50160 2.0-3.0%

There are no standard limitations set for current unbalance; however, by the attention
of NEMA MG-1 standard, the maximum unbalanced limit advised for current is 30%
owing to 3% unbalance in voltage i.e. 10 times of the voltage unbalance limit [15].

2.4.1 Special Characteristics for Japan:

Japan has some unbalance load limitation requirements set for three-phase three-wire
according to the indoor wiring code (JEAC 8001-2011) [16]. It states that in the case of a
triple-phase three-wire system, the facility unbalance ratio (FUR) shall be 30% or less.

Definition of facility unbalance ratio (FUR):

Max ( phase load) − Min( phase load)


FUR = ∗ 100 (2.18)
Average Phase load

where average phase load = (Total load/3) and 3 is the total number of phases accord-
ingly in a three-phase system. In Japan, the three-phase voltage is 200V.

© Uppsala University Mahir Mahdee


Theory 11

2.5 Sources of Unbalance and its Consequence


In a practical network, utilities struggle to maintain three-phase balanced currents
and voltages at all nodes/junctions. Network apparatus such as transformers, rotating
machines, and distribution lines mainly cause voltage unbalance while asymmetric loads
cause substantial current unbalance. On the contrary, voltage unbalances responses to
current unbalance event in the case of a perfectly balanced supply owing to asymmetrical
loads. As the scope of the thesis is to consider the unbalances from the customer
perspective, i.e. load aspect, thus this research will only discuss the unbalance issues
with respect to customer loads.

In the telecommunication sector, the customer loads, i.e. base station loads or the
site power systems use three-phase power where mostly single-phase loads (example
single-phase power supply units - AC/DC converters supplying telecom load(s)) are
distributed within the three phases.

Figure 2.4: Example showing (a) unbalance current due to uneven distribution of
one-phase rectifiers and (b) a site power system

© Uppsala University Mahir Mahdee


Theory 12

The connection of this low voltage loads, i.e. single phase might not assure even or
balanced distribution between the supply system phases as can be seen from example
figure 2.4. However, even in the case of three power supply units (PSU) of the same
characteristic connected to the three phases of the three-phase supply, and each of this
PSU is powering a radio connected to each of them, the traffic on this radios might not be
same at a certain instant. Generally, these PSUs’ would have been connected to a single
DC bus and at the same time charging backup batteries. In either case, this will result in
an uneven current consumption from the three phases of the supply system, causing the
three-phase unbalance. This can be of concern when it comes to the grid that is sensitive
to phase-unbalance, especially in case of having delta or open-delta transformers (open
delta explained in appendix B). In Japan, the grid is delta/open-delta without neutral
which may be a reason to believe in making the special requirement stated in section
2.4.1 relevant to consider.

In a perfectly balanced system, the three-phase current adds up to zero, and the only
sequence current that exists in the system is the positive sequence current (Section
2.2). Whenever there is unbalance, it gives rise to negative and zero sequence current
components where this currents flows back to the source of generation (-ve sequence
current) or circulate within the delta connected windings in 3-phase 3-wire delta system
(zero-sequence current) or will flow through the neutral of 3-phase 4-wire wye system
(zero-sequence current). The current in the neutral leads to the increase of power
loss, which leads to an increase of additional expense because of the need for neutral
protection. The unbalance current contribute to voltage unbalance and will reduce the
operational capacity of the distribution transformer and overheating (increased losses
due to unbalance) of the transformer core will reduce the transformer lifetime. More to
it, it will initiate unwanted additional power loss and CO2 emission and shall facilitate
on increased carbon footprint [17].

2.6 Environmental Aspects


This thesis deals with the three phase-balancing in three-phase installations. Phase
balancing is more interesting from the grid point of view as it saves energy and money
in the long run. The way electricity is consumed also affects the environment directly on
indirectly.

Unbalance in customer three-phase installations can run down to the distribution side
after the customer metering box and reach the lines and the transformers which may
facilitate additional losses; these losses can then be translated to an additional CO2 emis-
sion intensity. This section will illustrate an example of how a three phase installation
could possibly induce additional losses and hence CO2 emission.

© Uppsala University Mahir Mahdee


Theory 13

Appendix A shows a detailed layout of a typical three-phase 10kW installation that


will be used here for the calculation purpose. The three-phase currents due to such
installations are as follows: Phase A (Ia ) = 18.16 A, Phase B (Ib ) = 18.16 A and Phase
C (Ic ) = 9.09 A. With these, additional losses in the low voltage distribution lines and
transformer can be calculated as follows [18] :

Additional loss of the low voltage lines

Additional line losses can be found by separately calculating the losses during balanced
and unbalanced condition. Let I p be the average load current

I p = ( Ia + Ib + Ic )/3 (2.19)

In balanced condition,I p = Ia = Ib = Ic . Now if the resistance of the line is set to be R L ,


the losses in the lines can be expressed as

∆PBalance = 3 ∗ I p2 ∗ R L (2.20)

In unbalanced condition, the losses in the lines can be expressed as

∆PUnbalance = ( Ia2 + Ib2 + Ic2 ) ∗ R L (2.21)

Now, degree of unbalance in each phase is expressed by

Ia − I p
βa = ∗ 100% (2.22)
Ip

Subtracting equation 2.20 from 2.21 and substituting the degree of unbalance in the
expression, the average additional power loss of the low voltage line can be found with
the equation:

∆Pline = (3 + β2a + β2b + β2c ) ∗ I p2 ∗ R L = 371.75W

where the R L is equivalent to 0.3Ω collected from e.ons network design manual [19].

Additional loss of the low voltage transformer

Additional copper loss of the distribution transformer can be calculated with the follow-
ing equation:

∆Pcu = 2( β2a + β2b + β2c − β a β b − β b β c − β c β a ) ∗ I p2 ∗ Rt = 262.47W

where Rt is is equivalent to 0.12Ω i.e, the internal resistance of the secondary side of a
common distribution transformer.

© Uppsala University Mahir Mahdee


Theory 14

Using the carbon footprint data from [20], the carbon footprint can be calculated as:

[(∆Pcu + ∆Pline )*8760/1000]kWh * 0.01189 kgCO2 e per kWh = 0.07 metric tons of CO2 e.

This is the additional emission due to three-phase unbalance from only one example
three-phase site installation in one year. With the presence of thousands of such installa-
tion connected to different power grids could add up to a great deal of additional CO2
emission which could be interesting if considered for future studies.

© Uppsala University Mahir Mahdee


Market Overview 15

3 Market Overview

This chapter is intended to compile information collected while reviewing the market for probable
solution for three-phase balancing. It includes the companies information with the products
overview along with their standards, voltage levels, cost etc. It creates an understanding of where
a three-phase balancing is desired.

3.1 Merus Power


Merus Power is a company headquartered in Finland that manufactures, designs and
markets a wide range of products for power quality solutions [21]. They are also
technology partner with a Malaysian based company named ELCO Power [22], and they
both have developed a product which they call Active Load Balancer - ALB. The device
acts as a balancing box and can mitigate the unbalance due to single-phase loading
(connected phase-phase or phase-neutral) in three-phase systems (3 wire or 4 wire).
Figure 3.1 below shows the product from both these companies.

Figure 3.1: Product from: (a) ELCO Power, (b) Merus Power [21][22]

Apart from mitigating unbalanced load issue, the product is aimed for applications
mainly for customers that have cases of harmonic distortion, poor power factor and
voltage variation or flickering in the three-phase system. The products are currently

© Uppsala University Mahir Mahdee


Market Overview 16

used in commercial buildings (residential buildings, hospitals, data centres and IT, etc.)
industry (chemical, textile, renewable sector, etc.). The product commercial demands it
could benefit customers in terms of energy savings, productivity, reliable plant operation,
etc. The product can be easily scalable owing to different needs. They can be found with
ratings from 50A-200A with nominal voltage input rating from 200VAC-690VAC.

The products have compliance with global power quality standards: IEEE 519, G5/4,
EN/IEC 61000 3-2, 3-4, 6-4, 6-2, EN 50178. The concept behind the product working
principle is shown in the graph below. The topology behind such a device shall be
discussed in details in chapter 4, section 4.5 with schematics. The cost of the device is
approximately 10,000-24,000 Swedish krona per the price quotation collected during the
market survey.

Figure 3.2: Working Principle of ALB through current compensation [22]

3.2 CME transformateur


Based in France, CME is a leading company in transformer manufacturing. CME has
been developing and manufacturing all variety of transformers, inductors and windings
for DC and AC for all power applications [23]. They have designed a fully passive
solution that improves and mitigates load balancing and voltage levels related issues.
They named the device EQUI8 three-phase network balancer. EQUI8 is intended to be
connected in parallel with the 3-phase 4-wire low voltage network where the voltage
variations are the greatest.

© Uppsala University Mahir Mahdee


Market Overview 17

Figure 3.3 shows a typical image of how the product looks like in real life. It basically
has three separate 1:1 special transformer with no other power electronics which makes
the product totally passive. The product can reduce the neutral current and also balance
the phase currents over the three lines to a certain extent. It complies with standards
like EN 61140, EN 50160, EN 60068-2-11, IK10.

Figure 3.3: Product EQUI8 [23]

CME has installed EQUI8 in many places, and they claim to have installed one of this
device near telecom base station to mitigate the issue of voltage unbalances, as shown
in figure 3.4. The cost of the product is about 27,000 Swedish Krona quoted by Oliver
Constant (export sales representative) during a skype conference dated 05.06.2020. As
per Oliver, the device seems to perform best in remote areas a little far from the grid
where there is unbalance issues in three-phase installations.

Figure 3.4: EQUI8 installation near telecom base station (courtesy of CME)

© Uppsala University Mahir Mahdee


Market Overview 18

3.3 Ferroamp
Ferroamp, a Swedish company founded in 2010, started with the idea of building a
bidirectional converter, named EnergyHub ACE (patented), with a unique function to
make a phase balancing in three-phase systems [24]. The idea here was to measure the
phase current (at the main fuse) continuously and when the current rises in one of the
phases due to higher power consumption, for example in electric car charging as shown
in figure 3.5, you can transfer energy from the other two phases. They then convert that
energy into direct current and add it into the first-mentioned phase so that they can
balance the currents and keep the total power through the main fuse lower. A lower
effect through the main fuse can lead to lower costs due to smaller dimensions and
electricity subscriptions.

Figure 3.5: Ferroamp phase-balancing concept with the EnergyHub [24]

The bidirectional converter uses the topology of non-isolated neutral point clamped
converter topology, and it only works in a three-phase system that has a neutral. As
per Wasif Adnan (Control System Expert in Ferroamp Energyhub System), the product
cannot work in delta systems where they require an additional delta-wye transformer,
but with that too, only active power control is possible and not phase balancing. En-
ergyHub aims for functionality not just to save electricity bill but it has the features of
connecting solar cells and batteries where their aim is interconnecting and controlling
DC micro-grid, said Niclas Wade Stig (VP Sales Marketing ) over the skype meeting
dated 22.07.2020. The product fulfils all standards demanded within Europe, and it has
good demand in the Nordics and the Netherlands. The cost of the device starts from
20,000 Swedish kronor and it goes up depending on the scalability.

© Uppsala University Mahir Mahdee


Solutions 19

4 Solutions

This chapter provides a detailed investigation of different approaches that could be used to mitigate
the unbalance issues from the loads perspective in three-phase installations. For each solution,
mathematical analysis, system schematics, control flow charts, etc. are discussed.

4.1 Introduction
For improving and mitigating three-phase unbalance, there are different solutions pro-
posed by different literature’s. In most cases, these solutions are designed for industries
with high voltage and high power applications. In the telecommunication industry,
the base station or site power system loads are generally of low voltage, but similar
solutions can be used with some slight system adjustments. Based on the scope of this
thesis, three different ways of mitigating unbalance concerning three-phase installation
in the telecom industry have been proposed (but not limited to), as shown in the figure
below.

Figure 4.1: Different solution based on (a) Logically controlling DC side of PSU’s, (b)
3-phase rectifier and (c) Balancing Box in the primary side of loads

Figure 4.1 (a) shows the idea of logically communicating and controlling the DC side of
the PSU’s load to control and reflect upon the unbalance on the three-phase AC side.
Figure 4.1 (b) shows that by using a three-phase rectifier, the uneven consumption of
phase current can be resolved as the three phases of the rectifier will be equally drawing
current from each phase.

© Uppsala University Mahir Mahdee


Solutions 20

Figure 4.1 (c) shows a balancing box that can be connected parallelly in the primary side
before the telecom 3 phase installations. Such a balancing device can either be active or
of passive type. The idea here is to compensate for the unbalances induced from the load
side such that looking from the AC supply side, the system will look perfectly balanced.
These concepts are discussed in details in the next following sections.

4.2 Controlling the DC side of power supply units through


load sharing
In the telecom industry, load sharing is used and is desirable for several reasons: (a) It
provides de-ration for individual power supply units (PSU’s), and helps to improve their
potential long-term reliability. (b) It boosts the total amount of load current available. (c)
It minimizes the potential for power bus glitches after a PSU fails.

The two most common load sharing approach is the active and passive load share. In a
power cabinet, the passive load share is performed internally by the PSU’s (the AC/DC
rectifiers) themselves whereas active load sharing is done by external control circuitry.
In order to load share, the PSU’s shall operate in parallel. Generally, with load sharing,
each PSU supplies current proportional to its rating. For example, three 25A PSU serving
a 50A load will divide the current among each other and each will supply 16.67A.

In this section, the idea is to make a relation between the DC side load sharing with
the AC side three-phase load balancing and check how load-sharing can impact on
three-phase balancing. Some example data has been collected from one of Ericsson
products to conduct the necessary analysis.

Active load sharing: In most base station telecom power cabinets, PSU’s are set to
calculate its load in percent of its maximum load and then report it back to the system
where it exists. For example, if a 2.5 kW PSU is considered and if the output load on that
PSU is 1250W, it reports a 50% load. Then, another PSU could have a 45% load where
the control system now adjusts the output voltage on each PSU with just a few millivolts
and then try to equalize the load balance between each PSU. For active load sharing in
the example case, each PSU can deviate from the average power load by ±3%. So, the
reported power load in percent shall not deviate more than 3% where one can have an
average system load of 50% and one PSU can have a 47% load then the system shall stop
with its regulation because it is within limits. If the PSU now goes to 45%, then it shall
adjust the voltage levels again to come back to the limit. This is the active load sharing.

For analysis purposes, it is considered that three PSU are connected parallelly and in
the input side, each of PSU is respectively connected to one of the phases of a three-
phase AC system. With this scenario, the total unbalance limit of the system as a whole

© Uppsala University Mahir Mahdee


Solutions 21

can be calculated using some standards. Japanese standard JEAC 8001-2011 has been
considered here which has the limit for three-phase unbalances current set to 30% with
regards to equation 2.18.

Load sharing (especially in the RBS 6k products) is considered on the DC side of PSU’s.
A relation has to be made between the input AC side and the output DC side in order
to realize the load current on the AC side and hence calculate the ultimate total system
unbalance ratio. Equation 4.1 below is used to determine the input current of the PSU’s.

Output Power
%E f f iciency = ∗ 100 (4.1)
Input Current ∗ Input Voltage

Where, input voltage varies slightly from 230V. The efficiencies are considered from
a standard curve shown below for different loading conditions. In case of delta grid
with loads connected phase-to-phase the current in the AC side of the PSU’s has to be
calculated with certain circuit analysis discussed in appendix C.

Figure 4.2: Efficiency vs Load curve of the example system (courtesy of Ericsson AB)

© Uppsala University Mahir Mahdee


Solutions 22

Table 4.1: Power load calculation on the DC side with different loading conditions and
the respective AC side current and Unbalances

© Uppsala University Mahir Mahdee


Solutions 23

Table 4.1 shows the calculations where each PSU deviates from the average power load
by 3%. The actual power has been calculated at different loading conditions, and their
respective efficiency has been determined from the graph in figure 4.2. The AC side
current has then been calculated using equation 4.1. and formulation in Appendix C.
From table 4.1, it can be observed that from 20 to 100 percent average loading conditions,
the percentage of unbalance is within the limit of 30%. At lighter loads below 20%, the
limitation standard for three-phase unbalance is not fulfilled.

To fulfill the requirement, either it has to be checked if at very low load, the three-phase
unbalance is a permitted (depending on the definition of the standards) or to comply
with the standard, the percentage of load share shall even go down below 3%; but in
reality, having a load share below 3% is not practical or is difficult.

Passive load sharing: The same system is considered here as what is considered in the
active load share case having three PSU’s each connected to one of the phases of the
three-phase AC system. Here, for the example system specification, each PSU shall not
deviate more than 50mV, so each and every PSU is calibrated at 50% load. The voltage
droop, i.e. the output voltage is dependent on the output current in the PSU where their
output voltage decreases 10mV per ampere current. If one of the PSU will experience
a little higher load, then the output voltage in that PSU will drop a little. Then it will
drop to that extent so that the other two PSU’s will try to help that PSU. The voltage
level in the common DC bus will then be a specific value, and then all the PSU will
approximately have the same output power as well. This is the passive load sharing.
The passive sharing is only active when the control system is not in place, which is
the normal case. The limit for passive load share is 10% or ± 5% for the example case
considered.

With the above information, and in the same way as the active load share case, the
percentage unbalance for passive load share is also calculated and presented in table 4.2.
Here only the AC side currents and percentage unbalance is tabulated. From table 4.2,
it can be observed that after an average loading condition of 35%, passive load share
cannot fulfil the requirement with the standard which suggests that active load share
can maintain a better three-phase balancing in comparison to passive load share.

© Uppsala University Mahir Mahdee


Solutions 24

Table 4.2: Unbalance calculation of passive load share

The recent power supply units aimed for high performance/growth markets have the
competence to communicate with each other through CAN bus. The Controller Area
Network (CAN) is a serial communication bus primarily developed for the automotive
industry and is currently adopted into numerous industrial applications owing to its
robustness in harsh electrical environment, reduced cable wirings, unique message han-
dling, and fault protection abilities. Such systems do not have any specific requirements
for active load sharing but can realize load sharing up to a certain extent within 3-5%.
The systems developed with this architecture where the PSU communicates with each
other and adjusts their output load currents are compared to the old systems where
output power was adjusted. With this information, the same calculation has been done
considering the same efficiencies where now each PSU shall deviate from the average
load current by ±3%.

© Uppsala University Mahir Mahdee


Solutions 25

Table 4.3: Unbalance calculation using DC side load currents

Table 4.3 shows the unbalance calculation results using DC side load current deviation
instead of power loads deviations. It can be observed that here the overall system unbalance
is limited to 6.5%, which is better in comparison with table 4.1 and 4.2 data. This reflects
to the understanding that with load sharing considering the output load current of
the PSU’s instead of output power load as a percentage offers better unbalance ratio.
However, for either case, it is unlikely to attain a good three-phase balance under low
load conditions.

The percentage of load share depends on the accuracy of the measurement circuit which
is not good at low load conditions. Generally, the accuracy measurement is achieved
through a resistor connected in series at the output. This resistor is used to estimate and
reflect the current; however, the value of resistor chosen is to optimize the reading in
a specific power range. The accuracy of the resistor shall also be considered. Usually,
two grade of accuracy are selected, 1% or 0.1%, and this also affects the reading. For
example, a very small resistor is needed to get a higher voltage for a higher load; but the
reading of this resistor will be very low voltage and very low load condition resulting in
not having a good accuracy at low load condition. Additionally, the voltage resolution
at the output will also effect how the output current is measured which also limits the

© Uppsala University Mahir Mahdee


Solutions 26

accuracy. Thus, there is always a trade-off between how accurately the output load
can be measured to the overall impact on the percentage of load sharing which further
reflects on the system unbalance ratio.

4.2.1 Redundancy:

Active and passive load sharing can provide three-phase balancing up to a certain
amount of loading condition. This issue becomes more complex when a customer
requires redundancy for the system. A redundant system is required to avoid the issues
of failure of power supply units from the AC supply phase. One way of solving for such
redundancy is described below.

The solution for three-phase balancing could be achieved (with redundancy) when
power supply units with 3*n multiple numbers of units are available. Here n should be
equal to 2 and above. So, when there is 6 PSU and are equally distributed among the
three phases, i.e. 2 PSU per phase, the system shall look something like shown below:

Figure 4.3: A redundant system with 2 PSU´s per phase

Now if one of the PSU in phase A fails, the PSU’s connected to phases B and C will
reduce their output power such that they balance out the current drawn in each AC
phase, i.e. by equalizing the load on each phase. Therefore, this system only relies on
active load share and passive load share is not possible. The algorithm here is defined
by software. It shall be noted that this system will be able to work only in PSU’s starting
from 6,9,12 (3*n where n> 2), i.e. the total number of PSU’s shall be such that it shall be
equally distributed between the three phases. However, a situation where high power is
not required makes such a system very expensive, where always one extra PSU’s per
phase is requisite for redundancy to solve the phase unbalance issue.

© Uppsala University Mahir Mahdee


Solutions 27

One hypothesis is proposed in this report where only one extra PSU can be used as a
redundant block which will reduce the cost of the overall system. The idea is to use
three PSU’s connected parallelly one per phase and a fourth PSU connected parallelly to
all three PSU’s which shall always remain idle. Whenever one of the PSU’s from one
phase fails, the fourth PSU shall come online and mitigate the issue of unbalance. Figure
4.4 shows a typical schematic of the proposed system.

Figure 4.4: A system with only one PSU for redundancy

Here, the fourth PSU can connect to any of the three phases at any instant when one of
the PSU fails. This is possible through a switch selector circuit that comprises of three
controllable switches. It can be implemented with TRIAC switches, or with the help of
PLC along with an ADC converter and the control algorithm shall be defined through
software.

The system will require two separate control circuits, one for tracking the current of the
output of each PSU’s and comparing them with a reference current to check the demand
of the DC load bus. Typically PSU has the highest efficiency when loaded between
30-80% (efficiency graph figure 4.2 ); thus the PSU initially has to set in such that they
will always try to work around 80% during highest load. Now whenever one of the PSU
from one phase shall fail, the other two PSU’s can quickly support the DC load for few
milliseconds until the fourth PSU comes online.

© Uppsala University Mahir Mahdee


Solutions 28

The second control is required for the fourth PSU to detect when one of the PSU fails
from a phase so that it can quickly connect to that phase from where the failed PSU is
disconnected. The flowchart in figure 4.5 shows the algorithm on how the system shall
react when a PSU disconnects.

Figure 4.5: Flowchart for switching algorithm

The system described above is just a hypothesis that could be considered for future
implementations. Therefore, this theory could be evaluated and verified to check if it
could provide the necessary response and reliability during a PSU failure with mitigation
for three-phase balancing according to standards.

© Uppsala University Mahir Mahdee


Solutions 29

4.3 Three Phase Rectifier


Traditional telecom rectifiers are typical of single-phase two-stage design. The reason
for the two stages is that there are specific industry standards that the rectifiers have
to comply with, and the most common is the IEC61000-3-2 standards. This standard
was introduced to regulate the harmonic current drawn from the ac mains to avoid the
harmonic pollution of the utility. These harmonic currents lower the efficiency of the
power drawn from the ac mains and further excite resonance as well as overload the
circuit wiring and transformers. Currently, three-phase rectifiers are given attention to
where three-phase ac systems are being used as they are less susceptible to imbalances
and harmonics.

This section will discuss the main three-phase rectifier topologies currently used in the
industry. A typical composition of three-phase rectifier is shown in figure 4.6 below:

Figure 4.6: A typical composition of three-phase rectifier

A three-phase rectifier most commonly has two stages. In the first stage, power factor
correction PFC is attained, including the conversion of AC energy to DC energy which
makes sure of good power quality. In the second stage, the DC-DC part converts the DC
energy to adjustable or fast-response DC energy. A number of three-phase telecom power
supply exists today with each having its own drawbacks and advantages. However, this
section will discuss the two most common PFC rectifiers (suggested by suppliers during
the research) followed by the commonly used DC-DC converters of reduced voltage
adjustments according to telcom need.

4.3.1 Three Phase PFC

The two most common PFC rectifier (front end converter) currently in use are the three-
phase three-level Vienna PFC and the three-phase two-level full-bridge PFC. There are
tradeoffs between the three-level and two-level PFC in terms of cost and complexity of
the structure, but both solutions exist today.

© Uppsala University Mahir Mahdee


Solutions 30

Vienna PFC: Professor Johann W. Kolar proposed the Vienna topology in 1994. The
Vienna PFC is widely in use in the industry because of its advantages towards excellent
THD and high efficiency when compared to traditional three-phase PFC topology [25].
Figure 2 shows the critical layout of Vienna PFC topology.

Figure 4.7: Vienna PFC rectifier topology

Figure 4.7 illustrates the schematic of a basic Vienna PFC rectifier. This structure can
further be divided into two main types: the three-switch structure on the right (b) uses
four diode bridge and one switch structure (per leg) while the six-switch Vienna uses the
two-switch structure as shown in the left (a). Both the structure has the same function
where (b) with three switches, all the semiconductors are only stressed with half of the
output DC voltage. In both structures, the DC output voltage can never be shortened
which provides pretty high reliability of both the structure. In the case of the structure
with (b), two of the diodes will always be conducting which will account for high
conduction losses. This issue can be declined by the six-switch structure to reduce the
number of conducting diodes.

Vienna PFC claims to have a power factor higher than 0.99 with THD less than 5%
[26][27]. A reference control circuit of such PFC is interpreted in figure 4.8. Here several
reference control parameters have been taken into consideration. The three-phase input
voltage and current signals, the two output voltage signals and the control signals for

© Uppsala University Mahir Mahdee


Solutions 31

the gates of the switches. The example of the control circuit is further expanded in figure
4.9 where the full overview of the control of the Vienna PFC is disclosed.

Figure 4.8: Control of Vienna PFC showing the reference generation

Figure 4.9: Detailed control layout of Vienna PFC rectifier

© Uppsala University Mahir Mahdee


Solutions 32

In figure 4.9, the different control stages are marked and is interpreted in the list below:

1. OP voltage loop
2. Current reference
3. IP current
4. PWM
5. Voltage direction
6. Generate PWM signal
7. Voltage balance loop
8. Inductor current limiter

Three-phase full-bridge PFC: Three-phase full bridge PFC has high efficiency with con-
trollable output voltage. Compared to Vienna PFC rectifier, the control of three-phase
full bridge PFC rectifier is much simple. A schematic of three-phase full bridge PFC is
shown in figure 4.10 and a control principle is outlined in figure 4.11.

Figure 4.10: Schematic of three phase full bridge PFC

© Uppsala University Mahir Mahdee


Solutions 33

Figure 4.11: Control principle of full bridge PFC

To summarize, three-level PFC rectifiers outstand two-level PFC by lower voltage stress
on semiconductor switches and smaller current ripples which in turn facilitates in having
smaller boost inductor in the AC side. Moreover, owing to lower voltage stress on the
switches, the switching losses are reduced with reduced filtering efforts. Three-phase
two-level full-bridge PFC requires MOSFETs of very high voltage rating which reflects
back in its overall cost. But, as the cost of SiC MOSFET is going down, this solution is
going to be more attractive.

© Uppsala University Mahir Mahdee


Solutions 34

4.3.2 DC-DC Conversion Stage

LLC resonant converters are most commonly used for the DC-DC conversion stage
in telecom rectifiers. In numerous cases, downsizing of required voltage with high
efficiency, high output power (current), and low EMI noise is preferred. LLC converter
provides both soft-switching which helps in reducing the switching losses with the
controlled frequency range and zero voltage/current switching for high efficiency and
high power density [28].

A typical structure of half-bridge LLC resonant converter is shown in figure 4.12, which
is basically composed of a square wave generator circuit, a resonant circuit, a rectifier
and an output filter capacitor. The converter structure can be a full bridge or half-bridge
type (shown in a and b) depending on the application and need. The half-bridge LLC
is commonly used in medium power application with the advantage of having less
number of switches. On the other hand, for the same power level, full-bridge LLC has
reduced primary side current resulting in having reduced conduction losses. With the
centre-taped transformer, only two rectifying diodes are necessary at the secondary side
rectifying stage or else a full-wave rectifier would be essential.

Figure 4.12: Typical structure of LLC resonant converter

© Uppsala University Mahir Mahdee


Solutions 35

Parallel-connected configuration of LLC converters is being used to increase the output


power rating. To reduce the output current and voltage ripple, an interleaved operation
between parallelly connected LLC is generally adopted. Figure 4.13. shows an overview
of an interleaved full-bridge LLC resonant converter.

Figure 4.13: Interleaved LLC full-bridge resonant converter

The capacitor Cb in the interleaved LLC full-bridge resonant converter shown in figure
4.13 is introduced to fast adjust the input voltage of two LLC and to make sure to even
out the primary average current of the two parallel LLC.

© Uppsala University Mahir Mahdee


Solutions 36

4.3.3 Cost

The cost of a single three-phase rectifier with same power level compared to a one-phase
rectifier is high because of more number of components, the complexity of its control
and higher thermal requirement due to many switching components. However, when
it comes to system-level, things could be different. To make a fair comparison, a 6kW
system is considered having three 2kW rectifiers. The same system can be implemented
with a 6kW three-phase rectifier, but now the former system having three one-phase
rectifiers shall have more number of components (according to the total number of
switches, diodes, capacitors, etc.). Thus the system with three one-phase rectifiers is
supposed to be more expensive; in reality, it is not entirely true.

One-phase rectifiers have been in the telcom industry for many years, and it still seems
to dominate the telecom market. Three-phase rectifier, on the other hand, is not so
popular and thus developing an entirely new three-phase rectifier could depend on
many different factors. Even if we assume that different costs such as installation cost,
handling, BOM cost, etc. is constant for both one and three-phase rectifiers, the high price
of three-phase rectifier shall come from the high development/R&D cost. During the
study, it has been observed that the current market and requests for three-phase rectifiers
are limited, and an option in Japan only. This shall make systems with three-phase
rectifiers (with phase balancing) relatively high. Figure 4.14 shows how the current
cost of a system with three-phase rectifier shall look like. A calculation tool has been
developed in excel to estimate the cost of rectifiers shown in appendix D.

Figure 4.14: Cost of a 6kW system with one and three phase rectifiers (courtesy of
Ericsson AB)

© Uppsala University Mahir Mahdee


Solutions 37

4.4 Automatic Phase Balancing Technique - Solution Based on


the Primary Side of Load
This topology is established on network reconfiguration, which is considered to be the
most cost-effective approach for low voltage system applications [17]. The topology is
based on automatic switches to optimize the orientation of load design in a three-phase
system. It works both in three-phase wye configuration with neutral or three-phase
delta configuration. Different optimization techniques for the control of these automatic
switches have been developed with the advent of artificial intelligence. Several of
these optimized techniques were implemented based on the principles of dynamic
programming, fuzzy-based algorithms, neural network, genetic algorithms etc. [29][30].
However, the control discussed for this proposed topology focused on simple single
loop iterating algorithm.

4.4.1 Advantages of Automatic System

1. By continuous detection of the individual single-phase loads, it provides a state-of-


the-art balanced operation of three-phase system.

2. Helps in avoiding issues with losses and unbalance.

3. Easy to install with no significant complexity.

4. Allows users to monitor individual single-phase loads/currents if necessary.

4.4.2 Mathematical Study

This topology will consider both a wye and a delta configuration for the algorithm
design. One of the methods to detect unbalance in a three-phase system is obtained by
calculating the phase and sequence current of a three-phase system and checking the
value of neutral current. Here, in the case of a balanced system, the neutral current In
is equal to zero. The relationship between phase current, sequence current and neutral
current is shown below in equation 4.2, 4.3, 4.4 [31].
 0   
Ia 1 1 1 Ia
 1  2
 Ia  =  1 a a   Ib  (4.2)
 
2
Ia 1 a a2 Ic

0
In = 3I ph (4.3)

a = e j2π/3 (4.4)

© Uppsala University Mahir Mahdee


Solutions 38

Minimizing the average unbalance (AU) per phase is another way of interpreting a
balanced system. Average unbalance can be calculated by equation 4.5

| I ph1 − I ph2 | + | I ph2 I ph3 | + | I ph3 I ph1 |


AU = (4.5)
3

4.4.3 Phase Balancing Technique

The balancing technique is performed via load optimization by shifting single-phase


loads from an overloaded phase to a lightly loaded phase through a switching system
connected to it. The proposed method takes the required measurements of the current
consumption of each device through metering sensors where the data is then transferred
to a central computer unit via a communication interface (Communication standard
RS485) to the balancing algorithm. Once the data is collected and calculations are made,
the set of loads are selected, and a control signal is sent by means of a transmission
system to the control switches connected to each load shown in the figure. 4.17. The load
switch selector is nothing but a combination of three controllable triacs, that controls
the loads to switch between three phases as shown in the figure. 4.17. The desired
configuration is accomplished such that only one of the switches can be ON at a time, i.e.
each load can only be connected to one phase. The constraints of the switches can be
mathematically represented as [29].

3
∑ SWk1i − 1 = 0 (4.6)
n =1

3
∑ SWk2i − 1 = 0 (4.7)
n =1

3
∑ SWk3i − 1 = 0 (4.8)
n =1

Where, k = number of loads


. I = number of phases
. and swkji can be equal to 1 (ON) or 0 (OFF) only.

© Uppsala University Mahir Mahdee


Solutions 39

Figure 4.15: Switch selector circuit

The suggested method is established on a single loop load exchange from one phase
to another. Let’s say the loads connected to each phase behaves dynamically, and thus
their variations are non-uniform and cannot be controlled. The currents are measured
according to user-defined time, and the unbalance percentage is evaluated by equation
4.9 defined by NEMA and IEEE standards [15]. This equation can be modified accord-
ingly owing to the requirement of different standardization, but the balancing procedure
and algorithm will remain the same.

(highest current − avg. current)


%Unbalance = ∗ 100 (4.9)
avg. current

After the calculation of unbalance, if the percentage does not surpass the defied stan-
dards, no actions will be taken. Otherwise, the original values are sent to the algorithm
responsible for finding a balanced solution. As shown in figure 4.18, the first step is
user-defined as mentioned earlier, according to the type of loads in the system. The
system is said to be balanced when the determined values of β, AU and In calculated
according to the equation in 4.3, 4.5, 4.10 are satisfied. If these conditions are not satisfied,
the algorithm keeps on iterating [31].

s
Imax − Imin
β= (4.10)
Imax + Imin

where Imax and Imin are the sum of maximum and minimum currents per phase respec-
tively.

© Uppsala University Mahir Mahdee


Solutions 40

A new parameter ∆Ie f f is calculated using equation 4.11, and condition 4.12 or 4.13
should be satisfied such that the load can be effectively transferred from a highly load
phase to the phase with less load [17]. Highly load phase is defined by the phase with the
maximum sum of currents. The difference between the load ‘k’ and ‘∆Ie f f ’ is calculated
as the algorithm iterates over Imax . The load ‘k’ is selected to move to the phase with a
minimum sum of current if the difference is less than ∆Ie f f . Next, the values of β, AU
and In are redetermined and checked. This is repeated up till condition 4.12 can no
longer be enforced on the group of loads with the highest phase current. The search for
the competent load should satisfy condition 4.13 in this case.

Imax − Imin
∆Ie f f = (4.11)
2

Ii e Imax , | Ii − ∆Ie f f | < ∆Ie f f (4.12)

Ii e Imax , Ij e Imin , |( Ii − Ij ) − ∆Ie f f | < ∆Ie f f (4.13)

According to the algorithm, two loops iterate over the phases with the highest and
lowest current Imax and Imin , respectively. The effective load is found through calculating
the difference between any two loads current and ∆Ie f f , and if it is less than ∆Ie f f ,
that load in the highest phase is allowed to be moved to the lightly loaded line. The
algorithm is designed to break in cases when a threshold is reached which is defined
by the user. In such a case, the output is conveyed by combinations of loads having the
lowest constraints. Here In is considered as the highest weight as discussed previously
that the presence of neutral current could be defined as a parameter for unbalance
measurement. Figure 4.18. describes the flow chart of the balancing algorithm. The
switching status sw is the yield of this algorithm.

© Uppsala University Mahir Mahdee


Solutions 41

Figure 4.16: Flowchart of the balancing algorithm

Switching from one phase to another is carried out at zero crossings so that there is
no effect to any loads during switching. Special circuits can be used with triac based
switching using ‘break before make’ strategy.

4.4.4 Zero crossing

The switching mechanism: As shown in figure 4.17, at the moment t1, the defined
reconfiguration algorithm elects the load ik to change from phase 1 to phase 3. At zero
crossing, the switch controller turns OFF the connection with phase 1 and connect to
phase 3 at t3, which is the following zero-crossing time. This reconfiguration causes a
dip that lasts for 17 milliseconds which is acceptable by the standards.

© Uppsala University Mahir Mahdee


Solutions 42

Figure 4.17: Switching mechanism [32]

4.4.5 Component Study

Some standard components needed to implement the system has been tabulated below.
The data and price of all the components were found from electronic suppliers Farnell
and Digikey accessed [06-08-2020]. It can be observed that many components are
required, and the total number of switches desired also depends on the total number of
loads. Thus, there are more losses in this system which will reduce the overall system
performance in terms of efficiency. Even though the system seems simple, the cost of the
controllers and current sensors shall make the system very expensive. One additional
disadvantage of such system is that it works best when there are a lot of single-phase
loads and will not be a suitable solution say if the total number of loads are very few.

Table 4.4: List of common components required (**Farnell & Digikey).


Component Part Number Rating Price/100 piece Unit Price - SEK
SEK
TRIAC BT138-800, 127 800V, 12A 6.30 11.28

TRIAC DRIVER MOC 3021 400V 2.67 7.13

CURRENT SENSOR/CT - - 247 257

CONTROLLERS - - - 1400-2500

RESISTORS ROX7J2K0 750V, 2Kohm 7.20 13.97

CAPACITOR ALC70C102FP600 600V, 1000 µF 381 -

© Uppsala University Mahir Mahdee


Solutions 43

4.5 Active Load Balancer Based on Multilevel Converter-based


Topology - A Solution Based on the Primary Side of Load
The topology is established on the principle of active power filter for either a three-phase
three-wire system or three-phase four-wire system. As defined earlier, single-phase loads
on a three-phase electric power system (connected phase-neutral or phase-phase) trigger
unbalanced load conditions which result in voltage/current unbalance in the system.
For a balanced system, the only current that exists is the positive sequence current as
defined in the literature of sequence components. Any unbalance situation from the
load side will trigger a negative sequence current to flow back to the source from the
load, causing a voltage unbalance in the system, which is unacceptable. In the case with
a three-phase system with neutral wire (three-phase four-wire wye system), this further
facilitates in excessive neutral current flowing back to the source/transformer, which is
undesirable. According to some literature, the neutral current shall be limited to 20% of
the normal phase magnitude current [33].

In this proposed approach, the balancing strategy is based on a three-phase active filter
upon neutral-point-clamped (NPC) topology. The topology is adopted to compensate
for the unbalance current issues by injecting an opposite phase current to loads negative
sequence current and hence balance the voltage/current in the phases.

Three-level NPC converters are used universally in industrial applications over other
converter topologies owing to their advantages with lower losses, reduced filter size,
lower electromagnetic interference, higher switching frequency which means reduced
weight and lower costs [34]. Figure 4.20 shows an NPC multilevel converter-based
topology connected parallelly before the load.

4.5.1 Three-level NPC inverter

Three-level NPC topology consists of 6 diodes and 12 IGBTs. figure 4.20 (b) shows a
detailed overview of the multilevel converter layout. The capacitors connected in series
are both charged with Vdc . Each leg of the converter contains four series IGBTs with two
diodes clamped in series. Here the function of the diode is to clamp the upper switches
at higher voltage potential to the zero dc-link points. The output of the three-level NPC
inverter shall be obtained through a particular switching pattern.

As shown in figure 4.20 (b), the circuit diagram gives an idea about the arrangement of
the power electronics requires for the three-level NPC topology. The diodes arranged
in such that it outputs different stages of voltage with respect to the neutral point N.
Voltages divided according to capacitor C1 and C2 connected in series with the neutral

© Uppsala University Mahir Mahdee


Solutions 44

point N between them. The voltage division obtained as + Vdc /2, 0 and - Vdc /2, so the
name is given as three-level topology [35].

Figure 4.18: (a). A typical schematic of an active load balancer, (b) a close overview of
the converter layout

© Uppsala University Mahir Mahdee


Solutions 45

In the topology, when the IGBTs Sa1 and Sa2 are on, + Vdc /2 is obtained. For - Vdc /2,
IGBTs Sa1’ and Sa2’ need to be turned on. Consequently, to received voltage level 0,
0 0
IGBTs Sa1 and Sa2 needs to be switched ON. The power circuit will hence provide the
three-level topology [35].

4.5.2 Control Strategy overview

The control of active filter-based topology is implemented in three stages. The first
stage constitutes accurate system information collection through the essence of current
and voltage signals utilizing current transformers (CT’s), power transformers (PT’s),
Hall-effect sensors or isolation amplifiers.

In the second stage, compensating commands are derived based on voltage or current
levels through different control methods and active filter configuration [36].

The third stage of the control consists of the gating signal generation for the solid-state
devices for the active filter (AF) using techniques such as hysteresis, PWM, fuzzy-logic
based control techniques etc. Most prominent methods of gating signal controls are listed
in some literature [34][37]. The control of the AF is realized using discrete analogue
and digital devices or advanced microelectronic devices such as DSP’s, single-chip
micro-controller etc. [38].

4.5.3 Component Study

Some common components required to implement the active load balancer has been
tabulated below. Active load balancer based on multilevel converter topology is efficient
and has high accuracy. The drawback of such an active device is the requirement for
many components which in turn requires the need for bigger cooling systems and
additional costs. This makes the system very expensive and bulky.

Table 4.5: List of common components required (**Farnell & Digikey).


Component Part Number Rating Price/100 piece - SEK Unit Price - SEK
POWER MOSFET R6020KNJTL 600V, 20A 22.58 30.46

DIODE RFN20TJ6SGC9 600V, 20A 13.83 25.45

CAPACITOR ALC70C102FP600 600V, 1000 µF 381 -

GATE DRIVER IR2110 - 34.28 23.36

CURRENT SENSOR - - 247 257

CONTROLLERS - - - 1400-2500

© Uppsala University Mahir Mahdee


Solutions 46

4.6 Transformer Based Passive Solution - A Solution Based on


the Primary Side of Load
Many literatures suggests that transformer based solution is a simple way to resolve
three-phase unbalance issue [39] [40] [41]. This is because no power electronics circuity
is required in such systems and this type of solutions are fully passive. In a three-phase
four-wire system, there are used to attenuate the excess neutral current caused by the
uneven distribution or uneven consumption of current over the three phases of the
supply system.

4.6.1 Basic Theory

Zig-Zag transformers is a special connection of three single-phase transformers windings


or a three-phase transformers windings. A basic schematic of a Zigzag winding is shown
in figure 4.21. The turn ratio of such a transformer is 1:1, and thus the current entering
the primary winding is equal to the current leaving the secondary winding. Then
three equation can be derived (eq 4.14-4.16) such that three-phase current entering into
the three transformers must be identical. Ideally, a zigzag transformer has substantial
impedance to positive and negative sequence current and very low impedance to zero-
sequence current. Hence, the current flowing through the zigzag transformer is only a
zero sequence component.

Figure 4.19: Zig-Zag Transformer [41]

iza (t) = izb (t) (4.14)

izb (t) = izc (t) (4.15)

izc (t) = iza (t) (4.16)

© Uppsala University Mahir Mahdee


Solutions 47

4.6.2 Compensation Circuit

For different wiring structure, compensation circuit diagram has been developed as
shown in the the following figure 4.22 and 4.23.

Figure 4.20: Zig-zag transformer compensation system

In figure 4.22, Zsn and ZLn are the impedance of neutral conductor between the trans-
former and the source and transformer and loads respectively. Zs is the impedance of
phase conductors between the transformer and the source. Transformers are also able
to mitigate the voltage unbalance in delta grids and to check if zig-zag transformer (as
they are represented simply as inductors) is able to do that, a compensation model (as in
figure 4.23) of such system is also analysed. Chapter 5 will show the simulation based
on figure 4.22 and 4.23.

Figure 4.21: Zig-zag transformer compensation system

© Uppsala University Mahir Mahdee


Solutions 48

4.7 Summary of Brief Comparison of Solutions Presented


This section provides a summary of a brief comparison of the different solutions pre-
sented in the chapter. It exhibits the unbalance ratio for each solution with their advan-
tages and considerations.

Table 4.6: Brief comparison of the solutions presented in this Chapter

© Uppsala University Mahir Mahdee


Simulations 49

5 Simulations

This chapter provides the simulation and their results based on the study in chapter 4. section
4.6.

The main purpose of this chapter is to verify the validity of transformer-based solution
for three-phase balancing in three-phase three-wire (delta) and three-phase four-wire
(wye) system. Simulations are carried out using Matlab/SIMULINK platform and
divided into 5 cases. All simulations are done with a three-phase grid of 10KV RMS
which is then stepped down to 200V, and the voltage is assumed to be purely sinusoidal
with a 120 degrees phase shift between the phases.

Practically system balance can be achieved by having an equal number of one-phase rec-
tifiers evenly distributed within the phases of a three-phase system. But with the failure
of one of such one-phase rectifier or the addition of one extra rectifier for redundancy
or load requirement can initiate three-phase unbalance in a three-phase installation.
The five different cases presented in the simulation are five different practical scenarios
which are evaluated with the passive transformer-based solution to check if the solution
can fulfil the three-phase unbalance standard of 30%.

The distribution grid delta-wye and delta-delta are modelled with grid parameters
collected from e.ons network design manual [19]. The distribution lines on the secondary
side of the transformer are short lines (under 100m to the PCC with width 16mm2 per
phase), only a resistive and an inductive part is modelled in this simulation. The loads
in the following simulations are the single-phase rectifier of 2KW each.

The chapter is arranged in two sections; section 5.1 includes the simulation and results
based on the wye distribution system with neutral. Section 5.1 comprises the simulation
and results for the delta distribution system. The table 5.1 shows the major parameters
used in the simulation.

Table 5.1: Major parameters used in the simulation.

Utility Voltage 200V, 50Hz

Grid Impedance 0.1Ω, 3.18mH

Line Impedance 0.01Ω, 0.1mH

Utility Neutral Inductance 0.1mH

© Uppsala University Mahir Mahdee


Simulations 50

5.1 Three-Phase Four-wire


Case 1:
In this case, the loads are balanced and are equally distributed among the three phases.

Figure 5.1: Simulation with balanced load

Results:
In balanced condition, the neutral current in the load side and source side is zero which
can be observed from figure 5.2 (c) and (d) respectively.

Figure 5.2: Simulation result of three-phase under balanced loads, (a) utility currents, (b)
load currents, (c) utility side neutral current, and (d) load side neutral current

© Uppsala University Mahir Mahdee


Simulations 51

Case 2:
In this case, to create an unbalance, an extra 2kw rectifier is connected to phase 3.

Figure 5.3: Simulation with overloaded third phase

Results:
From figure 5.4, it can be observed that with the third phase overloaded, the current
in phase 3 is higher (b). This is balanced to a extent in the utility side (a) by the
compensation current from the zigzag transformer (c). The neutral current on the load
side is almost 9 ampere (e) which is reduced to almost zero at the utility side neutral (d).

Figure 5.4: Simulation result of three-phase under unbalanced loading, (a) utility
currents, (b) load currents, (c) Zig-Zag transformer compensation currents,
(d) utility side neutral current, and (e) load side neutral current

© Uppsala University Mahir Mahdee


Simulations 52

Case 3:
In this case, to create an unbalance, the load from phase 1 is disconnected.

Figure 5.5: Simulation with a load failure from phase 1

Results:
With one load disconnected, the compensation current from the zigzag transformer was
able to balance the utility side phase currents to an extent Figure 5.6 (a) and the reduce
the utility side neutral current to almost zero figure 5.6 (d).

Figure 5.6: Simulation result of three phase with load in phase A failure, (a) utility
currents, (b) load currents, (c) Zig-Zag transformer compensation currents,
(d) utility side neutral current, and (e) load side neutral current

© Uppsala University Mahir Mahdee


Simulations 53

5.2 Three-Phase Three-wire delta


Case 4:
In this case, the loads are connected phase-to-phase and an unbalance was established
by connecting one extra 2kw load in between phase 1 and 3.

Figure 5.7: Simulation with overloaded phase

Results:
It can be observed from figure 5.8 (c) that the load current are not balanced, which is
then balanced to an extent in the utility side (a).

Figure 5.8: Simulation result of three-phase under overload phases, (a) utility currents,
(b) Zig-Zag transformer compensation currents, (c) Load Currents

© Uppsala University Mahir Mahdee


Simulations 54

Case 5:
In this case, the load between phase 1-2 is disconnected to create an unbalance in the
system.

Figure 5.9: Simulation with a load disconnected from phase 1-2

Results:
Figure 5.10 shows with one load disconnected, how the load currents behave in (c).
However, the current in the utility side is less unbalanced (a) and is compensated by the
generated current from the zig-zag transformer (b).

Figure 5.10: Simulation result of three-phase under a load failure, (a) utility currents, (b)
Zig-Zag transformer compensation currents, (c) Load Currents

© Uppsala University Mahir Mahdee


Simulations 55

Table 5.2 presents the values of the currents and the percentage unbalance ratio of the
five cases with and without transformer.

Table 5.2: Current and unbalance ratio of the five cases

© Uppsala University Mahir Mahdee


Conclusion 56

6 Conclusion

• One goal of the research was to find the requirements for three-phase balancing
within different countries. It was found that Japan is the only country that has a
standard for phase balancing in three-phase installations. This standard JEAC8001
of having phase unbalance up to 30% is expected to be satisfied in three-phase
installations.

• According to the goal of the study, several approaches to solving three-phase


unbalance have been presented. The solution using existing one-phase rectifiers
seems more reliable and cost-effective. One-phase rectifiers have been in the
industry for many years, and it still seems to dominate the telecom market. The
study showed that the three-phase unbalance mitigation with one-phase rectifiers
is possible with active load sharing. However, there is room for improvements
in the trade-off on how accurately the output load can be measured to the load
unbalance ratio on the AC side.

• From a technical viewpoint, three-phase rectifiers suggest a good solution in


phase unbalance mitigation issues for three-phase installations. However, they are
expensive for their high development cost due to their small market in telecom.

• The existing infrastructure of telecom relies on one phase rectifiers. In future,


solution for phase balancing in existing systems may also be needed. In that case,
replacing one rectifier unit is more economical than replacing the complete station
(with three-phase rectifiers).

• Solution with a separate balancing unit (box) before the load appears to be in-
teresting. These balancing units can be of an active or passive type. The active
technique uses power electronics, converters and complex control circuits with
good balancing indication.

• A passive solution has been simulated, and it shows they work well for the range
of operations according to the Japanese standard. They are as much reliable as the
active device, and they make the system work in limits of the norm; although they
do not completely eliminate the unbalance.

• Market research shows there are several companies that propose particular balanc-
ing equipment for phase balancing. However, such equipment has very high costs,
and this may add an extra burden for the users if required in small systems.

© Uppsala University Mahir Mahdee


Conclusion 57

• Phase balancing in three-phase installations can be beneficial in terms of reducing


the main fuse ratings and also avoiding overloading of the fuse. Lower fuse rating
shall make the contract with distribution company also lower where one can have
additional economic savings in the contract with the distribution company.

• Phase balancing is also interesting when it comes to saving of environment in


terms of additional CO2 emission and carbon footprint. Thus one who wishes
to have three-phase balancing can earn something on the greener side too, but
putting an additional balancing unit in place also relates to cost which also needs
to be considered.

6.1 Future Work


• Active load sharing among the PSUs are currently carried out determining the
power loads. As studies have shown that load current is a better way of realising the
percentage of active load share which further depends on the accuracy measure-
ment of the load currents, future work can be done in this area by trying to reduce
the percentage of active load sharing which shall eventually lead to better phase
balancing. Besides, the hypothesis that was suggested in the study for redundancy
could be studied and tested to verify if it is feasible to implement such solution.

• Studies conducted during the research were solely theoretical. It would be interest-
ing to do some practical testing on the available market products suggested in the
study to see if the products fit in the telecom company portfolio.

© Uppsala University Mahir Mahdee


Literature 58

Literature

[1] F. Kong, C. Dong, X. Liu, and H. Zeng, “Quantity versus quality: Optimal
harvesting wind power for the smart grid”, Proceedings of the IEEE, vol. 102, no. 11,
pp. 1762–1776, 2014.

[2] L. Neukirchner, P. Görbe, and A. Magyar, “Voltage unbalance reduction in the


domestic distribution area using asymmetric inverters”, Journal of Cleaner
Production, vol. 142, November 2016.

[3] I. Humar, X. Ge, L. Xiang, M. Jo, M. Chen, and J. Zhang, “Rethinking energy
efficiency models of cellular networks with embodied energy”, IEEE Network,
vol. 25, no. 2, pp. 40–49, 2011.

[4] Domestic ict and sustainable development: The same logic,


https://journals.openedition.org/terminal/2911, (Accessed on 10/13/2020).

[5] The true cost of providing energy for telecom towers in india | telecomlead,
https://www.telecomlead.com/whitepaper/the-true-cost-of-providing-
energy-for-telecom-towers-in-india-10200, (Accessed on 10/13/2020).

[6] W. Vereecken, W. V. Heddeghem, M. Deruyck, B. Puype, B. Lannoo, W. Joseph,


D. Colle, L. Martens, and P. Demeester, “Power consumption in
telecommunication networks: Overview and reduction strategies”, IEEE
Communications Magazine, vol. 49, no. 6, pp. 62–69, 2011.

[7] Wind power from uppsala operates ericsson’s mobile stations,


https://www.nyteknik.se/energi/vindkraft-fran-uppsala-driver-ericssons-
mobilstationer-6416138, (Accessed on 10/13/2020).

[8] Y. Hase, Handbook of Power System Engineering. Wiley, 2007.

[9] A. K. Singh, G. K. Singh, and R. Mitra, “Some observations on definitions of


voltage unbalance”, in 2007 39th North American Power Symposium, 2007,
pp. 473–479.

[10] “Ieee standard definitions for the measurement of electric power quantities under
sinusoidal, nonsinusoidal, balanced, or unbalanced conditions - redline”, IEEE Std
1459-2010 (Revision of IEEE Std 1459-2000) - Redline, pp. 1–52, 2010.

[11] “Definitions of voltage unbalance”, IEEE Power Engineering Review, vol. 21, no. 5,
pp. 49–51, 2001.

© Uppsala University Mahir Mahdee


Literature 59

[12] C. Masetti, “Revision of european standard en 50160 on power quality: Reasons


and solutions”, in Proceedings of 14th International Conference on Harmonics and
Quality of Power - ICHQP 2010, 2010, pp. 1–7.

[13] P. V. Santos Valois, C. M. Vieira Tahan, N. Kagan, and H. Arango, “Voltage


unbalance in low voltage distribution networks”, in 16th International Conference
and Exhibition on Electricity Distribution, 2001. Part 1: Contributions. CIRED. (IEE
Conf. Publ No. 482), vol. 2, 2001, 5 pp. vol.2-.

[14] Ansi c84.1-2016 - electric power systems and equipment - voltage ratings (60 hz),
https://webstore.ansi.org/Standards/NEMA/ANSIC842016?gclid=
CjwKCAjwqML6BRAHEiwAdquMnfKYaZg1O13nmwNEIUb8GUHmv-
og_URedVDM8aQvNnWHnL5QAu5X5RoC5qgQAvD_BwE, (Accessed on
09/03/2020).

[15] H. Arghavani and M. Peyravi, “Unbalanced current-based tariff”, CIRED - Open


Access Proceedings Journal, vol. 2017, no. 1, pp. 883–887, 2017.

[16] t. M. o. C. A. B. Japan Electrotechnical Association needs the equipment, “Internal


regulations (jeac 8001-2011) revised information (part3)”, 2012.

[17] M. Fahim, M. E. Hassan, and M. B. E. Najjar, “Single phase load balancing in a


three phase system at distribution and unit level”, in 2018 IEEE International
Conference on Industrial Technology (ICIT), 2018, pp. 1297–1301.

[18] J. Meng, W. Hao, W. Xu, M. Wang, and E. Li, “Research on accounting method of
carbon emission reduction for three-phase load balancing project of low-voltage
power grid”, in 2018 IEEE 4th Information Technology and Mechatronics Engineering
Conference (ITOEC), 2018, pp. 1134–1138.

[19] Network design manual - e.on uk,


https://www.yumpu.com/en/document/read/8953808/network-design-
manual-eon-uk, (Accessed on 07/23/2020).

[20] 2020_07_emissions_factors_sources_for_2020_electricity_v1_3.pdf,
https://www.carbonfootprint.com/docs/2020_07_emissions_factors_sources_
for_2020_electricity_v1_3.pdf, (Accessed on 07/28/2020).

[21] Merus power – winning business with power quality,


https://www.meruspower.fi/company/, (Accessed on 07/20/2020).

[22] Elco systems active harmonic filter, power electronics-based device,


http://www.elcopower.com.my/, (Accessed on 07/20/2020).

[23] Cme transformer | manufacturer of electrical transformers [t]: 04.74.35.55.11,


http://cmetransformateur.com/, (Accessed on 07/20/2020).

© Uppsala University Mahir Mahdee


Literature 60

[24] Ace – adaptive current equalization – ferroamp, https://ferroamp.com/en/ace-2/,


(Accessed on 07/23/2020).

[25] J. W. Kolar and F. C. Zach, “A novel three-phase utility interface minimizing line
current harmonics of high-power telecommunications rectifier modules”, IEEE
Transactions on Industrial Electronics, vol. 44, no. 4, pp. 456–467, 1997.

[26] H. Xu, W. Yao, and S. Shao, “Improved svpwm schemes for vienna rectifiers
without current distortion”, in 2017 IEEE Energy Conversion Congress and
Exposition (ECCE), 2017, pp. 3410–3414.

[27] J. Adhikari, P. IV, and S. K. Panda, “Reduction of input current harmonic


distortions and balancing of output voltages of the vienna rectifier under supply
voltage disturbances”, IEEE Transactions on Power Electronics, vol. 32, no. 7,
pp. 5802–5812, 2017.

[28] C. Liang, T. Liang, K. Chen, and H. Liang, “Design and implementation of


interleaved llc half-bridge resonant converters”, in 2014 International Power
Electronics and Application Conference and Exposition, 2014, pp. 1509–1512.

[29] M. W. Siti, D. V. Nicolae, A. A. Jimoh, and A. Ukil, “Reconfiguration and load


balancing in the lv and mv distribution networks for optimal performance”, IEEE
Transactions on Power Delivery, vol. 22, no. 4, pp. 2534–2540, 2007.

[30] A. B. Knolseisen, J. Coelho, S. F. Mayerle, F. J. S. Pimentel, and


R. H. Guembarovski, “A model for the improvement of load balancing in
secondary networks”, in 2003 IEEE Bologna Power Tech Conference Proceedings,,
vol. 3, 2003, 7 pp. Vol.3-.

[31] C. L. Fortescue, “Method of symmetrical co-ordinates applied to the solution of


polyphase networks”, Transactions of the American Institute of Electrical Engineers,
vol. XXXVII, no. 2, pp. 1027–1140, 1918.

[32] D. V. Nicolae, M. W. Siti, and A. A. Jimoh, “Lv self balancing distribution network
reconfiguration for minimum losses”, in 2009 IEEE Bucharest PowerTech, 2009,
pp. 1–6.

[33] A. Negi, S. Surendhar, S. R. Kumar, and P. Raja, “Assessment and comparison of


different neutral current compensation techniques in three-phase four-wire
distribution system”, in 2012 3rd IEEE International Symposium on Power Electronics
for Distributed Generation Systems (PEDG), 2012, pp. 423–430.

[34] R. Teichmann and S. Bernet, “A comparison of three-level converters versus


two-level converters for low-voltage drives, traction, and utility applications”,
IEEE Transactions on Industry Applications, vol. 41, no. 3, pp. 855–865, 2005.

© Uppsala University Mahir Mahdee


Literature 61

[35] H. Dehghani Tafti, A. I. Maswood, A. Ukil, O. H. P. Gabriel, and L. Ziyou, “Npc


photovoltaic grid-connected inverter using proportional-resonant controller”, in
2014 IEEE PES Asia-Pacific Power and Energy Engineering Conference (APPEEC),
2014, pp. 1–6.

[36] B. Singh, K. Al-Haddad, and A. Chandra, “A review of active filters for power
quality improvement”, IEEE Transactions on Industrial Electronics, vol. 46, no. 5,
pp. 960–971, 1999.

[37] Ning-Yi Dai, Man-Chung Wong, and Ying-Duo Han, “Application of a three-level
npc inverter as a three-phase four-wire power quality compensator by generalized
3dsvm”, IEEE Transactions on Power Electronics, vol. 21, no. 2, pp. 440–449, 2006.

[38] H. Obara, K. Wada, K. Miyazaki, M. Takamiya, and T. Sakurai, “Active gate


control in half-bridge inverters using programmable gate driver ics to improve
both surge voltage and switching loss”, in 2017 IEEE Applied Power Electronics
Conference and Exposition (APEC), 2017, pp. 1153–1159.

[39] D. Sreenivasarao, P. Agarwal, and B. Das, “Neutral current compensation in


three-phase, four-wire systems: A review”, Electric Power Systems Research, vol. 86,
pp. 170–180, 2012.

[40] B. Liu, X. Shi, Y. Li, F. F. Wang, and L. M. Tolbert, “A line impedance conditioner
for saturation mitigation of zigzag transformer in hybrid ac/dc transmission
system considering line unbalances”, IEEE Transactions on Power Electronics,
vol. 32, no. 7, pp. 5070–5086, 2017.

[41] Hurng-Liahng Jou, Jinn-Chang Wu, Kuen-Der Wu, Wen-Jung Chiang, and
Yi-Hsun Chen, “Analysis of zig-zag transformer applying in the three-phase
four-wire distribution power system”, IEEE Transactions on Power Delivery, vol. 20,
no. 2, pp. 1168–1173, 2005.

© Uppsala University Mahir Mahdee


Appendices 62

Appendices

A Environmental analysis

Figure A.1 shows a 10KW three-phase installation and its matlab model to find the three
phase AC current is shown in figure A.2.

Figure A.1: 10kW 3-phase installation

Figure A.2: Matlab model of the 10KW system

© Uppsala University Mahir Mahdee


Appendices 63

B Open delta

The figure below considers an open delta transformer connected to a three-phase source
where the secondary of the transformer is open-circuited. The voltage across AB, BC,
CA shall be equal to that of the source voltage (generator). This implies that even with
the absence of the third winding, there is no impact on the phase-phase voltage on the
open delta system formed by two single-phase transformers. The phase relationship
in case of three-phase open delta system is precisely the same as what is expected in a
conventional three-phase delta system. The three-phase voltage and phase relationship
shall be induced from the primary to the secondary where the secondary terminals shall
now have three-phase voltage.

Figure B.1: open delta connection

To understand and mathematically prove open delta voltage, KVL is applied across the
secondary windings:

We know in a delta system,


Vab + Vbc + Vca = 0 (B.1)

Vca = −[Vab + Vbc ] (B.2)

Vca = −V∠0 − −V∠ − 120 (B.3)

Vca = −V − 0.5V − j0.866V (B.4)

Vca = (0.5 − j0.866)V (B.5)

Vca = V∠120 (B.6)

© Uppsala University Mahir Mahdee


Appendices 64

It can be realized from the above equation B.6 that the voltage across the leg ’ac’ with
the missing transformer is V∠120 which is indeed what one would have expected from
a conventional delta system.

Because in open delta we have a missing winding ‘ac’, the current in the two single-phase
transformer shall be equal to the line currents with 30 degrees leading and lagging. This
means the transformer will have an internal power factor of 86.6 percent (cos 30 = 86.6)
even with loads having unity power factor; that is, one transformer has a power factor of
cos(30+φ) and the other cos(30 -φ). Thus, the voltage regulation in both the transformer
is non-identical. This is a major reason the secondary voltage of open delta transformer
tends to be unbalanced and sensitive to three-phase unbalance.

© Uppsala University Mahir Mahdee


Appendices 65

C Delta modelling

A delta transformer/source with three loads connected phase to phase has been mod-
elled (figure C.1) where Vab = 0, Vbc = V∠120 and Vca = V∠ − 120. R a , Rb and Rc are
resistance to represent the voltage drops in the lines.

Figure C.1: Delta connected model

The loads in figure B.1 are represented as current source in figure B.2 where all the
directions of the currents are shown.

© Uppsala University Mahir Mahdee


Appendices 66

Figure C.2: Delta model with current flow directions

By applying Kirchhoff’s Voltage Law (KVL) and Kirchhoff’s Current Law (KCL), few
sets of equations are obtained as follows:

Ia = Iab − Ica (C.1)

Ib = Ibc − Iab (C.2)

Ic = Ica − Ibc (C.3)

VAB = Vab − Ia R a + Ib Rb (C.4)

VBC = Vbc − Ib Rb + Ic Rc (C.5)

VCA = Vca − Ic Rc + Ia R a (C.6)

Solving the equations above, the desired line current and phase current can be deter-
mined based on the standards. Here, the loads are telecom PSU’s which are defined as
current sources.

© Uppsala University Mahir Mahdee


Appendices 67

D Cost estimation tool

Figure D.1: Cost Estimation Tool

© Uppsala University Mahir Mahdee

You might also like