Documente Academic
Documente Profesional
Documente Cultură
Doctor of Philosophy
from
University of Wollongong
by
Engineering
March 2009
Dedicated to my parents...
Acknowledgements
visor, Associate Professor Sarath Perera of the University of Wollongong (UoW), for
support given throughout the study period in many ways. Your dedication, patience,
knowledge and experience could not have been surpassed. I admire your guidance
Thanks to my co-supervisor, Professor Danny Sutanto of the UoW, for the assis-
tance provided. I would also like to offer many appreciations to Dr. Duane Robinson
of Beca, Australia for proofreading this thesis. To Mr. Robert Koch of Eskom
Holdings Limited, South Africa and Dr. Zia Emin of National Grid Electricity Trans-
mission, United Kingdom go many thanks for their insightful technical contributions
and helpful attitude. LATEX assistance received from Dr. Timothy Browne, previ-
ously with the Integral Energy Power Quality and Reliability Centre (IEPQRC) at
Funding for this project was provided by SP AusNet, Victoria and the IEPQRC.
I am grateful to Mr. Dhammika Adihetti, Mr. Shiva Bellur and Mr. Sanath Peiris of
SP AusNet for arraigning this. Many thanks to Mr. Jeff Sultana, Mr. Shem Cardosa
and Mr. Mahinda Wickramasuriya of SP AusNet for the support given in collecting
Thanks to Dr. Vic Smith and Sean Elphick of the IEPQRC who have graciously
Engineering (SECTE) at the UoW, Tracey O’Keefe and Maree Burnett who are
former members of the SECTE staff, and Esperanza Riley of the IEPQRC for solving
Very special thanks go to my friend Dr. Sankika Tennakoon, previously with the
IEPQRC, for being generously supportive especially during hard times along the way.
My heartiest gratitude goes to my parents Mithrananda and Manike for all encour-
agements, guidance and sacrifices made on behalf of me to come this far. Finally, my
thanks go to the rest of my family and friends particularly Pinky, Dimuthu, Radley,
requirements for the award of Doctor of Philosophy, in the School of Electrical, Com-
own work unless otherwise referenced or acknowledged. This manuscript has not been
Prabodha Paranavithana
v
Abstract
tions on engineering practices governing voltage unbalance that would facilitate the
Technical Report IEC/TR 61000-3-13 which provides guiding principles for coordi-
nating voltage unbalance between various voltage levels of a power system through
the allocation of emission limits to installations. Although the IEC report is based
on widely accepted basic concepts and principles, it requires refinements and original
developments in relation to some of the key aspects. This thesis primarily focuses on
Similar to the counterpart IEC guidelines for harmonics (IEC 61000-3-6) and
flicker (IEC 61000-3-7) allocation, IEC/TR 61000-3-13 also apportions the global
apparent power, and the total available apparent power of the system seen at the
can arise as a result of both load and system (essentially lines) asymmetries, IEC/TR
tioned allowance. This factor Kue represents the fraction of the global emission
accounts for voltage unbalance which arises as a result of line asymmetries. Although
vi
vii
IEC/TR 61000-3-13 recommends system operators to assess the factors Kue and
K 0 ue for prevailing system conditions, a systematic method for its evaluation is not
provided other than a rudimentary direction. This thesis initially examines, employ-
ing radial systems, the influence of line asymmetries on the global emission levels
in medium voltage (MV) and high voltage (HV) power systems in the presence of
the factor K 0 ue is seen to be dependant not only on line parameters as evident from
IEC/TR 61000-3-13, but also on the downstream load composition. In essence, the
global emission levels in HV power systems is seen to arise as a result of both the local
and interconnected networks, for the assessment of the global emission in MV and
tions, quantitative measures of its propagation from higher voltage to lower voltage
levels in terms of transfer coefficients, and from one busbar to other neighbouring bus-
less than unity for industrial load bases containing large proportions of mains con-
nected three-phase induction motors, and a value of unity for passive loads in general.
Upon detailed examination, it is noted that a transfer coefficient > 1 can arise in the
influences exhibited by various load types under unbalanced supply conditions on the
The IEC allocation policy with regard to harmonics and flicker has been found not
of the set planning levels. This thesis reports that the above is an issue with voltage
referred to as ‘constraint bus voltage’ (CBV) method which closely aligns with the
IEC approach has been suggested for harmonics and flicker. The work presented in
this thesis extends the suggested CBV method to voltage unbalance allocation adding
appropriate revisions to address the additional aspect of the emission which arises as
networks already experiencing excessive voltage unbalance levels, the initial develop-
ment of insights into the influences made by various sources of unbalance is required.
case, deterministic studies are carried out in a systematic manner considering each of
the asymmetrical elements. Approaches for studying the voltage unbalance behaviour
established. These are employed to identify the most favourable line transposition
options for the study system. Further, this knowledge that facilitates the identifi-
developing techniques to assess the compliance with emission limits, which is another
As an essential tool for carrying out the studies, an unbalanced load flow program
based on the phase coordinate reference frame incorporating the component level load
ix
x
klv fraction of LV loads supplied by any higher voltage (MV, HV) busbar
km ratio between the rated motor load (in MVA) and the total
load (in MVA) supplied by an LV system
kmmv ratio between the rated motor load (in MVA) and the total
load (in MVA) supplied by an MV system
kpq ratio between the constant power load (in MVA) and the total
load (in MVA) supplied by an LV system
kpqmv ratio between the constant power load (in MVA) and the total
load (in MVA) supplied by an MV system
ks ratio between the positive and negative sequence impedances of the
aggregated motor load supplied by an LV system
ksc−s ratio between the short-circuit capacity (in MVA) at any busbar S
and the total load (in MVA) supplied by the busbar S
kz ratio between the constant impedance load (in MVA) and the total
load (in MVA) supplied by an LV system
kzmv ratio between the constant impedance load (in MVA) and the total
load (in MVA) supplied by an MV system
Kues:x fraction of the busbar emission allowance at any busbar x of any
sub-system S that can be allocated to installations
K 0 ues:x fraction of the busbar emission allowance at any busbar x of any
sub-system S that accounts for the emission arising as a result of
system inherent asymmetries
LF load flow
LV low voltage
ml refers to a MV-LV coupling transformer
MV medium voltage
NECA National Electricity Code Australia
NEMA National Equipment Manufacturer’s Association
PCC point of common coupling
PQ refers to a constant power load
PS refers to a passive load
xi
lines
Ug/s:x global emission arising as a result of system inherent asymmetries
at any busbar x of any sub-system S [VUF]
Uj/s:x emission level caused by any source of unbalance j
at any busbar x of any sub-system S [VUF]
result
Us:x resultant emission level at any busbar x of any sub-system S [VUF]
Ux voltage unbalance at any busbar x [VUF]
UIE International Union for Electricity Applications
US represents any upstream system of any sub-system S
(US = EHV, HV, MV)
[V ] matrix of nodal voltages
Vλ:x λ (= 0, +, −) sequence component of Vx [V]
Vλ:s−us λ (= 0, +, −) sequence voltage, referred to US, at any busbar S [V]
Vn−s nominal line-line voltage of any sub-system S [V]
Vx voltage at any busbar x [V]
lines
V−:g/s:x global negative sequence voltage arising as a result of line
asymmetries at any busbar x of any sub-system S [V]
V−:Ui /x negative sequence voltage at any busbar x caused by
the voltage unbalance Ui that exists at any other busbar i
V Rt voltage regulation of any line t
V Rtd voltage regulation of any line td
VUF voltage unbalance factor [%]
[Y ] matrix of nodal admittances
Yλ∆:xy λ − ∆ (λ, ∆ = 0, +, −) sequence coupling admittance
component of Yxy [S]
Yxy nodal admittance between any busbar x and any
other busbar y [S]
Y−−:x−im downstream negative sequence admittance seen at any
busbar x taking only induction motors into account [S]
Y−+:x downstream negative-positive sequence coupling
admittance seen at any busbar x [S]
xiii
October 2008.
5. Robert Koch, Alex Baith, Sarath Perera, and Prabodha Paranavithana. Volt-
age Unbalance Emission Limits for Installations - General Guidelines and Sys-
September-October 2008.
xiv
xv
2009.
July 2009.
Delivery.
1 Introduction 1
1.1 Statement of the Problem . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Research Objectives and Methodologies . . . . . . . . . . . . . . . . . 4
1.3 Outline of the Thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2 Literature Review 10
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.2 Definition of Voltage Unbalance . . . . . . . . . . . . . . . . . . . . . 11
2.3 Sources of Voltage Unbalance . . . . . . . . . . . . . . . . . . . . . . 13
2.4 Effects of Voltage Unbalance . . . . . . . . . . . . . . . . . . . . . . . 14
2.5 Mitigation Techniques of Voltage Unbalance . . . . . . . . . . . . . . 17
2.6 Measurement and Indices of Voltage Unbalance . . . . . . . . . . . . 18
2.7 Limits of Voltage Unbalance . . . . . . . . . . . . . . . . . . . . . . . 21
2.7.1 Compatibility Levels . . . . . . . . . . . . . . . . . . . . . . . 21
2.7.2 Voltage Characteristics . . . . . . . . . . . . . . . . . . . . . . 22
2.7.3 Planning Levels . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.7.4 Customer Emission Limits . . . . . . . . . . . . . . . . . . . . 26
2.8 Guiding Principles of IEC/TR 61000-3-13 [1] for Voltage Unbalance
Emission Allocation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
2.8.1 Basic Concepts Used in IEC/TR 61000-3-13 . . . . . . . . . . 28
2.8.2 Emission Limits: Stages 1, 2 and 3 . . . . . . . . . . . . . . . 30
2.8.3 Development of Stage 2 Emission Limits . . . . . . . . . . . . 31
2.8.4 Voltage Unbalance Transfer Coefficients . . . . . . . . . . . . 39
2.8.5 Factor K 0 ue . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
2.9 A Revised Harmonics/Flicker Allocation Technique Based on the IEC
Guidelines - A Preamble to Voltage Unbalance Allocation . . . . . . . 43
2.10 Chapter Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
xvii
xviii
Appendices
H Test Case Description of the Radial HV-MV-LV System (Fig. 4.2) 224
xxi
xxii
5.1 Variation of Tmv−lv with ksc−lv obtained for constant power loads using
unbalanced load flow analysis . . . . . . . . . . . . . . . . . . . . . . 95
5.2 Radial system considered for the illustration of transfer coefficients . 97
5.3 Variation of Tmv−lv with ksc−lv for constant current loads: I - 0.99
lagging pf, II - 0.9 lagging pf . . . . . . . . . . . . . . . . . . . . . . . 104
5.4 Variation of Tmv−lv with ksc−lv for constant power loads: I - 0.99 lagging
pf, II - 0.9 lagging pf . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
5.5 Variation of Tmv−lv with ksc−lv for induction motor loads with ks = 6.7
and pf = 0.9 lagging . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
5.6 Variation of Tmv−lv with ksc−lv : I - for a load base dominated by in-
duction motors, II - for a load base dominated by passive elements . . 108
5.7 Variation of Tmv−lv with km for ksc−lv ≈ 25 and ksc−lv ≈ 10: I - for
load mixes of Z and IM loads, II - for load mixes of P Q and IM loads 109
5.8 Variation of Tmv−lv with km established using the IEC method, (5.19),
(5.20) and unbalanced load flow analysis . . . . . . . . . . . . . . . . 110
5.9 Variation of Thv−mv with klv for ksc−mv = 12 (loads are supplied directly
at the MV busbar): I - for load mixes of Z and IM loads, II - for load
mixes of PQ and IM loads . . . . . . . . . . . . . . . . . . . . . . . . 115
5.10 Variation of Thv−mv with klv for ksc−mv = 4 (loads are supplied directly
at the MV busbar): I - for load mixes of Z and IM loads, II - for load
mixes of PQ and IM loads . . . . . . . . . . . . . . . . . . . . . . . . 116
5.11 Variation of Thv−mv with klv (LV loads are supplied through MV lines):
I - for ksc−mv = 12, II - for ksc−mv = 4 . . . . . . . . . . . . . . . . . . 116
5.12 Radial MV-LV system (reproduction of Fig. 3.2) . . . . . . . . . . . . 117
5.13 Variation of ksend−rec with km for the cases where klv = 1, klv = 0.5
and klv = 0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
5.14 Interconnected sub-system S (reproduction of Fig. 2.6) . . . . . . . . 122
5.15 System representation of any busbar x of the MV system shown in
Fig. 5.14 (reproduction of Fig. 3.6) . . . . . . . . . . . . . . . . . . . 124
5.16 Three-bus MV test system considered for applying the proposed method-
ology (reproduction of Fig. 3.7) . . . . . . . . . . . . . . . . . . . . . 127
5.17 Variations of k1−2 and k1−3 with km:2 for the three-bus MV test system 127
5.18 IEEE 14-bus test system (reproduction of Fig. 4.9) . . . . . . . . . . 128
5.19 Influence coefficients k4−x (x = 1 − 14, x 6= 4) for the IEEE 14-bus
test system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
6.1 Three-bus HV test system considered for examining the IEC/TR 61000-
3-13 approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
6.2 A comparison of the influence coefficients for the test system derived
using the proposed method: (5.37), and unbalanced load flow analysis 135
6.3 A comparison of the K 0 uex factors for the test system derived using
the proposed method: (4.16), and unbalanced load flow analysis . . . 138
xxiii
6.1 Influence coefficients for the test system shown in Fig. 6.1 . . . . . . 135
6.2 Shv:x , Shv:x−total and Ug/hv:x for the test system shown in Fig. 6.1 . . . 135
6.3 lines
Ug/hv:x , K 0 uex and Kuex for Case 2 of the test system shown in Fig. 6.1137
6.4 Ehv:x according to IEC/TR 61000-3-13 for the test system shown in
Fig. 6.1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
reult
6.5 Ug/hv:x arising as a result of the IEC/TR 61000-3-13 allocation proce-
dure for the test system shown in Fig. 6.1 . . . . . . . . . . . . . . . 139
6.6 Values of the RHS of (6.8) in relation to the test system shown in Fig.
6.1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143
6.7 ka for the test system shown in Fig. 6.1 . . . . . . . . . . . . . . . . . 143
6.8 Kuex and Ehv:x according to the revised allocation method for the test
system shown in Fig. 6.1 . . . . . . . . . . . . . . . . . . . . . . . . . 143
reult
6.9 Ug/hv:x arising as a result of the revised allocation procedure for the
test system shown in Fig. 6.1 . . . . . . . . . . . . . . . . . . . . . . 144
xxv
xxvi
Introduction
unbalanced installations and system inherent asymmetries can cause damage to, and
degradation and maloperation of, customer and utility equipment. Despite the exis-
tence of voltage unbalance regulatory codes, some network service providers are facing
difficulties in complying with stipulated levels. This emphasises the need for recom-
The IEC, one of the world’s leading organisation for standardisation on power
quality, has recently released the Technical Report IEC/TR 61000-3-13 [1] which
provides guiding principles for coordinating voltage unbalance between various voltage
The philosophy of this voltage unbalance allocation process is similar to that of the
counterpart IEC approaches to harmonics (IEC 61000-3-6 [2]) and flicker (IEC 61000-
3-7 [3]) allocation. The absorption capacity or the allowed global emission of a sub-
1
In the context of the thesis, this is limited to negative sequence unbalance.
1
2
system of a power system is established such that the total emission level derived using
the general summation law, taking the upstream contribution in terms of a transfer
coefficient into account, at any point is maintained at or below the set planning
level. The global emission allowance of the sub-system is allocated to its busbars in
proportion to the ratio between the total apparent power to be supplied by the busbar
under evaluation, and the total available apparent power of the sub-system as seen
at the busbar. Voltage unbalance contributions from neighbouring busbars are taken
into account using influence coefficients in determining the total available apparent
power of the sub-system as seen at the busbar. This busbar emission allowance is then
apparent power, and the total apparent power supplied by the busbar.
In the case of voltage unbalance, the global emission at a busbar generally arises
not only as a result of unbalanced installations but also as a result of system inherent
installations as in the case of harmonics and flicker can lead to exceedances of the
referred to as ‘Kue’ to the apportioned allowance. This factor Kue represents the
fraction of the emission allowance that can be allocated to customers, whereas the
factor K 0 ue (= 1 − Kue) accounts for the emission which arises as a result of system
Kue and K 0 ue for prevailing system conditions in their specific networks. However,
a systematic method for its evaluation is not provided other than a rudimentary
The Technical Report IEC/TR 61000-3-13 gives a method for estimating the MV
to LV transfer coefficient considering the system and load characteristics and the
downstream load composition. This suggests a value less than unity for the trans-
3
fer coefficient in the presence of industrial load bases containing large proportions
ematically trivial for constant impedance loads, its validity has not been cautiously
examined in relation to constant current and constant power loads which may exhibit
ods for assessing the HV to MV and EHV to HV transfer coefficients and influence
The IEC allocation policy with regard to harmonics and flicker has been found not
ensure non-exceedance of the set planning levels [4, 5] 2 . Overcoming this problem, an
method which closely aligns with the IEC approach has been suggested for harmon-
ics and flicker [4, 5]. Being based on a common philosophy, the above problem is
of IEC/TR 61000-3-13 which also involves an additional aspect, i.e. the emission aris-
ing due to system inherent asymmetries. Extension of the CBV method to voltage
networks already experiencing excessive voltage unbalance levels, the initial develop-
ment of insights into the influences made by various sources of unbalance is required.
tion is not a usual practice, the emission which arises as a result of system inherent
asymmetries would not allow an equitable share of busbar emission allowances to in-
2
References [4, 5] are the only sources which provide evidence in support of this statement.
4
stallations. The present knowledge which describes the voltage unbalance behaviour
els arising as a result of line asymmetries themselves are excessive, making decisions
challenge.
The aim of the work presented in this thesis is to make contributions for further
of voltage unbalance.
ipated problem of exceedance of the set planning levels, and proposing appro-
priate revisions.
on the global voltage unbalance levels in MV and HV power systems and the de-
carried out in relation to radial power systems. A basis towards the development
which arises as a result of line asymmetries is established through the extension of the
nodal equations [I] = [Y ][V ] to the sequence domain. This basis is integrated with the
outcomes obtained from the preliminary studies for ascertaining the methodologies.
theoretical bases which describe the behaviour of these load types under unbalanced
supply conditions. Employing these, the impact of a load base which consists of
employing a simple three-bus test system. Consideration to cases both with and with-
out the inclusion of the influence of system inherent asymmetries is given. Firstly, the
emission limits to installations are calculated using the prescribed approach together
with some of the methodologies proposed in this thesis. Secondly, the resulting bus-
bar voltage unbalance levels are established using the general summation law when
all installations inject their allocated limits, and examined against the set planning
accomplished by introducing its principles while addressing the emission which arises
To develop theoretical bases which provide an insight into the problem of voltage
by unbalanced load flow analysis are carried out employing a 66kV sub-transmission
system that is known to experience excessive voltage unbalance levels. Using a new
concept termed ‘voltage unbalance emission vector’ which is derived based on IEC/TR
61000-3-13, the behaviour of each of the lines treating as standalone lines and also as
elements in the interconnected system, and of each of the loads operating in the inter-
are established. Employing the linearity of negative sequence variables, these global
emission vectors of individual unbalanced sources are added forming a basis which
entire system.
fects and mitigation techniques of voltage unbalance is given. Various standards and
voltage unbalance are reviewed. The key section of this chapter describes concepts,
principles and related aspects prescribed in IEC/TR 61000-3-13 establishing the back-
grounds for Chapters 3 - 6. The last section discusses fundamental deficiencies of,
and suggested revisions to, the IEC allocation policy with regard to harmonics and
7
Chapter 3 addresses the aspect of the global voltage unbalance emission which
carried out employing a simple radial network on the influence of line asymmetries on
the global emission levels and its dependency on various load types/bases is presented,
the evaluation of this global emission at nodal level is proposed. Results established
using this method for a three-bus test system are compared with those obtained using
on the influence of the downstream MV line asymmetries on the global emission levels.
Dependency of the global voltage unbalance levels on the local HV line asymmetries
61000-3-13 are emphasised. A methodology for evaluating the global emission which
arises as a result of both the local and downstream line asymmetries is proposed.
Results established using this method for a three-bus test system and also for the
IEEE 14-bus test system are compared with those obtained using unbalanced load
flow analysis.
the propagation from upstream higher voltage to downstream lower voltage levels
which include different passive components and three-phase induction motors, gen-
eralised expressions for their estimation are proposed. Ranges of variation of these
transfer coefficients are demonstrated. The accuracy of this new method for esti-
given in IEC/TR 61000-3-13. Secondly, the propagation from one busbar to other
proposed. Results established using this method for a three-bus test system and also
for the IEEE 14-bus test system are compared with those obtained using unbalanced
principles employing a three-bus test system. The calculation procedure of the emis-
sion limits to installations using the prescribed formulae together with some of the
levels when all installations inject their allocated limits are derived, and examined.
Secondly, the principles of the suggested CBV allocation policy are introduced to
voltage unbalance ensuring a robust allocation. These new allocation principles are
Chapter 7 establishes theoretical bases for studying the problem of voltage un-
load asymmetries are separately described. Outcomes from these studies are presented
proposed approach is applied for assessing the study system on major contributors
to voltage unbalance levels, and line transposition options which effectively correct
the asymmetrical network. These assessments are validated employing the results
Finally, Chapter 8 summarises the major outcomes of the work presented in the
Literature Review
2.1 Introduction
and a critical discussion on IEC/TR 61000-3-13 on which the thesis is primarily based.
used in different standards and documents for its quantification is given in Section 2.2.
Sections 2.3 - 2.5 cover sources, effects and mitigation techniques of voltage unbalance
respectively as reported in the literature. The widely used IEC 61000-4-30 and other
of voltage unbalance are examined in Section 2.6. Various categories of voltage unbal-
ance limits: compatibility levels, voltage characteristics, planning levels and customer
emission limits are discussed, and a review on limiting values is given in Section 2.7.
The key section of this chapter, Section 2.8, describes concepts, principles and related
3 - 6. Section 2.9 discusses fundamental deficiencies of, and suggested revisions to, the
IEC allocation policy with regard to harmonics and flicker forming the background
10
11
The Technical Report IEC/TR 61000-3-13 [1] defines voltage unbalance as a condition
phase voltages and/or the associated phase angles of separation are not equal. This
is a steady-state condition, and hence short-term unbalance which can occur during
events such as unsymmetrical faults does not fall under the definition [1]. Voltage
unbalance can exist in two forms in three-phase power systems: zero and negative
sequence unbalance. Where there exists a path for the flow of zero sequence currents
such as in grounded-neutral systems, the presence of zero sequence voltage can become
an issue [6, 7] especially when the coupling transformer allows zero sequence currents
to flow from higher voltage to lower voltage systems and vice-versa. Zero sequence
neutral systems and dual-phase installations, and also as it can be controlled through
system design and maintenance [1, 8]. As the negative sequence voltage propagates
through all power system components similar to the positive sequence voltage, it is
The modulus of the ratio of the fundamental negative sequence (V− ) to posi-
tive sequence (V+ ) voltage components, which is known as ‘voltage unbalance factor’
negative sequence voltage unbalance. This seems to be in agreement with most other
V−
V U F = (2.1)
V+
Reproducing from IEC 61000-4-30 [19], IEC/TR 61000-3-13 gives a practical method
for establishing the VUF using the three fundamental line-line rms voltage magni-
tudes as: s √
1 − 3 − 6
V UF = √ (2.2)
1 + 3 − 6
where,
|Vab |4 +|Vbc |4 +|Vca |4
= (|Vab |2 +|Vbc |2 +|Vca |2 )2
Alternative methods for the quantification of voltage unbalance are given by the
trical and Electronics Engineries (IEEE)4 . The NEMA definition which is known
as ‘line voltage unbalance rate’ (LVUR), and the IEEE definition which is known as
‘phase voltage unbalance rate’ (PVUR) that exists in two different forms (P V U R1 and
P V U R2 ) are given by (2.3), (2.4) and (2.5) respectively. However, the recent IEEE
power quality monitoring standard IEEE 1159 [23] lists both the P V U R1 and the VUF.
Although angle unbalance is excluded, the LVUR which does not take the presence
of zero sequence voltage into account is similar to the VUF or the true value for
more realistic levels of voltage unbalance [22, 24]. However, the PVUR which is
influenced by the presence of zero sequence voltage deviates significantly away from
the true value in the presence of zero sequence voltage even at lower levels of voltage
unbalance [24]. Among the two IEEE definitions, the P V U R1 is reasonably close to
V−
Although the absolute value of the ratio V+
or the VUF is the parameter in general
use, it is worthwhile noting that voltage unbalance is also associated with a phase
angle. One may, in the same way, define this phase angle as the angle between the
fundamental negative and positive sequence voltage components [25]. This concept
Voltage unbalance is caused mainly by the uneven distribution and/or the uneven
connection of single-phase and dual-phase loads6 across the three phases and the op-
currents or negative sequence currents into the system. Unequal mutual impedances
31, 32] transmission and distribution [33, 34] overhead lines, which lead to unbal-
anced voltage drops across the three phases, is also a well known source of voltage
posed overhead transmission lines [30, 35] and asymmetrical transformer banks [36]
in particular open-wye open-delta transformer banks [37] have also been reported as
tion motors is well documented [38, 39, 40]. When an induction motor is exposed to
unbalanced voltages, the negative sequence voltage component produces an air gap
flux that rotates against the rotor which is forced by the positive sequence torque,
thus generating an unwanted reverse torque. This results in a reduction of the net
motor torque and speed, in addition to torque and speed pulsations and increased
motor vibration and noise. Further, due to the relatively small negative sequence mo-
the supply voltage unbalance [20] causing increased motor losses and heating. On the
whole, the motor efficiency and lifetime (primarily as a consequence of the prolonged
overheating) will be reduced. To be able to deal with this extra heating, the motor
the International Union for Electricity Applications (UIE) [16]8 , an induction motor
by Fig. 2.1.
43] and arc furnaces [42] produce uncharacteristic triplen harmonics in addition to
the characteristic harmonics in the input current in the presence of supply voltage
8
The derating curve given in [16] is preferred, as it uses the VUF in quantifying voltage unbalance,
in comparison to other recommendations such as given in the standards NEMA MG1 [20] (which
uses the LVUR) and AS 1359.31 [41]/IEC Report 892 (which uses the P V U R1 ).
9
e.g. adjustable speed drives.
15
Please see print copy for image
unbalance. Significant third harmonic currents can increase harmonics and resonance
problems in power systems, and require large filter ratings. As the degree of voltage
unbalanced and changes from a double pulse waveform to a single pulse waveform as
a result of the asymmetric conduction of the diodes. This results in excessive currents
in one or two of the phases10 , which can lead to the tripping of overload protection
circuits, under voltage and increased ripple on the dc-link, and decreased lifetime of
rectifier front ends generate a second order harmonic component on the dc-link when
they are exposed to supply voltage unbalance [45]. This results in increased ripple
on the dc-link affecting the life and size of the dc-link capacitor. Further, this second
order harmonic component reflects in the input current and also in the inverter output
10
Measurements taken on an adjustable speed drive system has shown 50% over-current for a
supply voltage unbalance where the highest voltage magnitude was 3.6% higher than the lowest
voltage magnitude [44].
16
respectively.
The impact of some fault conditions (other than the traditionally studied three-
phase fault) on the transient stability of synchronous generators has been seen to
be more severe in the presence of voltage unbalance [46, 47]. This indicates the
requirement of advanced algorithms and computer programs for power system stabil-
ity studies.
tribution overhead lines and cables and transformers can also be affected by voltage
unbalance, which is intensified by the fact that a small degree of unbalance in phase
motors in the presence of negative sequence current resulting in excess machine losses
and heating and possible hazards to structural components [48]. According to the
the rated current and the ratio of the negative sequence current component and the
rated current does not exceed a value between 5% and 10% depending on the type
of construction, the method of cooling and the machine capacity. Flow of negative
sequence currents in overhead lines, cables and transformers increases power losses
lowering their capacity [50, 51]. From a more theoretical point of view, current un-
balance affects the definitions and the measurement techniques of apparent power
and power factor [52, 53] influencing the aspects of the power system economics. In
addition, current unbalance has been seen to result in a degraded power factor [53].
11
These sub-harmonics will be replaced by a dc component when the inverter output frequency is
equal to twice the system frequency.
12
This is based on IEC 34-1: Rotating electrical machines - part 1 - rating and performance.
17
across the three phases. This has been facilitated through the development of tech-
niques for the manual or automatic reconfiguration of distribution systems [54, 55], the
phase rearrangement between distribution transformers and primary feeders [55, 56]
arrangements which provide an equilateral triangular spacing between the three phase
conductors (for single-circuit lines) nullify the emission which arises as a result of
line asymmetries [57]. However, as these ideal conditions can rarely be achieved in
more appropriate design options in terms of the tower configuration [30, 32], and the
Furthermore, the increase of the fault level at the point of common coupling (PCC)
of inherently unbalanced large loads (e.g. traction loads and arc furnaces) can make
In cases where excessive voltage unbalance levels are unavoidable, special balanc-
ing equipment can be installed at the utility and/or plant level. Power electronic
the injection or the absorbtion of reactive power to or from the system have been pro-
posed with the development of suitable control algorithms for dynamically correcting
13
e.g. passive static var compensators (SVC) [59], active static synchronous compensators (STAT-
COM) [60] and distribution STATCOMs (DSTATCOM) [61].
18
correction through the injection of a compensating voltage signal in series with the
supply have also been reported as a means for mitigating voltage unbalance. Com-
prehensive techniques such as unified power quality conditioners (UPQC) [66, 67]
and hybrid active and passive filters [68] which are capable of compensating various
power quality disturbances simultaneously have been further advanced also to handle
voltage unbalance.
using an equivalent balanced section and a two-phase section has been employed in
reducing the influence of large unbalanced loads (e.g. traction loads) by the use of
equipment. Site indices are used to provide a statistical description of the disturbance
at a particular site. System indices which are derived using site indices of various
over a part of the power system. This section discusses measurement procedures and
The widely accepted standard IEC 61000-4-30 [19] for the measurement of power
quality disturbances prescribes also the voltage unbalance measurement and evalu-
ation procedure for instruments with Class A17 performance. Measurement of the
fundamental component of the three line-line rms voltages over 10-cycle and 12-cycle
intervals for 50Hz and 60Hz systems respectively is specified. A minimum mea-
surement period of one week is recommended. Aggregated values are obtained over
standard time intervals of 3-second, 10-minute and 2-hour18 . The method of quan-
tification is as per (2.2). For instruments with Class B19 performance, the above
the presence of other background disturbances in the input electrical signal to the
measuring instrument. Thus, IEC 61000-4-30 defines limits for the uncertainty of
a specified range of variation. For the measurement of voltage unbalance, when other
disturbances exist in the input signal fulfil the requirements given in Table 2.1, except
for voltage unbalance levels in the range of 1% to 5% of the declared input voltage
(Udin ), an instrument shall present an uncertainty less than ±0.15%. For instruments
Derivation of site indices using a high percentile (e.g. 95%, 99%) of the aggre-
and documents [9, 10, 12, 19]. The 95% percentile of the 10-minute aggregated val-
for most power quality disturbances including voltage unbalance. This is the only
17
That is, precise measurements such as for the verification of compliance with standards.
18
A 2-hour value is obtained by combining twelve number of 10-minute values.
19
That is, less precise measurements such as for statistical surveys.
20
index used in the European standard EN 50160 [9]. IEC 61000-4-30 proposes a num-
ber of voltage unbalance site indices for contractual applications including the 95%
percentile of the 10-minute and 2-hour aggregated values over a week. The issue
of voltage unbalance indices has also been addressed by the CIGRE/CIRED Joint
Working Group C4.07 [12], and the above index together with the 95% percentile of
the 3-second values over a day has been recently recommended. The South African
standard NRS 048-2 [10] uses the highest of the 10-minute values over a week in
addition to the above index as preliminary site indices. Site indices over long mea-
surement periods are typically calculated as the highest of the daily or weekly indices
Among various methods of calculating system indices [25], the choice of a high
percentile of site indices is seen to be popular [12, 70, 71]. The IEC electromagnetic
compatibility standards IEC 61000-2-2 [70] and IEC 61000-2-12 [71] use the 95%
percentile of the 95% site indices as the system index. In addition to the above system
index, [25] recommends the highest of the 95% or 99% site indices as a system index
Working Group C4.07 [12] for a system index in assessing a set voltage unbalance
limit as a low percentage of sites (e.g. 1% and 5%) that exceeds the limit.
21
the emission of the disturbance must be limited to a level which is tolerable by the
balance between the emission and the immunity is the compatibility level. Equip-
the compatibility level, and utilities are required to maintain the disturbance at or
below the compatibility level. Due to the stochastic nature of the power quality phe-
locations with a set limit is not sensible. Thus, the compatibility levels are generally
set allowing a small exceeding probability (e.g. 5%) as illustrated in Fig. 2.2 where
the probability density function21 of the disturbance level which represent both time
and space variations and the probability density function of the equipment immunity
The IEC compatibility standards IEC 61000-2-2 [70] and IEC 61000-2-12 [71] give
a value for the voltage unbalance compatibility level in LV and MV power systems
single-phase loads are connected. These are based on the 95% non-exceeding proba-
bility level of statistical distributions which represent both time and space variations
of the disturbance. The IEC does not define compatibility levels for HV and EHV
systems. One of the CIGRE papers (Working Group 36.05) [72] proposes a com-
patibility level of 1% for HV networks together with the 2% level for LV and MV
20
IEC 61000-2-2 [70] and IEC 61000-2-12 [71].
21
Inherently normal distributions.
22
systems22 . Use of the compatibility levels is seen to be the usual practice in limiting
voltage unbalance at LV, and the value of 2% is commonly applied (e.g. Belgium,
Voltage characteristics are limits within which any user of a public power system
can expect the voltage to remain at the point of utilistion under normal operating
conditions [12]. That is, these limits should not be exceeded for 100% of locations for
The European standard EN 50160 [9] defines main voltage characteristics for LV
and MV distribution systems. According to this, the 95% of the 10-minute VUF
values during each measurement period of one week shall be limited at 2%. An
installations. The CIGRE/CIRED Joint Working Group C4.07 [12] recommends the
EN 50160 limit for MV networks, and further defines limits of 2% and 1.5% for HV
and EHV systems respectively based on the 95% percentile of the weekly 10-minute
values.
The South African standard NRS 048-2 [10] uses the EN 50160 limit for LV net-
works, and extends this limit to MV and HV levels together with the 3% excursion as
long as the 2% limit is not exceeded for more than 80% of time over the assessment
period. This also gives a limit of 1.5% for EHV networks. The compliance with these
limits is assessed based on the highest of the 95% weekly 10-minute values over the
The National Electricity Code Australia (NECA) [11] specifies various voltage
measured at a connection point should not vary by more than 0.5%, 1.3% and
2% for Vn > 100kV , 10kV < Vn ≤ 100kV and Vn ≤ 10kV (Vn - nominal supply
measured at a connection point should not vary by more than 0.7%, 1.3% and
2% for Vn > 100kV , 10kV < Vn ≤ 100kV and Vn ≤ 10kV respectively when
by more than 1%, 2% and 2.5% for Vn > 100kV , 10kV < Vn ≤ 100kV and
more often than once per hour by more than 2%, 2.5% and 3% for Vn > 100kV ,
24
Beyond the NECA, different states in Australia have adopted their own electricity
oped by the NEMA states that electrical supply systems should be designed and oper-
ated to limit the maximum voltage unbalance23 to 3% when measured at the electric-
utility revenue meter under no-load conditions [43]24 . Concurrently, the NEMA stan-
dard MG1-1993 [20] recommends that three-phase induction motors should be derated
for voltage unbalance levels25 greater than 1%. Some other ANSI/IEEE standards26
indicate that some electronic equipment, such as computers, may experience problems
As given in [12] and [48], voltage unbalance limits used in some other coun-
on the 10-minute values over a minimum measurement period of one week [12].
• Quebec (Canada) - 2%, 1.5% and 1% limits for MV, HV and EHV networks
respectively based on the 95% percentile of the 2-hour values over a week [12].
23
Note that this is given in terms of the LVUR.
24
This 3% value stated in ANSI C84.1-1995 is based on the minimum combined cost to utilities
and manufacturers for voltage unbalance related issues.
25
In terms of the LVUR.
26
e.g. ANSI/IEEE standard 141-1993 and ANSI/IEEE standard 241-1990.
25
• Russia - 2% and 4% based on the 95% percentile and the maximum value
respectively [48].
Planning level is a concept adopted by the IEC27 , and is a limit set for a particular
voltage level by the body responsible for the planning and operation of the supply
system. This can be considered as an internal quality objective of the system opera-
tor. Setting of the planning levels is aimed at the coordination of voltage unbalance
between various voltage levels such that the MV/LV compatibility level is not ex-
ceeded. Planning level is usually equal to or lower than the compatibility level, and
may differ from case to case depending on the network structure and circumstances.
In practice, planning levels are reflected in MV, HV and EHV systems. A statistical
Figure 2.3: Statistical interpretation of the planning level (IEC 61000-2-2, IEC 61000-
2-12)
27
IEC 61000-2-2 [70] and IEC 61000-2-12 [71].
26
As given in [12] and [16], planning levels used in some countries are listed below:
RED Joint Working Group C4.07 [12] recommends planning levels of 2%, 1.5% and
1% for MV, HV and EHV systems respectively based on the 95% daily 3-second
values and/or the 99% weekly 10-minute values. Indicative values for planning levels
Beyond the three basic categories of voltage unbalance limits: compatibility levels,
voltage characteristics and planning levels, some countries impose emission limits on
major customers such as high speed railway systems as a means for managing the
global limits. The limits applied in various countries are stated below.
work, the mean VUF value during a 1-minute period is not allowed to exceed
speed trains are 1%, and 1.5% for periods equal to or greater than 15 minutes
normally met if any connected equipment does not cause emission levels higher
than 0.7% for time ranges of minutes and 1% for time ranges of seconds [16].
28
Reported on in [12].
27
• Italy - if the subscribed power Sj of an individual user does not meet the
fault level at the PCC), emission limits of 0.7% for time ranges of minutes and
• Spain - for high speed train systems, a voltage unbalance limit of 1% is applied
higher voltage level must ensure that the current in any phase of a three-phase
electrical installation does not deviate away from the average current by more
than 2%. This limit is 5% for a customer connected at a voltage level lower
than 30kV. The Victorian distribution code [73] defines this limit as 5% and
Furthermore, IEC/TR 61000-3-13 [1] prescribes guiding principles for the assess-
rather than simply specifying limiting values. This will be critically reviewed in the
next section.
The recently released IEC Technical Report IEC/TR 61000-3-13, which is based on
the work that has been undertaken by the CIGRE/CIRED Joint Working Group
C4.103 [13, 14], is the most comprehensive document available in international con-
sensus governing voltage unbalance (negative sequence). The Scope of this report
29
Excursions up to 10% and 4% respectively are permitted for periods less than 2 minutes.
28
covers the provision of guiding principles to system operators, which can be used as a
basis for determining the requirements for the connection of unbalanced installations
to MV, HV and EHV public power systems such that adequate service quality to
all connected customers is ensured. The report addresses the coordination of voltage
unbalance between various voltage levels of a power system through the allocation of
lation, i.e. including both balanced and unbalanced parts, causing voltage unbalance.
dressed, and the distribution of these loads evenly across the three phases is considered
effects of these emissions will have on the quality of the voltage. The concepts of
compatibility level30 , planning level and emission level are used to evaluate the voltage
quality.
EHV systems and measurement results31 , IEC/TR 61000-3-13 gives indicative plan-
ning levels for MV, HV and EHV systems as reproduced in Table 2.2. These levels
consider the need to provide margins between LV, MV and HV-EHV for the purpose
provides the general guidance for adopting planning levels for specific systems32 . The
30
Compatibility levels given in IEC/TR 61000-3-13 are reproductions of the IEC compatibility
standards IEC 61000-2-2 [70] and IEC 61000-2-12 [71] which were discussed above in Section 2.7.1.
31
Reported on in [74] by the GIGRE/CIRED Joint Working Group C4.103.
32
Refer to Annex A of IEC/TR 61000-3-13.
29
measurement and evaluation procedure for the assessment of planning levels against
actual voltage unbalance levels is as per IEC 61000-4-30 [19] for instruments with
• 95% of the weekly 10-minute values should not exceed the planning level.
• The highest of the 99% daily 3-second values should not exceed the planning
system is defined as the magnitude of the unbalanced voltage vector which the con-
is the magnitude of the vector Uj illustrated in Fig. 2.437 . When this vector results
in an increased38 voltage unbalance level (i.e. |Upost-connection | > |Upre-connection |), the
emission level as defined above (i.e. |Uj | in terms of the VUF) is required to be at
approach relies on individual emission limits which are being derived39 from plan-
ning levels, the compliance against these emission limits should be assessed based
on the same measurement and evaluation procedure and indices applied for assessing
IEC/TR 61000-3-13 specifies three stages governing the approval for the connection
of unbalance, which fulfil the criteria given by (2.6), can be accepted without a
Suj
≤ 0.2% (2.6)
Ssc
39
The procedure of the derivation of individual emission limits will be discussed in Section 2.8.3.
31
where,
ply with an emission limit imposed based on a certain criteria by the system
in Section 2.8.3.
This stage of the IEC/TR 61000-3-13 voltage unbalance management procedure ap-
of this allocation approach is similar to that of the counterpart IEC harmonics [2]
and flicker [3] allocation methods. However, an additional aspect is involved in the
case of voltage unbalance, i.e. the emission which arises as a result of system inher-
ent asymmetries40 . The general principle of the proposed allocation approach is such
that, when a system is utilised to its designed capacity and all connected installations
inject their individual limits, taking also the emission arising due to system inherent
asymmetries into account, the resultant emission level which arises at any point of
• Sharing of the system capacity to absorb voltage unbalance between various sub-
manner taking into account the emission which arises as a result of system
inherent asymmetries.
practices is not random, it also varies with the load. Thus, representing these vectors
as stochastic quantities, the following general summation law which avoids the need
qX
result α
Us:x = (Uj/s:x )α (2.7)
41
i.e. numerous unbalanced installations and system inherent asymmetries.
33
where,
result
Us:x - resultant emission level (a probabilistic quantity) at a busbar x of a system S
The exponent α depends upon the chosen value of probability for the actual volt-
age unbalance level not to exceed the calculated value, and the degree to which the
Considering the 95% non-exceeding probability and the fact that the operation of
sions in practice, IEC/TR 61000-3-13 gives an indicative value for α in the absence of
specific information42 as 1.4. This indicative value would be more applicable in cases
simple radial system shown in Fig. 2.5. Each of the busbars (US, S and DS) represents
a sub-system43 , where S is the sub-system under evaluation and US and DS are the
the emission arising due to unbalanced sources that exist in the considered sub-system
and its downstream. Then, two sources which contribute to voltage unbalance in the
42
e.g. when it is known that unbalances are likely to be in-phase and coincident in time, a
summation exponent closer to unity should be used.
43
e.g. US - HV, S - MV and DS - LV.
34
system S can be identified: the global emission (Ug/s ), and the voltage unbalance
(UUus /s ) which propagates from US. By combining these two emissions using the
general summation law, the resultant voltage unbalance level Usresult at S can be
established as:
Voltage unbalance UUus /s is expressed as a product of a transfer factor and the voltage
unbalance Uus prevailing at US, i.e. UUus /s = Tus−s Uus . This factor Tus−s referred
Section 2.8.4. Rearranging (2.8) while restricting the voltage unbalance levels Uus
and Us to the respective planning levels Lus and Ls , the global emission allowance of
p
α
Ug/s = (Ls )α − (Tus−s Lus )α (2.9)
This is the level of voltage unbalance that unbalanced installations supplied by the
lines which exist in the sub-system and its downstream are allowed to cause at any
Avoiding this intermediate step, IEC/TR 61000-3-13 directly gives the formula for
allocating the global emission allowance to individual customers. However, the inclu-
44
This can be a single-bus or a multi-bus system.
35
Please see print copy for image
The underlying technique employed for apportioning the global emission allowance
of a sub-system45 to its busbars uses two power definitions, which are described below:
all power flows leaving the busbar while ignoring all power flows between the
considered busbar and other busbars. Considering the sub-system S (Figs. 2.6 -
2.7), this power Ss:x for the busbar x includes customers to be supplied directly
at the busbar (Ss:x−local ) and also at the downstream system (Ss:x−ds ), i.e. Ss:x =
Ss:x−local + Ss:x−ds .
for this power is given as the sum of power flows leaving the busbar while in-
cluding all power flows between the considered busbar and other busbars. This
ing busbars make a direct impact on the considered busbar, which is intended
the following method given by (2.10) which takes the influences of neighbouring
busbars using a transfer factor into account is recommended for assessing this
power.
Ss:x−total = k1−x Ss:1 + k2−x Ss:2 + ... + Ss:x + ... + ki−x Ss:i + ... + kn−x Ss:n (2.10)
where,
Ss:x−total - total available power of the entire sub-system as seen at the busbar x
ki−x - voltage unbalance influence coefficient between the busbar i and the
busbar x, that is defined as the voltage unbalance which arises at the busbar x
The global emission allowance Ug/s is then apportioned to the busbar x in proportion
s
α
Ss:x
Ug/s:x = Ug/s (2.11)
Ss:x−total
This is the level of voltage unbalance that unbalanced installations supplied by the
existing in the sub-system and the downstream system supplied by the busbar x are
Busbar 2
Upstream
system Busbar 1 Busbar x Busbar n
… …
Busbar 3
Sub-system S
Busbar x
Ss:x-local
Downstream system
supplied by the busbar x
Ss:x-ds
Figure 2.7: System representation of any busbar x of the system S shown in Fig. 2.6
38
As in the case of the IEC harmonics and flicker allocation, IEC/TR 61000-3-13 also
considers that the allocation of the busbar allowance Ug/s:x to a customer to be con-
nected at the busbar x based on the ratio between the agreed apparent power and
the total power to be supplied by the busbar as an equitable criteria. However, not-
ing that the global voltage unbalance at the busbar is generally caused not only by
loads lines
unbalanced installations (Ug/s:x ) but also by system inherent asymmetries (Ug/s:x ) as
expressed by (2.12), the allocation of the total Ug/s:x to installations may result in
(Ug/s:x )α = (Ug/s:x
loads α lines α
) + (Ug/s:x ) (2.12)
IEC/TR 61000-3-13 addresses this issue by introducing a new factor Kues:x . This
factor which is defined by (2.13) represents the fraction of Ug/s:x that can be allocated
represents the fraction of Ug/s:x that accounts for the emission arising as a result of
loads
!α
Ug/s:x
Kues:x = (2.13)
Ug/s:x
lines
!α
Ug/s:x
K 0 ues:x = (2.14)
Ug/s:x
where,
from upstream higher voltage (Uus ) to downstream lower voltage (UUus /s ) systems
defined as:
UUus /s
Tus−s = (2.16)
Uus
underestimation of these results in emissions above the set planning levels, whereas
In addition to the indicative values of 0.9 and 0.95 given for the MV to LV and
for estimating the MV to LV transfer coefficient (Tmv−lv ) using the system and load
1
Tmv−lv = (2.17)
ks −1
1 + km ksc−lv +1
where,
km - ratio between the rated motor load (in MVA) and the total load (in MVA) sup-
ks - ratio between the positive and negative (which is inductive) sequence impedances
ksc−lv - ratio between the short-circuit capacity (in MVA) at the LV busbar and the
Equation (2.17) indicates a value less than unity for Tmv−lv in the presence of indus-
trial load bases containing large proportions of mains connected three-phase induction
motors, and a unity transfer coefficient in relation to passive loads49 in general. That
is, motor loads help attenuating voltage unbalance as it propagates from higher volt-
presence of passive loads. Fig. 2.8 illustrates the variation of Tmv−lv with km de-
rived using (2.17) for various combinations of ks and ksc−lv values demonstrating that
Tmv−lv can be as small as 0.6. This attenuation has also been seen from measure-
ments taken by the CIGRE/CIRED Joint Working Group C4.103 [74] at a remote
1.0
0.9
0.8
Tmv-lv
0.7 ks = 5, ksc-lv = 20
ks = 5, ksc-lv = 10
0.6 ks = 5, ksc-lv = 5
0.5 ks = 7, ksc-lv = 20
ks = 7, ksc-lv = 10
0.4
0.0 0.2 0.4 0.6 0.8 1.0
km
Figure 2.8: Variation of Tmv−lv with km established using (2.17) for various combina-
tions of ks and ksc−lv values
48
Typically, ks can be in the range of 5 to 7.
49
e.g. constant current, constant impedance, constant power loads.
41
2.8.5 Factor K 0 ue
busbar emission allowance that accounts for the emission arising as a result of sys-
determine this factor based on the line construction practices and the system char-
acteristics in their specific networks. In any case, system operators are responsible
to maintain their networks such that K 0 ue allows an equitable share of the busbar
The technical report gives a rudimentary direction towards evaluating this factor
together with a set of indicative values which are reproduced in Table 2.8.5. Consid-
ering a simple radial network having an asymmetrical line50 , the direction mentioned
above indicates that the global emission caused by the line at its receiving end busbar
where,
Table 2.3: Indicative values for the factor K 0 ue given in IEC/TR 61000-3-13
System characteristics K 0 ue
- Highly meshed systems with generation locally connected near load
centers
- Transmission lines are fully transposed, otherwise lines are very
short (few km) 0.1-0.2
- Distribution systems supplying high density load areas with short
lines or cables
- Meshed systems with some radial lines which are either fully or
partly transposed
- Mix of local and remote generation with some long lines 0.2-0.4
- Distribution systems supplying a mix of high density and suburban
areas with relatively short lines (< 10km)
- Long transmission lines which are generally transposed, generation
mostly remote
- Radial sub-transmission lines which are partly transposed or
untransposed
- Distribution systems supplying a mix of medium and low density 0.5-0.4
load areas with relatively long lines (> 20km)
- three-phase motors account for only a small part of the peak load
(e.g. 10%)
43
ance Allocation
As stated earlier, the IEC approach for managing continuous power quality distur-
bances (IEC/TR 61000-3-13 [1] for voltage unbalance, IEC 61000-3-6 [2] for harmon-
ics, IEC 61000-3-7 [3] for flicker) in power systems through the allocation of emission
section reviews the work that has been undertaken on fundamental deficiencies of, and
suggested revisions to, the counterpart harmonics and flicker allocation approaches.
Being based on a common philosophy, these deficiencies and revisions may appli-
• Firstly, the level of disturbance at any point of any part of the power system
should not lead to the LV compatibility level is being exceeded. Planning levels
• Secondly, the allocation should not distinguish between different types of cus-
• Thirdly, the allocation should be equitable in some sense, i.e. larger installations
7 [76] are essentially based on the respective IEC technical reports on harmonics and
these standards, rigorous studies addressing associated deficiencies have been carried
out by the Integral Energy Power Quality and Reliability Centre at the University of
the handbook HB-264-2003 [4] which gives more prescriptive procedures for the use of
these standards. Arising as a result of these studies, it has been revealed that the ap-
plication of the standards to even a simple radial network, let alone relatively complex
has been seen to lead to situations where the set planning levels are being exceeded
even when no customer exceeds the allocated emission limit. This behaviour has been
identified when using a uniform planning level across the entire network (at a partic-
ular voltage level) as per the IEC approach. Reference [5] demonstrates, employing
a simple example, that it is not possible to derive a practical set of emission limits
such that all busbars in a network reach an uniform planning level when all emission
limits are met. Evidently, a requirement exists for either non-uniform planning levels
A revised allocation technique for both harmonics and flicker, which closely aligns
with the IEC policy, whereby emission levels at network busbars are explicitly forced
to be at or below the set planning level when all loads inject their limits derived
under the new approach, has been introduced [4, 5]. The principles of this technique
which is referred to as ‘constraint bus voltage (CBV) method’ will be reviewed in the
Consider the requirement that two customer installations (say, m and n) of agreed
apparent power Ss:x−m and Ss:x−n supplied at a busbar x shall receive the same
45
the same busbar. According to the general summation law, the combined emission
limit Es:x−mn can be written in terms of the individual emission limits Es:x−m and
Es:x−m as:
Now, suppose that the emission limit of an installation p is some function f of its
That is, f α is associative. The simplest way to satisfy (2.21) is to make the function
p
f proportional to α Ss:x−p :
p
α
f (Ss:x−p ) = Es:x−p = ka Ss:x−p (2.22)
The allocation policy as suggested by the IEC gives the emission limit Es:x−p as52 :
Ug/s p
Es:x−p = √
α
α
Ss:x−p (2.23)
Ss:x−total
By comparing (2.22) and (2.23), it can be observed that the allocation constant under
Ug/s
ka = √
α
(2.24)
Ss:x−total
Note that, under the IEC allocation policy, ka is a busbar dependant parameter as
The simplest way to ensure that busbar emission levels do not exceed the set
planning level is the relaxation of the constraint imposed by (2.24). Instead, the
result
Ug/s:x ≤ Ug/s for every busbar x (2.25)
result
The resulting global emission level Ug/s:x at any busbar x can be derived using the
general summation law in terms of busbar emission limits Es:i 54 for i = 1, 2, ..., x, ..., n
and influence coefficients ki−x between busbars i and the busbar x as:
result
p
α
Ug/s:x = (k1−x Es:1 )α + (k2−x Es:2 )α + ... + (Es:x )α + ... + (kn−x Es:n )α (2.26)
result
p
α α α α
Ug/s:x = ka k1−x Ss:1 + k2−x Ss:2 + ... + Ss:x + ... + kn−x Ss:n (2.27)
53
See Section 2.8.3.
54
i.e. the combined emission limit of a load of which the agreed apparent power is equal to the
total apparent power Ss:i supplied by the busbar.
47
Ug/s
ka = v (2.28)
u n
u X
α
max Ss:x + t
α ki−x Ss:i
i=1,i6=x
using (2.28). This new policy meets the four key allocation objectives stated above.
Firstly, (2.25) ensures that the set planning levels are not exceeded when all con-
sumers inject their allocated emission limits. Secondly, based on (2.22), customer
installations of equal MVA demand, whether connected at the same busbar or dif-
ferent busbars, receive identical emission limits. Thirdly, (2.22) ensures that larger
customer installations (in MVA demand terms) to receive larger emission levels than
smaller installations. Finally, the absorption capacity of the network is fully utilised
in the sense that at least one busbar will reach the network planning level.
This chapter has provided general information in relation to voltage unbalance, which
The key section of the chapter has given a critical discussion on the IEC/TR
61000-3-13 guidelines for voltage unbalance allocation on which the thesis is primarily
based. Step by step procedure of the development of Stage 2 emission limits together
with the related aspects, i.e. the propagation from higher voltage to lower voltage
levels, the propagation from one busbar to other neighboring busbars of a particular
sub-system, and the emission arising as a result of system inherent asymmetries, has
been clearly described establishing the backgrounds for the remaining chapters.
48
The last section of the chapter has covered the counterpart IEC approaches to
suggested revisions. The principles of this revised allocation technique for harmon-
Inherent Asymmetries
3.1 Introduction
lines
!α
Ug/s:x
K 0 uex = (3.1)
Ug/s:x
where,
K 0 uex - factor K 0 ue at any busbar x of any sub-system S (see Figs. 2.6 and 2.7)
exist in the sub-system S and the downstream system supplied by the busbar x
49
50
The emission which arises as a result of the asymmetries associated with LV feeders
is not of significant concern owing to their shorter lengths and smaller loading levels.
on which this chapter is focused, it is only the local MV lines that are responsible for
lines
the emission Ug/mv:x .
line at its receiving end busbar (labelled ‘rec’) can be expressed as:
V t
t −:g/mv:rec
Ug/mv:rec = (3.2)
V+:rec
where,
t t
|V−:g/mv:rec | = |V−:g/mv:send − (Z−+:t I+:t + Z−−:t I−:t/t + Z−0:t I0:t )| (3.3)
t
V−:g/mv:rec - negative sequence voltage caused by the line at its receiving end busbar
Z−−:t - negative sequence impedance of the line, which is inherently equal to its pos-
I+:t , I0:t - positive and zero sequence currents respectively in the line
I−:t/t - negative sequence current in the line arising as a result of the asymmetry
(labelled ‘send’), which arises as a result of the flow of negative sequence current
1
Impedance Z−−:t will be replaced with Z++:t hereafter.
51
(that is caused by the line) through the transformer coupling the upstream system
Upstream
HV system
send
MV line
(t)
MV system under
consideration
rec
Assuming that zero sequence unbalance in the network is controlled through sys-
Z−0:t I0:t can be set to zero. Then, by comparing (3.3) with the IEC approach stated
in Chapter 2 by (2.18), the IEC approach is seen to assume that the negative se-
This seems equitable when the line supplies primarily passive loads (e.g. constant
impedance loads) due to the high negative sequence impedance2 associated with such
loads. However, the validity of the above simplification is questionable when the line
three-phase induction motors, considering the fact that the current unbalance can
2
Which is equal to the positive sequence impedance.
52
be 6 to 10 times the supply voltage unbalance [20] for an induction motor due to
its relatively low negative sequence impedance. Referring to Table 2.8.5 (last entry),
counts for system inherent asymmetries it has some degree of load dependency, an
three-phase induction motors. The work carried out in this regard in relation
works, for the evaluation of the global voltage unbalance in MV systems that
The proposed methodology is verified in relation to a three-bus test system using un-
balanced load flow analysis in Section 3.4. Section 3.5 summarises the work presented
Consider the radial MV-LV system shown in Fig. 3.24 where the MV line is untrans-
t
posed. The purpose is to assess the voltage unbalance Ug/mv:rec caused by the MV line
3 lines
i.e. emissions Ug/mv:x referring to the system shown in Fig. 3.5.
4
Loads, MV-LV coupling transformers and downstream LV busbars supplied by the MV line are
represented as aggregated elements.
53
at its receiving end busbar while the voltage at the sending end busbar is balanced.
For this, the loads Smv:rec−local and Smv:rec−ds are considered as balanced.
t t
The emission Ug/mv:rec can be expressed by (3.2) where |V−:g/mv:rec | given by (3.3)
t
can be simplified for the considered scenario (i.e. V−:g/mv:send = 0) as:
t
|V−:g/mv:rec | = |Z−+:t I+:t + Z++:t I−:t/t | (3.4)
The behaviour of the negative sequence current I−:t/t or the influence of the term
t
Z++:t I−:t/t on |V−:g/mv:rec | may depend on the load type/base supplied by the line.
Thus, the aim of this section is to develop theoretical bases which describe the be-
haviour exhibited by four basic load types5 and also various load bases6 in this regard.
send
MV line
(t)
rec
Smv:rec-local (MVA)
Downstream
LV system
Smv:rec-ds (MVA)
5
i.e. constant impedance, constant current, constant power and three-phase induction motor
loads.
6
i.e. mixes of various load types.
54
When the loads Smv:rec−local and Smv:rec−ds (Fig. 3.2) represent constant impedance
loads (balanced: decoupled and equal positive, negative and zero sequence impedances)7 ,
(3.4) can be re-expressed8 using the decoupled nature of sequence impedances as:
t
Z −−:rec
|V−:g/mv:rec | ≈ Z−+:t I+:t (3.5)
Z−−:send
receiving end and the sending end busbars respectively of the MV line. Employing
the fact that the negative sequence impedances associated with all system components
are equal to their positive sequence impedances, (3.5) can be re-expressed as:
t
|V−:g/mv:rec | ≈ |Z−+:t I+:t | (1 − V Rt ) (3.6)
where, V Rt - voltage regulation of the line, which is defined as the ratio between
the positive sequence voltage drop across the line (i.e. Z++:t I+:t ) and the sending
end positive sequence voltage. That is, for constant impedance loads, the negative
sequence current I−:t/t behaves in such a manner that the term Z++:t I−:t/t causes
t
the negative sequence voltage |V−:g/mv:rec | to be smaller than the term |Z−+:t I+:t |
The negative sequence current I−:t/t can be considered to be negligible when the line
supplies constant current loads (balanced)10 , as such loads draw equal magnitudes of
three phase currents regardless of the prevailing voltage condition. Hence, (3.4) can
t
|V−:g/mv:rec | ≈ |Z−+:t I+:t | (3.7)
That is, in contrary to the case of constant impedance loads, the IEC approach: (2.18)
As the linearisation of system equations and simplifying assumptions (as in the case
of constant current loads) are not supported by constant power loads11 , through
careful examination of the results obtained from unbalanced load flow analysis, (3.8)
t
|V−:g/mv:rec | ≈ |Z−+:t I+:t | (1 − V Rt )β (3.8)
where, β ≈ −1 and −2 for low (∼ 0.9) and high (∼ 1) lagging power factor (pf)
conditions respectively. That is, for constant power loads, the term Z++:t I−:t/t causes
10
i.e. the three phase supply is connected to equal loads where each of the loads is modelled as
per O.5 - O.6 (Appendix O, Section O.4.2) with λp = λq = 1. In this case, the three-phase load
bank draws equal magnitudes of three phase currents regardless of the voltage condition (including
unbalance) which prevails at the terminals of the load bank.
11
i.e. when the three phase supply is connected to equal loads where each of the loads is modelled
as per O.5 - O.6 (Appendix O, Section O.4.2) with λp = λq = 0. In this case, the power drawn by
each of the three phases of the load bank is equal and does not depend on the voltage condition
(including unbalance) which prevails at the terminals of the load bank.
56
t
the negative sequence voltage |V−:g/mv:rec | to be greater than the term |Z−+:t I+:t |
constant13 sequence impedances. Hence, (3.4) can be re-expressed14 in the form given
by (3.5). This can be expressed15 in terms of the system, line and load characteristics
t 1
|V−:g/mv:rec | ≈ |Z−+:t I+:t | (3.9)
V Rt 1
1+ 1−V Rt 1
+k 1
ks sc−lvagg
where,
ks - ratio between the positive and negative sequence impedances of the aggregated
Smv:rec−ds - total load (in MVA) supplied by the aggregated LV busbar (see Fig. 3.2)
(Vn−lv )2
Ssc−lvagg = |Z++:sys−lv |
, the short-circuit capacity (in MVA) at the aggregated LV
busbar, that is derived using the positive sequence system impedance Z++:sys−lv which
exists between the busbar under evaluation and the downstream LV busbar17
Equation (3.9) implies that the negative sequence current I−:t/t in the presence of
induction motor loads behaves in such a manner that the term Z++:t I−:t/t causes
t
the negative sequence voltage |V−:g/mv:rec | to be smaller than the term |Z−+:t I+:t |
Z loads: LF
Z loads: Eq. (3.6) IEC approach:
|Z-+:t I+:t|
80 I loads: LF
I loads: Eq. (3.7)
PQ loads: LF
PQ loads: Eq. (3.8)
IM loads: LF
|V -:g/mv:rec| (V)
40
t
0
|I+:t| (A) 50 150 250 350 450 550 650
VRt (%) 1 3 5 6 8 10 12
t
Figure 3.3: Variation of |V−:g/mv:rec | with |I+:t | (V Rt values corresponding to various
|I+:t | are also indicated) for the four basic load types
3.2.5 Discussion
Taking a generalised view, the additional factors associated with the above proposed
for the three passive load types (where γ = 1 for constant impedance loads, γ = 0
for constant current loads, and −2 ≤ γ ≤ −1 for constant power loads). Consider-
ing most practical circumstances where V Rt < 10%, the factor (1 − V Rt )γ can be
approximated to unity (in other words I−:t/t ≈ 0) supporting the IEC approach for
The additional factor associated with the proposed expression for induction motor
loads is considerably smaller18 than that for passive loads, implying that the IEC
t
Fig. 3.3 illustrates the variation of the negative sequence voltage |V−:g/mv:rec | with
the current |I+:t | for the four load types19 obtained employing the test system20 de-
scribed in Appendix B. This shows the variations established using the proposed
expressions (3.6) - (3.9) in comparison to the results obtained using unbalanced load
flow (LF) analysis21 , justifying the new formulation and also the above discussion.
When the MV line supplies a mix of passive loads (at MV and/or LV) and induction
motors (at LV), the negative sequence current I−:t/t can be decomposed as:
(PS) and induction motor (IM) branches respectively arising as a result of the asym-
t
|V−:g/mv:rec | = |Z−+:t I+:t + Z++:t I−:t/ps + Z++:t I−:t/im | (3.11)
As discussed in Section 3.2.5, the current component I−:t/ps associated with passive
t
|V−:g/mv:rec | ≈ |Z−+:t I+:t + Z++:t I−:t/im | (3.12)
t
Z −−:rec−im
|V−:g/mv:rec | ≈ Z−+:t I+:t (3.13)
Z−−:send−im
the receiving end and the sending end busbars respectively of the MV line taking
into account only induction motors supplied by the downstream LV busbar. Re-
expressing22 (3.13):
t 1
|V−:g/mv:rec | ≈ |Z−+:t I+:t | (3.14)
V Rt klv
1+ 1−V Rt 1
+k 1
ks km sc−lvagg
where,
km - ratio between the rated motor load (in MVA) and the total load (in MVA) sup-
t
The emission Ug/mv:rec can be established in a generalised form by substituting (3.14)
in (3.2) as:
t |Z−+:t I+:t | 1
Ug/mv:rec ≈ × (3.15)
V Rt klv |V+:rec |
1+ 1−V Rt 1
+k 1
ks km sc−lvagg
22
Derivation of (3.14) is given in Appendix C.
60
t
Fig. 3.4 illustrates the variation of Ug/mv:rec with km established using (3.15) in
comparison to the results obtained using unbalanced load flow analysis for the test
• klv = 0.5
• klv = 0 (i.e. Smv:rec−ds = 0, which implies that no motor loads are supplied by
the MV line)
The results presented in Fig. 3.4 correspond to a selected operating scenario where
the MV line supplies a total of 10MVA load at 0.9 lagging pf resulting in |I+:t | ≈ 470A,
V Rt ≈ 8.5% and |V+:rec | ≈ 7.2kV (1pu). These results confirm the above basis given
by (3.15) which describes the behaviour of various load bases in relation to the global
untransposed lines (t12 , t13 , t23 , ..., tij , ...), the system representation of any busbar x
of the system can be taken as per Fig. 3.625 where the downstream system represents
an aggregated LV system. For the purpose of assessing the global emission arising
as a result of the line asymmetries, the voltage at the upstream system and all loads
0.7
klv = 0.5
U g/mv:rec (%)
0.6
t
klv = 1 (Smv:rec-local = 0)
0.5
Load flow results
Eq. (3.15)
0.4
0.0 0.2 0.4 0.6 0.8 1.0
km
t
Figure 3.4: Variation of Ug/mv:rec with km for the cases where klv = 1, klv = 0.5 and
klv = 0
Busbar 2
Upstream
HV system Busbar 1 Busbar x Busbar n
… …
Busbar 3
MV sub-system
Busbar x
Smv:x-local
Downstream
system supplied
by the busbar x
Smv:x-ds
Figure 3.6: System representation of any busbar x of the MV system shown in Fig. 3.5
Employing the linearity of negative sequence variables [1], the resultant negative
lines
sequence voltage V−:g/mv:x arising as a result of the interaction of the untransposed
lines t12 , t13 , t23 , ..., tij , ... at any busbar x can be written as:
ij t
where, V−:g/mv:x - negative sequence voltage caused by any line tij on its own at the
lines
busbar x. Then, the emission Ug/mv:x can be expressed as:
V lines
lines −:g/mv−x
Ug/mv:x = (3.17)
V+:x
Extending the nodal equations [I] = [Y ][V ]26 to the sequence domain, xth element
of [I], (x, y)th element of [Y ] and xth element of [V ] can be written respectively for
I0:x
Ix =
I+:x
(3.18)
I−:x
Y00:xy Y0+:xy Y0−:xy
Yxy =
Y+0:xy Y++:xy Y+−:xy
(3.19)
Y−0:xy Y−+:xy Y−−:xy
V0:x
Vx =
V+:x
(3.20)
V−:x
where,
Ix - xth element of [I] or the nodal current vector at the busbar x, which is considered
to the busbar x (= y)
• for x 6= y, Yxy is equal to the negative value of the admittance of the network
Yλ∆:xy - λ−∆ sequence coupling admittance element of Yxy for λ 6= ∆, and λ sequence
The nodal equations [I] = [Y ][V ] of which the elements are given by (3.18) - (3.20)
can be expanded while setting zero sequence voltages and currents to zero to establish
−I−:1 Y Y−+:12 . . . Y−+:1n V
−+:11 +:1
−I−:2 Y−+:21 Y−+:22 . . . Y−+:2n V+:2
= . +
. .. .. .. ..
. .
. . . . . .
−I−:n Y−+:n1 Y−+:n2 . . . Y−+:nn V+:n
Y Y−−:12 . . . Y−−:1n V
−−:11 −:1
Y−−:21 Y−−:22 . . . Y−−:2n V−:2
+ (3.21)
.. .. .. .. .
.
. . . . .
Y−−:n1 Y−−:n2 . . . Y−−:nn V−:n
Equation (3.21) can be rewritten in a concise form also employing the inherent rela-
Emphasising that these negative sequence nodal currents and voltages arise as a re-
sult of the untransposed lines of the considered network, the matrices [I− ] and [V− ]
lines
are relabelled as [I−:lines ] and [V−:g/mv ] respectively for consistency. As shown in Sec-
tion 3.2 above, the influence of the negative sequence currents [I−:lines ] on the negative
lines
sequence voltages [V−:g/mv ] should be taken into account in the presence of consider-
able proportions of induction motor loads. Thus, the negative sequence nodal current
I−:lines/x at any busbar x can be generally written when the busbar supplies a mix of
passive (at MV and/or LV) and motor (at LV) loads as:
65
lines
I−:lines/x ≈ Y−−:x−im V−:g/mv:x (3.23)
taking into account only induction motors supplied by the downstream LV busbar.
This admittance Y−−:x−im , which is inherently inductive owing to the inductive nature
of the associated induction motor negative sequence impedances and MV-LV trans-
former impedances, can be expressed27 in terms of the system and load characteristics,
where,
0
lines
[V−:g/mv ]n×1 ≈ −[Y++ ]−1
n×n [Y−+ ]n×n [V+ ]n×1 (3.25)
where,
0
Y++:xy ≈ Y++:xy + Y−−:x−im for x = y
0
Y++:xy = Y++:xy for x 6= y
27
Derivation of (3.24) is given in Appendix D.
28
Additional subscript ‘x’ indicates the quantities corresponding to the busbar x.
66
That is, taking the nodal positive sequence voltages as known quantities as they
can be easily obtained from conventional balanced load flow analysis, the negative
lines
sequence voltage V−:g/mv:x at any busbar x can be established using (3.25).
three-wire) shown in Fig. 3.7. Considered operating scenario and resulting positive
sequence system conditions29 are also indicated in Fig. 3.7. Lengths of the lines
which are taken as identical in construction (including the phase positioning) and
untransposed are shown alongside the lines. Relevant admittance data of the lines
are30 :
and constant power elements. That is, km:x = 0 implying that Y−−:x−im ≈ 0 for
x = 1 and 3. Busbar 2 supplies loads at the LV level which account for 40% of the
total load supplied by the system. Two cases based on the type of LV loads are
considered:
• Case 1 - LV loads represent passive elements31 . That is, km:2 = 0 implying that
Y−−:2−im ≈ 0.
listed above. This shows the results established using the proposed methodology32 in
comparison to the results obtained using unbalanced load flow analysis validating the
proposed technique. Further, these results reveal that the presence of considerable
lines
proportions of induction motor loads increases the emission Ug/mv:x when x represents
the busbar that is directly connected to the upstream system (e.g. busbar 1 of the
system shown in Fig. 3.7), compared to the case where only passive loads are supplied
lines
by the network. In addition, induction motor loads tend to reduce the emission Ug/mv:x
at all other busbars (e.g. busbars 2 and 3 of the system shown in Fig. 3.7), compared
204A
MV busbar 2
(0.98pu, -6.740) LV (460V)
4MVA
0.9 lagging pf
10km
Upstream HV 135A
(66kV) system MV busbar 1
(1.05pu, 00) (1.05pu, -4.070)
10km
59A
5km
152A
4MVA
0.9 lagging pf
2MVA
0.9 lagging pf
MV busbar 3
(1.01pu, -5.570)
HV-MV coupling transformer – 12MVA, winding resistance = 1%, leakage reactance = 10%,
secondary tap setting = 1.05pu
MV-LV coupling transformer – aggregated representation of fully loaded 1MVA transformers with
winding resistance = 1%, leakage reactance = 5% and secondary tap setting = 1.05pu
Figure 3.7: Three-bus MV test system considered for applying the proposed method-
ology
32
Application of the methodology to the test system is described in Appendix E.
68
0.4
U glines
0.2
0.0
1 2 3
Busbar
lines
Figure 3.8: Emissions Ug/mv:x for the three-bus MV test system for the two cases
where km:2 = 0 and km:2 = 1
This chapter has addressed the global voltage unbalance in MV power systems which
arises as a result of line asymmetries. This is a key aspect in assessing emission limits
The dependency of the global emission on various load types/bases including three-
line. The following major conclusions can be drawn from the study:
only when the system supplies primarily passive loads. In this case, the impact
• The IEC approach has been seen to be conservative when the network supplies
a large proportion of induction motor loads. In this case, the negative sequence
on the proportion of motor loads, and secondarily on the system and motor
characteristics.
ating the global emission caused by line asymmetries at nodal level taking the line,
system and load characteristics, system operating conditions and downstream load
composition into account has been proposed. The results established using the pro-
posed methodology in relation to a three-bus test system has been seen to be in close
agreement with the results obtained using unbalanced load flow analysis. Further,
directly connected to the upstream system, compared to the case where only
• Induction motor loads tend to reduce the global emission levels at all other
busbars of the network, compared to the case where only passive loads exist.
Chapter 4
Inherent Asymmetries
4.1 Introduction
The global voltage unbalance in MV power systems which arises as a result of line
motor loads has been investigated, emphasising the limitations associated with the
also the MV lines that is present in the downstream system supplied by the busbar.
HV line (labelled ‘t’) which supplies an untransposed MV line (labelled ‘td ’) at the
1
Stated in Chapter 2 by (2.18).
70
71
t+td
downstream, the emission Ug/hv:rec caused by the lines2 at the receiving end busbar
t+t
V d
t+td −:g/hv:rec
Ug/hv:rec = (4.1)
V+:rec
where,
t+td t td
|V−:g/hv:rec | = |V−:g/hv:rec + V−:g/hv:rec | (4.2)
t+td
V−:g/hv:rec - negative sequence voltage caused both by the HV and MV lines at the
busbar
td
V−:g/hv:rec - negative sequence voltage caused only by the MV line4 at the receiving
V+:rec - positive sequence voltage at the receiving end busbar of the HV line
t t
V−:g/hv:rec = V−:g/hv:send − (Z−+:t I+:t + Z−−:t I−:t/t ) (4.3)
td
The symbols in (4.3) are as defined in Chapter 3 for (3.3)6 . The second term V−:g/hv:rec
td td
V−:g/hv:rec = V−:g/hv:send − Z−−:t I−:td /t (4.4)
where,
I−:td /t - negative sequence current in the HV line, which arises as a result of the
busbar (labelled ‘send’) of the HV line, which arises as a result of the flow of the
negative sequence current (that is caused by the MV line) through the transformer
Upstream
system
send
rec
MV line
(td)
By comparing (4.2) of which the two terms are given by (4.3) and (4.4) respectively,
with the IEC approach, it can be identified that the IEC approach infers the following
• Negligible influence from the negative sequence currents that are caused by the
tion can lead to a significant degree of error when the load supplied by an HV
• Negligible influence from the negative sequence currents that are caused by the
downstream MV lines. In other words, the IEC approach leaves out the presence
networks. However, the asymmetry associated with MV lines, which was seen
to influence the local emission in the presence of large motor proportions, can
• To investigate the influence of line asymmetries, which include the local HV lines
of induction motor loads. The work carried out in this regard in relation to a
that is caused by the local and downstream line asymmetries at nodal level.
The proposed methodology is applied to a three-bus test system and the results are
compared with those obtained using unbalanced load flow analysis in Section 4.4.
Section 4.5 further verifies the proposed methodology employing the IEEE 14-bus
test system which supplies passive loads locally8 . Section 4.6 summarises the work
Consider the radial HV-MV-LV network shown in Fig. 4.29 where the HV and MV
t+td
lines of interest are untransposed. The purpose is to assess the emission Ug/hv:rec that
is caused by the line asymmetries at the receiving end busbar of the HV line while
the voltage at the sending end busbar is balanced. For this, all loads supplied by the
t+td
The emission Ug/hv:rec can be expressed by (4.1) where the two components of
t+td
the resultant negative sequence voltage |V−:g/hv:rec | which are given by (4.3) and (4.4)
t
respectively can be simplified for the considered scenario (i.e. V−:g/hv:send = 0 and
td
V−:g/hv:send = 0) as10 :
t
V−:g/hv:rec = −(Z−+:t I+:t + Z++:t I−:t/t ) (4.5)
td
V−:g/hv:rec = −Z++:t I−:td /t (4.6)
8
i.e. at the HV level itself.
9
Fig. 4.2 gives a generalised representation of the downstream system supplied by the HV line.
LV busbars, where motor loads are connected, which can be supplied through MV lines (labelled
‘LVr ’ where subscript ‘r’ indicates the receiving ends of the MV lines) and also directly by HV-MV
coupling transformers (i.e. not through MV lines, labelled as ‘LVs ’ where subscript ‘s’ indicates
the sending ends of the MV lines) are separately considered. Loads, HV-MV and MV-LV coupling
transformers, and downstream MV lines and MV and LV busbars supplied by the HV line are
represented as aggregated elements.
10
The impedance Z−−:t is replaced with Z++:t .
75
send
HV line
(t)
rec
MV line
(td)
LVs
Downstream
LVr system supplied
by the HV line
t+td
The resultant negative sequence voltage |V−:g/hv:rec |, which is the absolute value of
the sum of (4.5) and (4.6), can be generally expressed11 when the network supplies a
mix of passive loads (at HV, MV and/or LV) and induction motors (at LV) as:
t+td
|V−:g/hv:rec | ≈ |Z−+:t I+:t (1 − µ − ζ)| (4.7)
where,
Z++:t
µ = Z−−:send−im
11
Derivation of (4.7) is given in Appendix F.
76
Z−−:send−im - downstream negative sequence impedance seen at the sending end bus-
bar of the HV line taking into account only induction motors supplied by the LV
ζ = µktd kn σ
Z−+:td −hv
σ= Z−+:t
(a complex quantity)
the MV line
ktd - ratio between the total load supplied by the MV line, and the total load supplied
kn - ratio between the downstream negative sequence impedance12 seen at the sending
end busbar of the MV line taking into account the total motor load13 supplied by the
HV line, and the downstream negative sequence impedance seen at the sending end
busbar of the MV line taking into account only the motor load14 supplied the MV line
t+td
The emission Ug/hv:rec can be established in a generalised form by substituting (4.7) in
(4.1) as:
t+td
Z−+:t I+:t (1 − µ − ζ)
Ug/hv:rec ≈
(4.8)
V+:rec
The factors µ and ζ (= µktd kn σ) represent the impact of the term Z++:t I−:t/t in
(4.5) and of the term given by (4.6) or of the downstream MV line respectively on the
t+td
emission Ug/hv:rec in the presence of induction motor loads. On the whole, the term
of which the magnitude can be greater or smaller than µ and ζ noting the phase
angle involved with σ. For the purpose of demonstrating the level of this influence,
12
Ignoring the presence of any passive loads.
13
i.e. the motor load supplied both by the busbars LVs and LVr .
14
i.e. the motor load supplied only by the busbar LVr .
77
the factor µ is expressed15 in terms of the system and load characteristics, system
operating conditions and downstream load composition for a simplified case where
motor loads are supplied only by the MV line at the busbar LVr 16 as:
1
µr ≈ „ « (4.9)
1 1
(1−V Rt )(1−V Rtd )2 ksr kmr
+k
sc−lvragg
1+ V Rt klvr
where,
µr - factor µ corresponding to the simplified case where motor loads are supplied only
ksr - ratio between the positive and negative sequence impedances of the aggregated
kmr - ratio between the rated motor load (in MVA) and the total load (in MVA)
klvr - ratio between the total load (in MVA) supplied by the busbar LVr , and the
total load (in MVA) supplied by the HV busbar rec under evaluation
ksc−lvragg - ratio between the short-circuit capacity19 (in MVA) at the aggregated
busbar LVr , and the the total load (in MVA) supplied by the busbar LVr
15
Derivation of (4.9) is given in Appendix G.
16
i.e. the busbar LVs , which is directly supplied by the HV-MV coupling transformer, supplies
primarily passive loads resulting in a kn = 1.
17
Voltage regulation is defined as the ratio between the positive sequence voltage drop across the
line (e.g. Z++:t I+:t for the HV line), and the sending end positive sequence voltage.
18
Typically, ksr can be in the range of 5 to 7.
19
Which is derived using the positive sequence system impedance Z++:sys−lv = Z++:hm−lv +
Z++:td −lv +Z++:mlr −lv that exists between the HV busbar rec under evaluation and the downstream
busbar LVr , where Z++:hm−lv , Z++:td −lv , Z++:mlr −lv are the positive sequence impedances, referred
to LV, of the HV-MV coupling transformer, MV line and MV-LV coupling transformer supplying
the busbar LVr respectively.
78
t+td
Fig. 4.3 illustrates the variation of Ug/hv:rec with klvr established using (4.8) in
comparison to the results obtained using unbalanced load flow analysis for the sim-
klvr are given in Appendix H. Further, the levels of the influence of the factors µ21
(i.e. of the term Z++:t I−:t/t ) and ζ 22 (i.e. of the downstream MV line) on the emission
in each case are also indicated in Fig. 4.323 . Arising from these results, the following
can be concluded:
t+td
• For passive loads, the influence of the factors µ and ζ on the emission Ug/hv:rec
is insignificant implying that the IEC approach can be accepted only when the
line supplies primarily passive loads also in assessing HV systems. That is, in
the case of passive loads, the local HV lines are totally responsible for the global
t+td
• The presence of induction motors affects the level of emission Ug/hv:rec given by
the IEC approach noticeably24 . This dependency of the global emission on the
20
Where, |Z−+:t | = 0.5226Ω, ksr = 6.7, kn = 1, ktd = klvr .
21
For the case of kmr = 0, µ is the difference between the results established using (4.8) for
kmr = 0, and those obtained using unbalanced load flow analysis for kmr = 0 without introducing
the effects of the downstream MV line asymmetry. For the case of kmr = 1, µ is the difference
between the results established using (4.8) for kmr = 0, and those obtained using unbalanced load
flow analysis for kmr = 1 without introducing the effects of the downstream MV line asymmetry.
22
For the both cases of kmr = 0 and kmr = 1, ζ is the difference between the results established
using unbalanced load flow analysis with and without introducing the effects of the downstream MV
line asymmetry.
23
The overall influence introduced by motor loads on the emission for the test case is given by the
summation µ + |ζ| noting that the phase angle (see Table H.1) involved with σ is zero (i.e. phase
angles of the impedances Z−+:t and Z−+:td of the HV and MV lines respectively are equal).
24
e.g. motor loads weighted only 20% cause 10% reduction in the emission compared to that when
only passive loads exist.
79
motor proportion can be seen in two forms which are explained by the factors µ
and ζ respectively. Prominently, in the case of motor loads, not only the local
HV line but also the downstream MV line are equally responsible in determining
0.6 d
for kmr = 1
kmr = 1
caused only by t
ζ | Z − +:t I +:t |
0.5
kmr = 1
d
caused by both t and t d for kmr = 1
0.4
0.0 0.2 0.4 0.6 0.8 1.0
k lvr
t+td
Figure 4.3: Variation of Ug/hv:rec with klvr for the two cases where kmr = 0 and
kmr = 1
Considering the network shown in Fig. 4.425 as an HV system with asymmetrical lines
(t12 , t13 , t23 , ...), the system representation of any busbar x of the system can be taken
as per Fig. 4.5. This system supplied by the busbar x consists of a radial26 MV-LV
network with an untransposed MV line td:x . For assessing the global emission arising
25
Reproduction of Fig. 2.6.
26
Which is the usual practice.
80
as a result of the line asymmetries, the voltage at the upstream system and all loads
Busbar 2
Upstream
EHV system Busbar 1 Busbar x Busbar n
… …
Busbar 3
HV sub-system
determined not only by the local line asymmetries but also by the downstream MV
lines asymmetries when the network supplies considerable proportions of motor loads.
lines
Thus, the resultant negative sequence voltage V−:g/hv:x which arises as a result of the
(4.10)
where,
tij
V−:g/hv:x - negative sequence voltage caused by any local HV line tij on its own at
the busbar x
81
td:i
V−:g/hv:x - negative sequence voltage caused by any downstream MV line td:i on its
lines
Then, the emission Ug/hv:x and the factor K 0 uex can be expressed in the forms which
Busbar x
MV line
(td:x)
LVs:x
Downstream
system supplied by
the HV busbar x
LVr:x
Figure 4.5: System representation of any busbar x of the HV system shown in Fig. 4.4
([I−:lines ]) which arise as a result of the line asymmetries can be written in terms
of the nodal negative-positive sequence coupling admittances ([Y−+ ]), nodal positive
lines
sequence admittances ([Y++ ]) and nodal positive ([V+ ]) and negative [V−:g/hv ] se-
lines
−[I−:lines ] = [Y−+ ][V+ ] + [Y++ ][V−:g/hv ] (4.11)
The current I−:lines/x which is the xth element of [I−:lines ] or the negative sequence
nodal current at the busbar x can be generally written when the busbar supplies a
mix of passive (at HV, MV and/or LV) and motor (at LV) loads as:
lines
I−:lines/x ≈ Y−−:x−im V−:g/hv:x + Y−+:x V+:x (4.12)
the busbar x taking into account only induction motors that are normally supplied
for the simplified case where motor loads are supplied only by the MV line td:x at the
√ !
klvr:x 3 |I+:x |
Y−−:x−imr ≈ −j (4.13)
(1 − V Rtd:x )2 1
+ 1 Vn−hv
ksr:x kmr:x ksc−lvragg :x
where,
mittance seen at the busbar x, which arises as a result of the asymmetry of the MV
line td:x . This admittance can be generally expressed29 by (4.14) and (4.15):
27
Derivation of (4.13) can be described using (D.1), (D.3) and (G.6).
28
Additional subscript ‘x’ indicates the quantities corresponding to the busbar x.
29
Refer to Appendix I for the derivation.
83
√ !
3 |I+:x |
|Y−+:x | ≈ ktd:x kn:x |Y−−:x−im Z−+:td:x −hv | (4.14)
Vn−hv
θY−+:x ≈ 900 + θZ−+:td:x + θpf :x (4.15)
where,
MV line td:x
θY−+:x , θZ−+:td:x , - phase angles of the admittance Y−+:x and the negative-positive se-
0
lines
[V−:g/hv ]n×1 ≈ −[Y++ ]−1
n×n [Y−+ ]n×n [V+ ]n×1 (4.16)
where,
0
Y++:xy ≈ Y++:xy + Y−−:x−im for x = y
0
Y−+:xy ≈ Y−+:xy + Y−+:x for x = y
0
Y++:xy = Y++:xy for x 6= y
0
Y−+:xy = Y−+:xy for x 6= y
That is, taking the nodal positive sequence voltages as known quantities, the negative
lines
sequence voltage V−:g/hv:x at any busbar x can be established using (4.16).
30
− and + for lagging and leading conditions respectively.
84
erators and synchronous condensers in a system force the negative sequence voltage
ance. Equation (4.16) which gives the nodal negative sequence voltages arising as a
result of line asymmetries does not consider the presence of such components, and
thus requires suitable adjustments such that the influence of zero voltage unbalance
(or of voltage controlled components) at given busbars on the emission levels at other
busbars is accommodated.
the considered HV network (see Fig. 4.4). Hence, the negative sequence voltage V−:i
at the busbar i is zero, and there are only (n − 1) number of busbars at which the
(3.21) given in Chapter 3, the negative sequence current (labelled I−ci :x 31 ) at any
I−ci :x = (Y−+:x1 V+:1 + ... + Y−+:xi V+:i + ... + Y−+:xn V+:n ) + (Y++:x1 V−:1 + ...
Noting that the term associating the admittance Y++:xi is absent in (4.17), the ma-
trix equation (3.22) can be modified accordingly to incorporate the influence of the
busbars by:
0
• Reducing the dimension of the matrix [Y++ ] down to (n − 1) × (n − 1) by
System
three-wire) shown in Fig. 4.6. Considered operating scenario and resulting positive
sequence system conditions32 are also indicated in Fig. 3.7. Lengths of the HV
lines which are taken as identical in construction (including the phase positioning33 :
(0.2061∠600 Skm)
and constant power elements directly at the HV-MV coupling transformers. This
implies that Y−−:x−im ≈ 0 and Y−+:x ≈ 0 for x = 1 and 3. Busbar 2 supplies loads at
the LV level through 3.2187km of untransposed MV lines35 , which account for 40%
of the total load supplied by the system. Two cases based on the type of LV loads
are considered:
32
Nodal voltages and line currents, which are obtained using load flow analysis.
33
This shows the considered arrangement of the three phase conductors (a, b and c) of the hori-
zontal tower.
34
Refer to Appendix H for further details.
35
Refer to Appendix B for the tower construction and conductor data.
86
208A
HV busbar 2
(1.01pu, -10.70) LV (460V)
10MVA
0.9 lagging pf
tap = 1.1pu
10MVA
0.9 lagging pf
50km MV lines –
Upstream EHV 130A 12.47kV
(230kV) system HV busbar 1 3.2187km
(1.1pu, 00) (1.07pu, -8.30) 150A
VR = 9%
40km
78A
20km
167A
tap = 1pu
tap = 1.03pu
10MVA
0.9 lagging pf
HV busbar 3
(1.04pu, -9.60)
10MVA 10MVA
0.9 lagging pf 0.9 lagging pf
EHV-HV coupling transformer – 60MVA, winding resistance = 1%, leakage reactance = 20%,
secondary tap setting = 1.1pu
HV-MV coupling transformers – 12MVA, winding resistance = 1%, leakage reactance = 10%
MV-LV coupling transformers – aggregated representation of fully loaded 1MVA transformers with
winding resistance = 1% , leakage reactance = 5%, and secondary tap setting = 1.08pu
Figure 4.6: Three-bus HV test system considered for applying the proposed method-
ology
87
• Case 1 - LV loads represent passive elements36 . That is, km:2 = 0, implying that
lines
Fig. 4.7 illustrates the emissions Ug/hv:x at the HV busbars established using the
flow analysis for the two cases listed above, clearly indicating the influence of the
j0.1093) × 10−3 S.
lines
Fig. 4.8 illustrates the emission Ug/hv:x at each of the HV busbars established
using the proposed methodology and unbalanced load flow analysis for the case where
km:2 = 1 (i.e. when busbar 2 supplies motor loads) in relation to two different phase
arrangements of the MV lines, indicating their influence40 in each case on the resultant
emission levels.
Note that this phase positioning is as same as that of the local HV lines.
36
With equal compositions of constant impedance and constant power loads.
37
Application of the methodology to the test system is described in Appendix J.
38
Which is the difference between the results established using unbalanced load flow analysis with
and without introducing the effects of the downstream MV line asymmetries.
39
See Appendix B.
40
Which is the difference between the results established using unbalanced load flow analysis with
and without introducing the effects of the downstream MV line asymmetries.
88
0.5
/ hv:x (%)
lines
Figure 4.7: Emissions Ug/hv:x for the three-bus HV test system for the cases where
km:2 = 0 and km:2 = 1
quence coupling impedance and admittance values are 0.0349∠1500 Ω/km and
Figs. 4.7 and 4.8 demonstrate that there is a good agreement between the results
obtained using the proposed technique and unbalanced load flow analysis. Further,
based on the results presented in Figs. 4.7 and 4.8, the following can be revealed:
(e.g. busbar 1 of the system shown in Fig. 4.7) and lower at all other busbars
(e.g. busbars 2 and 3 of the system shown in Fig. 4.7), compared to the case
0.5
Caused only by the HV lines:
methodology
/ hv: x (%)
0.4
load flow
lines
Figure 4.8: Emissions Ug/hv:x for the three-bus HV test system for the case where
km:2 = 1 in relation to the Phase arrangements I and II of the MV lines
in Section 4.2 above. This influence of the downstream MV lines can either
decrease (e.g. for Phase positioning I in Fig. 4.8) or increase (e.g. for Phase
lines
positioning II in Fig. 4.8) the resultant emission Ug/hv:x compared to the local
Test System
The proposed methodology is further applied to the IEEE 14-bus test system shown
in Fig. 4.9 which consists of positive sequence voltage controlled busbars (busbars
1, 2, 3, 6 and 8), taking it as a 66kV, 60Hz and three-wire network supplying constant
90
power loads at the HV level. The system data are as per [77] with appropriate
and minor modifications, which are given in Appendix K together with the nodal
Lengths43 of the lines are established such that their positive sequence impedance
lines
Fig. 4.10 illustrates the emission Ug/hv:x at each of the HV busbars established
using the methodology in comparison to the results obtained using unbalanced load
the global voltage unbalance in HV power systems which arises as a result of line
recommendations.
The dependency of the global emission on the local HV lines as well as on the
downstream MV lines in the presence of passive and induction motor loads has been
examined in relation to a simple radial network. The following major conclusions can
0.5
Load flow results
Methodology
0.4
/ hv:x (%)
0.3
U glines
0.2
0.1
0.0
1 2 3 4 5 6 7 8 9 10 11 12 13 14
Busbar
lines
Figure 4.10: Emissions Ug/hv:x for the IEEE 14-bus test system
92
metrical radial line on the global emission can be applied also to an HV system
however only when the network supplies primarily passive loads. In this case,
the local HV lines are totally responsible for the global emission, and the con-
• The presence of induction motors has been seen to affect the level of emission
given by the IEC approach noticeably. This dependency of the global emission
emission in HV networks.
global emission caused both by the local and downstream line asymmetries at nodal
level has been proposed. The results established using the proposed methodology
in relation to a three-bus test system has been seen to be in close agreement with
the results obtained using unbalanced load flow analysis. Furthermore, these results
busbar which is directly connected to the upstream system and lower at all other
busbars of the network, compared to the case where only passive loads exist.
• The influence of the downstream MV lines can either decrease or increase the
resultant emission levels with respect to the local emission levels depending on
93
The proposed methodology is further validated in relation to the IEEE 14-bus test
system.
Chapter 5
5.1 Introduction
• Influence coefficients which give a measure of the propagation from one busbar
The influence coefficient ki−x between busbars i and x is defined as the voltage
unbalance which arises at the busbar x when 1pu of negative sequence voltage
94
95
determining the total available apparent power Stotal−x of the entire sub-system
as seen at the busbar x under evaluation to take into account the contributions
1.15
1.10
1.05
Tmv-lv
1.00
0.95
0.90
10 13 16 19 22 25
ksc-lv
Figure 5.1: Variation of Tmv−lv with ksc−lv obtained for constant power loads using
unbalanced load flow analysis
cient = 1 is mathematically trivial for constant impedance loads. However, it may not
be valid for other load types such as constant power and constant current loads, owing
to the different behaviours exhibited by these load types under unbalanced supply
the variation of the transfer coefficient Tmv−lv with ksc−lv 2 ) established when an LV
system supplies a load base primarily having constant power loads with 0.9 lagging pf
using unbalanced load flow analysis, compared against unity. Noting that the transfer
coefficient Tmv−lv is considerably greater than unity at relatively lower ksc−lv values
2
As defined in Chapter 2 for (2.17), ksc−lv is the ratio between the short-circuit capacity (in
MVA) at an LV busbar and the total load (in MVA) supplied by the LV busbar.
96
(i.e. heavily loaded LV systems), it is evident that the assumption of a unity transfer
coefficient for passive loads used in the IEC method cannot be generally applied with
a high degree of accuracy. That is, a requirement exists for cautious examination
systematic approaches for assessing other transfer coefficients (e.g. HV to MV) and
load types with regard to the propagation of voltage unbalance from higher
in Section 5.2.
Section 5.3 presents this, together with a comparison of the results obtained us-
ing the proposed technique and unbalanced load flow analysis in relation to an
Considering the system shown in Fig. 5.23 where the US (upstream higher voltage
system) to S (lower voltage system under evaluation) transfer coefficient Tus−s which
where,
US
Uus/s
Ss-local
DS
Ss-ds
Figure 5.2: Radial system considered for the illustration of transfer coefficients
3
Reproduction of Fig. 2.5.
98
The positive sequence voltage ratio VV+:s−us can be expressed in a general form,
+:us
V+:s−us 1
V+:us ≈ (5.2)
Z −s
1 + j ++:tf ∠θ
Z++:s pf :s
where,
Z++:s - downstream positive sequence impedance (or equivalent) seen at the busbar S
Z++:tf −s
The impedance ratio Z++:s can be written as:
Z++:tf −s 1
Z++:s = ksc−s (5.3)
where,
Ssc−s
ksc−s = Ss
(Vn−s )2
Ssc−s = |Z++:tf −s |
, the short-circuit capacity (in MVA) at the busbar S
(Vn−s )2
Ss = |Z++:s |
, the total load (in MVA) supplied by the system S
Then, the positive sequence voltage ratio VV+:s−us can be rewritten as:
+:us
V+:s−us 1
V+:us ≈ (5.4)
1
1 + j ksc−s ∠θpf :s
4
− and + for lagging and leading conditions respectively.
99
The negative sequence voltage |V−:Uus /s−us | can be generally written, ignoring zero
where,
Z−−:tf −us - negative sequence impedance5 , referred to US, of the US-S coupling trans-
former
I−:Uus /tf - negative sequence current (referred to US) in the US-S coupling trans-
Based on the evidence from Chapter 3 (Section 3.2), the behaviour of the negative
sequence current I−:Uus /tf seem to be influenced by the load type/base supplied by
the system S making the transfer coefficient Tus−s dependant on the load type/base.
The IEC approach6 for assessing the MV to LV transfer coefficient accounts for this
influence, however it distinguishes only motor loads from passive loads, or in other
words it does not take the differences7 that exist between various passive load types
into account. This section addresses four basic load types8 and various load bases
When the system S supplies constant impedance loads (balanced), (5.5) can be re-
V
arranged such that the negative sequence voltage ratio −:UVus /s−us
−:us
is equal to the
5
Which is inherently equal to the positive sequence impedance Z++:tf −us of the transformer.
6
Given in Chapter 2 by (2.17).
7
As seen from Fig. 5.1.
8
i.e. constant impedance, constant current, constant power and three-phase induction motor
loads.
100
positive sequence voltage ratio given by (5.4), as the positive and negative sequence
impedances of the loads and the US-S coupling transformer are equal:
V−:Uus /s−us 1
V−:us = (5.6)
1
1 + j ksc−s ∠θpf :s
Tus−s = 1 (5.7)
That is, for constant impedance loads, the term Z++:tf −us I−:Uus /tf causes the down-
stream negative sequence voltage |V−:Uus /s−us | to be smaller than the upstream neg-
1
ative sequence voltage |V−:us | by the factor ˛
1
˛, leading to a unity Tus−s .
∠θpf :s ˛
˛ ˛
˛1+j k
sc−s
As in Section 3.2.2, the negative sequence current I−:Uus /tf can be considered to be
negligible when the system supplies constant current loads (balanced), as such loads
draw equal magnitudes of three phase currents regardless of the prevailing voltage
condition. Hence, (5.5) can be simplified for constant current loads as:
V−:Uus /s−us
V−:us ≈ 1 (5.8)
1
Tus−s ≈ 1 + j ∠θpf :s (5.9)
ksc−s
That is, in contrary to the case of constant impedance loads, the downstream negative
sequence voltage |V−:Uus /s−us | is equal to the upstream negative sequence voltage
101
As in Section 3.2.3, through careful examination of the results obtained using unbal-
anced load flow analysis, (5.10) is established as a close approximation to the negative
V
sequence voltage ratio −:UVus /s−us
−:us
:
V−:Uus /s−us 1
V−:us ≈ (5.10)
β
1
1 + j ksc−s ∠θpf :s
where, β ≈ −1 and −2 for low (∼ 0.9) and high (∼ 1) lagging pf conditions respec-
1−β
1
Tus−s ≈ 1 + j
∠θpf :s (5.11)
ksc−s
That is, for constant power loads, the term Z++:tf −us I−:Uus /tf causes the downstream
negative sequence voltage |V−:Uus /s−us | to be greater than the upstream negative se-
1
quence voltage |V−:us | by the factor ˛˛ ˛β , resulting in an increase in Tus−s
˛1+j k 1 ∠θpf :s ˛
˛
sc−s
1−β
1
by the factor 1 + j ksc−s ∠θpf :s relative to unity.
Discussion
Taking a generalised view, the scaling factors applied to the downstream negative
sequence voltage |V−:Uus /s−us | relative to the upstream negative sequence voltage
γ
1
|V−:us | can be expressed in a form 1 + j ksc−s ∠θpf :s for the three passive load types,
where γ = 1 for constant impedance loads, γ = 0 for constant current loads, and
−2 ≤ γ ≤ −1 for constant power loads. Note that this exponent γ, which distin-
102
guishes various load behaviours, is similar to that established in Section 3.2 in relation
for constant impedance loads, τ = 1 for constant current loads, and 2 ≤ τ ≤ 3 for
constant power loads. Noting that the factor ksc−s can be in the range of 59 to 2510
for various systems and the exponent τ can vary in the range of 0 to 3 for different
assumed in the IEC approach cannot be used to represent all load types with a high
τ
1
degree of accuracy. The accuracy of the formulation Tus−s ≈ 1 + j ksc−s ∠θpf :s will
When the system S supplies three-phase induction motors11 which can be repre-
sented using decoupled, unequal and constant12 sequence impedances, (5.5) can be
written as:
V−:Uus /s−us 1
V−:us = (5.12)
Z++:tf −s
1 + Z−−:s
where, Z−−:s - downstream negative sequence impedance seen at the busbar S. When
US and S represent MV and LV (i.e. s = lv) systems respectively (i.e. the case of
impedance Z−−:im of the aggregated motor load supplied by the aggregated LV sys-
tem13 . In the case of the HV to MV (i.e. s = mv) propagation, the impedance Z−−:mv
9
e.g. fully loaded 60MVA HV-MV transformer with 20% impedance.
10
e.g. fully loaded 400kVA MV-LV transformer with 4% impedance.
11
Which are usually supplied at the LV level.
12
For a given motor speed.
13
Note that the busbar DS of Fig. 2.5, which represents the downstream system of the system S,
does not exist in this case.
103
is not simply equal to Z−−:im due to the additional system impedance Z++:sys−lv 14
that exists between the MV busbar under evaluation and the downstream LV systems
at which motor loads are supplied. That is, transfer coefficients, in the presence of
motor loads, seem to depend on the systems in which the propagation is being con-
This section considers that the busbars US and S (Fig. 5.2) represent MV and LV
systems respectively (i.e. the subscripts ‘us’ and ‘s’ used in the above formulae (5.1)
Passive Loads
τ
1
As derived above, Tmv−lv ≈ 1 + j ksc−lv ∠θpf :lv for passive loads. Representing most
practical circumstances, the factor ksc−lv for LV systems can take a value in the
range of 1015 to 2516 . Figs. 5.3: I − II and 5.4: I − II illustrate the variation of
Tmv−lv with ksc−lv established using this formulation in comparison to the results
obtained using unbalanced load flow analysis for constant current and constant power
loads respectively, where sub-figures I and II correspond to 0.99 and 0.9 lagging pf
conditions respectively. These illustrate the accuracy of the new formulation, while
demonstrating the deviation of the actual transfer coefficient from the unity value
1.05 1.05
Tmv-lv
Tmv-lv
1.00 1.00
Figure 5.3: Variation of Tmv−lv with ksc−lv for constant current loads: I - 0.99 lagging
pf, II - 0.9 lagging pf
1.10 1.10
1.05 1.05
Tmv-lv
Tmv-lv
1.00 1.00
Load flow results Load flow results
0.95 0.95
Proposed formulation Proposed formulation
0.90 0.90
10 13 16 19 22 25 10 13 16 19 22 25
ksc-lv ksc-lv
Figure 5.4: Variation of Tmv−lv with ksc−lv for constant power loads: I - 0.99 lagging
pf, II - 0.9 lagging pf
105
Equation (5.12) is reproduced here for the propagation of the negative sequence volt-
V−:Umv /lv−mv 1
= (5.13)
V−:mv Z++:ml−lv
1 + Z−−:im
where, the subscript ‘ml’ in Z++:ml−lv is a replacement of the subscript ‘tf ’ used
inductive nature of the impedances Z++:ml−lv and Z−−:im , (5.13) can be rewritten as:
V−:Umv /lv−mv 1
= (5.14)
V−:mv ks
1 + ksc−lv
where, ks - ratio between the positive and negative (which is inductive) sequence
That is, for induction motor loads, the term Z++:ml−mv I−:Umv /ml causes the down-
stream negative sequence voltage |V−:Umv /lv−mv | to be smaller than the upstream neg-
1
ative sequence voltage |V−:mv | by the factor „
ks
«. Noting that 5 < ks < 7, this
1+ k
sc−lv
results in a Tmv−lv < 1. This reduction in Tmv−lv relative to unity is significant com-
pared to the increment in Tmv−lv caused by passive loads18 . Fig. 5.5 illustrates the
variation of Tmv−lv with ksc−lv , for motor loads with ks = 6.719 and pf = 0.9 lagging,
17
Typically, ks can be in the range of 5 to 7.
18
e.g. Tmv−lv = 0.81 for motor loads with ks = 6.7, ksc−lv = 25 and pf = 0.9 lagging, whereas
Tmv−lv = 1.04 for constant power loads with ksc = 25 and pf = 0.9 lagging.
19
e.g. 50hp motors described in Appendix B.
106
established using the IEC method20 and (5.15) in comparison to the results obtained
using unbalanced load flow analysis. This demonstrates that both the IEC method
and the proposed new formulation are satisfactory for estimating Tmv−lv for mo-
tor loads.
1.10
1.00
0.90
Tmv-lv
0.80
Figure 5.5: Variation of Tmv−lv with ksc−lv for induction motor loads with ks = 6.7
and pf = 0.9 lagging
Consider a mix of constant impedance, constant current, constant power and induc-
tion motor loads is supplied by the LV system. Then, the negative sequence current
where, I−:z , I−:i , I−:pq , I−:im - negative sequence currents (referred to MV side) in Z,
The impact of the components Z++:ml−mv I−:L for the four load elements can be com-
V−:Umv /lv−mv 1
≈ β (5.18)
V−:mv
kpq
kz k m ks
1 + j ksc−lv ∠θpf :z 1 + j ksc−lv ∠θpf :pq 1 +
ksc−lv
where,
km - ratio between the rated motor load (in MVA) and the total load (in MVA) sup-
kz , kpq - ratios of the constant impedance and constant power loads (in MVA) respec-
θpf :z , θpf :pq - power factor angles of the constant impedance and constant power loads
β
kz kpq km k s
The terms 1 + j ksc−lv ∠θpf :z , 1 + j ksc−lv ∠θpf :pq and 1 + in the numera-
ksc−lv
tor of (5.18) account for the influence of the constant impedance, constant power and
induction motor elements respectively in the mix on the propagation of the negative
sequence voltage. Note that these terms are modified versions of the respective terms
for the individual load types22 , where the modifications are introduced by multiply-
1
ing ksc−lv of each of the terms for the individual load types by the respective load
proportion kL (L = z, pq, im). Substituting (5.4) and (5.18) in (5.1), the transfer
21
See Appendix
L for details.
γ
22 1
i.e. 1 + j ksc−lv ∠θpf :lv with different γ values for the various passive loads types, and
ks
1 + ksc−lv for motor loads.
108
Figs. 5.6: I − II illustrate the variation of Tmv−lv with ksc−lv for two load bases
which are dominated by motor loads (Z − 10%, I − 5%, P Q − 15% and IM − 70%)
using the IEC method, the proposed formulation (5.19) and unbalanced load flow
analysis. A lagging pf of 0.9 for all load components and a ks = 6.7 for motor loads
are assumed. These results demonstrate that although the IEC method provides an
accurate estimation to Tmv−lv for load bases containing large proportions of induc-
tion motors, it associates a considerable degree of error for load bases dominated by
passive elements. Furthermore, the proposed new formulation gives a more accurate
0.85
1.00
Tmv-lv
Tmv-lv
0.80
Load flow results 0.95 Load flow results
0.75 Proposed formulation Proposed formulation
IEC method IEC method
0.70 0.90
10 13 16 19 22 25 10 13 16 19 22 25
ksc-lv
ksc-lv
Figure 5.6: Variation of Tmv−lv with ksc−lv : I - for a load base dominated by induction
motors, II - for a load base dominated by passive elements
109
Figs. 5.7: I−II illustrate the variation23 of Tmv−lv with km , considering load mixes
of constant impedance and motor loads (i.e. km = 1 − kz ) and constant power and
motor loads (i.e. km = 1 − kpq ) respectively, for two extreme cases where ksc−lv ≈ 10
and ksc−lv ≈ 25. A lagging pf of 0.9 for all load components and ks = 6.7 for motor
loads are assumed. These demonstrate that the transfer coefficient Tmv−lv can be in
0.9 0.9
Tmv-lv
Tmv-lv
0.8 0.8
0.6 0.6
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
km = 1-kz km = 1-kpq
Figure 5.7: Variation of Tmv−lv with km for ksc−lv ≈ 25 and ksc−lv ≈ 10: I - for load
mixes of Z and IM loads, II - for load mixes of P Q and IM loads
The generalised expression (5.19) can be simplified, easing the computation, for
industrial load bases which contain large proportions of motor loads by neglecting
the negative sequence current components I−:z , I−:i , and I−:pq compared to I−:im in
(5.16) as:
1
1 + j ksc−lv ∠θpf :lv
Tmv−lv ≈ (5.20)
1 + kksc−lv
m ks
Fig. 5.8 provides a comparison of the variations of Tmv−lv with km established using
the IEC method, the generalised expression (5.19), the simplified expression (5.20)
for industrial load bases, and unbalanced load flow analysis. Load mixes of constant
23
Derived using the proposed formulation.
110
power and motor loads, a lagging pf of 0.9, ks = 6.7 and ksc−lv ≈ 10 are considered.
This shows that the estimation from the various methods are in close agreement for
km values above 0.5. For lower km values, (5.19) gives the closest estimation to the
actual24 Tmv−lv , and (5.20) is seen to provide a better estimation compared to the
IEC method.
0.9
0.8
0.7
0.6
0.0 0.2 0.4 0.6 0.8 1.0
km
Figure 5.8: Variation of Tmv−lv with km established using the IEC method, (5.19),
(5.20) and unbalanced load flow analysis
This section considers that the busbars US, S and DS of Fig. 5.2 represent HV, MV
and LV25 systems respectively (i.e. the subscripts ‘us’ and ‘s’ in the above formulae
Passive Loads
τ
1
As derived above, Thv−mv ≈ 1 + j ksc−mv ∠θpf :mv for passive loads. Representing
most practical circumstances, the factor ksc−mv for MV systems can take a value in
the range of 526 to 1527 . Noting that the values of ksc−mv for MV systems are usually
smaller than the ksc−lv values for LV systems, the amplification which takes place in
Considering that the MV system under evaluation supplies only induction motors at
the aggregated LV busbar, (5.12) is reproduced here for the HV to MV propagation as:
V−:Uhv /mv−hv 1
= (5.21)
V−:hv
1 +
Z++:hm−mv
Z−−:mv
where, the subscript ‘hm’ in Z++:hm−mv is a replacement of the subscript ‘tf ’ used
the impedance Z−−:mv is equal to the sum of the impedances Z−−:im and Z++:sys−lv
(referred to MV), the rearrangement of (5.21) in terms of the load and system char-
acteristics gives :
V−:Uhv /mv−hv 1
= (5.22)
V−:hv
1
ks
1 + ksc−mv 1+ ks
ksc−lv
agg
26
e.g. fully loaded 60MVA transformer with 20% impedance.
27
e.g. fully loaded 10MVA transformer with 6.5% impedance.
112
where, ksc−lvagg - ratio between the short-circuit capacity28 (in MVA) at the aggregated
LV busbar, and the total load (in MVA) supplied by the LV system. Substitution of
That is, for induction motor loads, the term Z++:hm−hv I−:Uhv /hm causes the down-
stream negative sequence voltage |V−:Uhv /mv−hv | to be smaller than the upstream neg-
ks
1+ k ks > 1, a value less than unity can be expected for the transfer coefficient
sc−lvagg
passive loads29 . Note that the degree of this reduction in the HV to MV transfer
coefficient Thv−mv is lower than that in the HV to MV transfer coefficient Tmv−lv for
similar system and load characteristics30 . However, usually ksc−mv < ksc−lv , and thus
Consider that the MV system under evaluation supplies a mix of constant impedance,
constant current, constant power loads (at MV and/or LV) and induction motors (at
LV). Based on the same approach32 used to establish (5.18) in the case of the MV to
LV propagation, the various load behaviours in the above load mix can be combined33 ,
forming the resultant impact of the load mix on the HV to MV propagation, as:
V−:Uhv /mv−hv
≈
V−:hv
1 2 0 13
˛ ˛˛ ˛β “ ”
kzmv kpqmv kmmv ks
∠θpf :zmv ˛˛1+j k ∠θpf :pqmv ˛ 1+ k
˛ ˛˛ ˛ 4 @ A5
˛1+j k
sc−mv sc−mv sc−mv 1+ km ks
ksc−lv
agg
(5.24)
where,
kzmv , kpqmv , kmmv - ratios of the constant impedance, constant power and motor loads
(in MVA) respectively to the total load (in MVA) supplied by the MV system under
evaluation
θpf :zmv , θpf :pqmv - power factor angles of the constant impedance and constant power
Substituting (5.4) and (5.24) in (5.1), and replacing kmmv = km klv (where klv is
the fraction of LV loads supplied by the MV system under evaluation), the transfer
1
1 + j ∠θ
ksc−mv pf :mv
Thv−mv ≈ β
kzmv kpqmv km klv ks
1 + j ksc−mv ∠θpf :zmv 1 + j ksc−mv ∠θpf :pqmv 1 + ksc−mv
km ks
1+ ksc−lv
agg
(5.25)
Figs. 5.9: I - II and 5.10: I - II illustrate the variation of Thv−mv with klv for two
where the loads are supplied directly at the MV busbar34 . Sub-figures I and II cor-
respond to load mixes of constant impedance and motor loads, and constant power
and motor loads respectively. Each figure shows the variation established using (5.25)
in comparison to the result obtained using unbalanced load flow analysis for three
sub-cases where:
• km = 0.5 (i.e. LV system supplies equal proportions of passive and motor loads)
A case where ksc−lvagg = 2035 , pf = 0.9 lagging for all load components and ks = 6.7
Figs. 5.11: I - II illustrate the variation of Thv−mv with klv for the two cases where
ksc−mv = 12 and ksc−mv = 4 respectively considering systems where the LV loads are
supplied through MV lines36 . Each figure shows the variation for the three sub-cases
obtained using unbalanced load flow analysis. Passive loads are represented as a mix
of constant impedance and constant power elements with equal compositions. A case
34
LV loads are supplied through MV-LV transformers, i.e. Z++:sys−lv accounts for the impedances
of the MV-LV coupling transformers.
35
e.g. aggregation of fully loaded 1MVA transformers with 5% impedance.
36
i.e. Z++:sys−lv accounts for the impedances of the MV lines and the MV-LV coupling transform-
ers.
115
where ksc−lvagg = 637 , pf = 0.9 lagging for all load components and ks = 6.7 for
According to Figs. 5.9 - 5.11, the proposed formulation (5.25) provides an estima-
tion to the transfer coefficient Thv−mv , which is seen to be in close agreement with the
values given by unbalanced load flow analysis except for the minor discrepancies38
arise when a system supplies primarily constant power loads under heavy loading
or lower ksc−mv conditions. Further, these demonstrate that the transfer coefficient
Thv−mv can be a value in the range of 0.5 to 1.4, which is a wider range than that of
Tmv−lv (0.6 - 1.1), depending on the prevailing system and load characteristics and
1.2 I - Load mixes of Z and IM loads 1.2 II - Load mixes of PQ and IM loads
1.1 1.1
km = 0
km = 0.5
1.0 1.0 km = 0
km = 0.5
Thv-mv
Thv-mv
0.9 0.9
km = 1
0.8 0.8
Load flow results Load flow results
0.7 Proposed formulation 0.7
km = 1 Proposed formulation
0.6 0.6
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
klv klv
Figure 5.9: Variation of Thv−mv with klv for ksc−mv = 12 (loads are supplied directly
at the MV busbar): I - for load mixes of Z and IM loads, II - for load mixes of PQ
and IM loads
37
e.g. the MV lines described in Appendix B supplying 10MVA loads per line with 10% voltage
regulation through fully loaded 1MVA transformers with 5% impedance and 1.1pu secondary tap
setting.
38
10% error in the case considered in Fig. 5.10: II.
116
Thv-mv
Thv-lv
1.1 km = 0.5 km = 0
0.9 km = 0.5 0.9
0.7 0.7 km = 1
Figure 5.10: Variation of Thv−mv with klv for ksc−mv = 4 (loads are supplied directly
at the MV busbar): I - for load mixes of Z and IM loads, II - for load mixes of PQ
and IM loads
1.0 km = 0.5
0.9
0.9
0.8 0.8 km = 1
km = 1 0.7
Load flow results Load flow results
0.7 0.6
Proposed formulation Proposed formulation
0.6 0.5
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
klv klv
Figure 5.11: Variation of Thv−mv with klv (LV loads are supplied through MV lines):
I - for ksc−mv = 12, II - for ksc−mv = 4
117
on Load Types/Bases
Consider the radial MV-LV system shown in Fig. 5.1239 where the voltage at the
sending end busbar (labelled ‘send’) of the line is taken as unbalanced. The purpose
is to assess the voltage unbalance that propagates from the sending end busbar to
the receiving end busbar (labelled ‘rec’) of the MV sub-system, or the influence
coefficient ksend−rec between the busbars send and rec. For this, the loads Smv:rec−local
send
MV line
(t)
rec
Smv:rec-local (MVA)
Downstream
LV system
Smv:rec-ds (MVA)
39
Reproduction of Fig. 3.2.
118
Based on the definition40 , the influence coefficient ksend−rec can be expressed as:
V−:Usend /rec
ksend−rec = (5.26)
V−:send
where,
V−:Usend /rec - negative sequence voltage at the receiving end busbar that propagates
V−:Usend /rec = V−:send − Z++:t I−:Usend /t (5.27)
V−:send - negative sequence voltage that exists at the sending end busbar
I−:Usend /t - negative sequence current in the line, which arises as a result of the voltage
fer coefficients, where the impedance Z−−:tf −us (= Z++:tf −us ) of the US-S coupling
transformer is replaced here with Z++:t of the line. Thus, as in the case of transfer co-
V send /rec
efficients, the negative sequence voltage ratio −:U
V−:send
, i.e. the influence coefficient
V−:Usend /rec 1
V−:send = ksend−rec ≈ (5.28)
γ
Z++:t
1 + Z++:rec
where,
busbar rec
40
See Section 5.1.
41
Ignoring zero sequence unbalance, and replacing the negative sequence impedance Z−−:t of the
line with the positive sequence impedance Z++:t .
119
1 for constant impedance loads
0 for constant current loads
γ=
β for constant power loads, where β = −1 and − 2 for low (∼ 0.9) and
high (∼ 1) lagging pf conditions respectively
ksend−rec ≈ (1 − V Rt )γ (5.29)
where, V Rt - voltage regulation of the line42 . That is, the influence coefficient ksend−rec
is equal to unity, smaller than unity by the factor (1−V Rt ), and greater than unity by
the factor (1 − V Rt )β for constant current, constant impedance, and constant power
loads respectively. In other words, the term Z++:t I−:Usend /t causes the receiving end
negative sequence voltage V−:Usend /rec to deviate away from the sending end negative
sequence voltage |V−:send | by the factor (1 − V Rt )γ for passive loads. Note that an
identical behaviour was observed in Chapter 3 (Section 3.2) for the negative sequence
current or the term Z++:t I−:t/t which arises as a result of an asymmetrical line, where
t
the term Z++:t I−:t/t was seen to cause the negative sequence voltage |V−:g/mv:rec | to
deviate away from the principal term |Z−+:t I+:t | by the factor (1 − V Rt )γ for passive
loads. Employing this similarity, for three-phase induction motor loads supplied at
the LV level43 , the influence of the term Z++:t I−:send/t on ksend−rec can be written44 as:
1
ksend−rec ≈ (5.30)
V Rt 1
1+ 1−V Rt 1
+k 1
ks sc−lvagg
42
Voltage regulation is defined as the ratio between the positive sequence voltage drop across the
line (i.e. Z++:t I+:t ), and the sending end positive sequence voltage.
43
i.e. Smv:rec−local = 0.
44
Refer to (3.9).
120
unity (in other words I−:Usend /t ≈ 0), resulting in a unity ksend−rec for passive loads
in general. However, the influence coefficient ksend−rec for induction motor loads is
considerably smaller than unity45 , implying that the impact of the negative sequence
the case of passive loads. Based on this, ksend−rec for a mix of passive (at MV and/or
LV) and motor (at LV) loads can be expressed46 in a generalised form as:
1
ksend−rec ≈ (5.31)
V Rt klv
1+ 1−V Rt 1
+k 1
ks km sc−lvagg
Fig. 5.13 illustrates the variation of ksend−rec with km established using (5.31) in
comparison to the results obtained using unbalanced load flow analysis for the test
• klv = 0.5
• klv = 0 (i.e. Smv:rec−ds = 0, which implies that no motor loads are supplied by
the MV line)
The results presented in Fig. 5.13 correspond to a selected operating scenario where
8.5%. Fig. 5.13 confirms the above basis given by (5.31) which describes the be-
proportion.
45
e.g. 0.6 for ksc−lvagg = 20, ks = 6.7 and V Rt = 10%.
46
Refer to (3.14).
47
Where ksc−lvagg ≈ 19, ks = 6.7, passive load composition - equal proportions of constant
impedance and constant power elements, and the MV line is taken as ideally transposed.
121
1.0
klv=0.5
0.9
ksend-rec
0.8
Figure 5.13: Variation of ksend−rec with km for the cases where klv = 1, klv = 0.5 and
klv = 0
Consider the network shown in Fig. 5.1448 . For the purpose of assessing the voltage
unbalance that propagates from any busbar i to other busbars 1, 2, ..., n, i.e. influence
coefficients ki−1 , ki−2 , ..., ki−n , the voltage at the upstream system, all loads supplied
by the system, and all lines in the network49 are considered to be balanced.
Based on a similar approach used to establish (3.21) in Chapter 3 (Section 3.3), the
can be written in terms of the nodal negative sequence admittances Y−−:xy 51 and
nodal negative sequence voltages52 V−:Ui /x at the busbars x, and the nodal negative
V−:Ui /1
−I−:Ui /1 Y Y++:12 . . . Y++:1i . . . Y++:1n V
−:Ui /2
++:11
..
−I−:U /2 Y++:21 Y++:22 . . . Y++:2i . . . Y++:2n
.
i
=
.
.. .. .. .. .. .. ..
. . . . . . V−:i
..
−I−:Ui /n Y++:n1 Y++:n2 . . . Y++:ni . . . Y++:nn
.
V−:Ui /n
(5.32)
Busbar 2
Upstream
system Busbar 1 Busbar x Busbar n
… …
Busbar 3
Sub-system S
−I−:Ui /1 Y Y++:12 . . . Y++:1n V Y
++:11 −:Ui /1 ++:1i
−I−:U /2 Y++:21 Y++:22 . . . Y++:2n V−:U /2 Y++:2i
i i
= + V
−:i
.. .. .. .. .. .. ..
. . . . . . .
−I−:Ui /n Y++:n1 Y++:n2 . . . Y++:nn V−:Ui /n Y++:ni
(5.33)
the negative sequence currents [I−:Ui ] on the negative sequence voltages [V−:Ui ] should
loads. Thus, the current I−:Ui /x at any busbar x can be generally written when the
busbar supplies a mix of passive (at HV, MV and/or LV) and motor (at LV) loads as:
taking only induction motors into account. As discussed in Chapters 3 (Section 3.3)
and 4 (Section 4.3) with regard to MV and HV systems respectively, this admittance
Y−−:x−im is inductive in nature, and dependant on the system and load characteris-
taking the network shown in Fig. 5.14 as an MV network with Fig. 5.15 representing
the system54 supplied by any busbar x, the admittance Y−−:x−im can be expressed as
54
Where the downstream system represents an aggregated LV system.
124
]−1
0
[ki−x ](n−1)×1 ≈ [Y++:xz [Y ] (5.37)
(n−1)×(n−1) ++:xi (n−1)×1
where,
0
Y++:xz ≈ Y++:xz + Y−−:x−im for x = z
0
Y++:xz = Y++:xz for x 6= z
Busbar x
Smv:x-local
Downstream
system supplied
by the busbar x
Smv:x-ds
force the negative sequence voltage at the connected busbars to be zero disregarding
the existence of sources of unbalance. Equation (5.37) which gives the influence
55
Refer to Chapter 3 for the definitions of the symbols.
125
coefficients ki−x between busbar i and other neighbouring busbars x does not consider
the presence of such components, and thus requires suitable adjustments such that
the impact of zero voltage unbalance (or of voltage controlled components) at given
taken in Chapter 4, the matrix equation (5.37) can be modified to incorporate the
by56 :
0
• Reducing the dimension of the matrix [Y++:xz ] down to (n − 2) × (n − 2) by
the j th row.
System
three-wire) shown in Fig. 5.1657 for evaluating the voltage unbalance which propa-
gates from busbar 1 to busbars 2 and 3, i.e. influence coefficients k1−x for x = 2, 3.
Lengths of the lines which are taken as identical in construction58 and ideally trans-
posed are shown alongside the lines. The positive sequence admittance per km of the
with equal compositions of constant impedance and constant power elements. That
is, km:x = 0 implying that Y−−:x−im ≈ 0 for x = 3. Busbar 2 supplies a mix (4MVA
56
Note that there is only (n − 2) number of influence coefficients are to be determined as the
influence coefficient ki−j is known to be zero.
57
Reproduction of Fig. 3.7.
58
See Appendix B for further details.
126
at 0.9 lagging pf) of passive loads59 and induction motors (ks = 6.760 ) at the LV
level, which accounts for 40% of the total load supplied by the system. Note that
Fig. 5.17 illustrates the variations of k1−2 and k1−3 with km:2 established using the
flow analysis, demonstrating the accuracy of the proposed technique. Further, these
results reveal that motor loads help reducing the voltage unbalance that propagates
between neighboring busbars compared to the case where only passive loads exist.
System
The proposed methodology is further applied to the IEEE 14-bus test system shown
in Fig. 5.18 which consists of positive sequence voltage controlled busbars (busbars
1, 2, 3, 6 and 8), taking it as a 66kV, 60Hz and three-wire network supplying constant
power loads at the HV level. System and line62 data are given in Appendix K.
Fig. 5.19 illustrates the influence coefficients ki−x where i = 4, x = 1−14 and x 6=
463 established using the proposed methodology in comparison to the results obtained
using unbalanced load flow analysis, further validating the proposed technique.
59
Which are represented using a mix of constant impedance and constant power elements with
equal compositions.
60
See Appendix B for further details.
61
Application of the methodology to the test system is described in Appendix M.
62
Which are taken as identical in construction as described in Appendix H, and ideally transposed.
The positive sequence admittance per km of the lines = (0.2729 + j1.0244) × 102 pu (based on a
100M V A base).
63
i.e. the propagation of voltage unbalance from busbar 4 to other busbars.
127
204A
MV busbar 2
(0.98pu, -6.740) LV (460V)
4MVA
0.9 lagging pf
10km
Upstream HV 135A
(66kV) system MV busbar 1
(1.05pu, 00) (1.05pu, -4.070)
10km
59A
5km
152A
4MVA
0.9 lagging pf
2MVA
0.9 lagging pf
MV busbar 3
(1.01pu, -5.570)
HV-MV coupling transformer – 12MVA, winding resistance = 1%, leakage reactance = 10%,
secondary tap setting = 1.05pu
MV-LV coupling transformer – aggregated representation of fully loaded 1MVA transformers with
winding resistance = 1%, leakage reactance = 5% and secondary tap setting = 1.05pu
Figure 5.16: Three-bus MV test system considered for applying the proposed method-
ology (reproduction of Fig. 3.7)
1.1
k1-3
1.0
0.9
k1-2, k1-3
0.8
0.7 k1-2
Load flow results
0.6
Methodology
0.5
0.0 0.3 0.5 0.8 1.0
km:2
Figure 5.17: Variations of k1−2 and k1−3 with km:2 for the three-bus MV test system
128
0.7
Load flow results
0.6
Methodology
0.5
0.4
k4-x
0.3
0.2
0.1
0.0
1 2 3 5 6 7 8 9 10 11 12 13 14
Busbar x
Figure 5.19: Influence coefficients k4−x (x = 1 − 14, x 6= 4) for the IEEE 14-bus test
system
129
This chapter has addressed the propagation of voltage unbalance. This is a key aspect
and LV public power systems essentially based on the IEC/TR 61000-3-13 recommen-
dations. This propagation from higher voltage to lower voltage systems in terms of
transfer coefficients (Section 5.2), and from one busbar to other neighbouring busbars
Theoretical bases which describe the behaviour exhibited by four basic load types
with regard to transfer coefficients and influence coefficients have been developed.
power loads respectively. Noting that the factor ksc−s can be a value in the
range of 5 to 25 for various systems and the exponent τ varies in the range of 0
• Transfer coefficients, for induction motor loads, have to be scaled down by the
1 01
factors „ « and 2 13 relative to unity in the cases of
1+ k ks
“ ”
41+ 1 ks
sc−lv ksc−mv
@
ks
A5
1+
ksc−lv
agg
the MV to LV and HV to MV propagation respectively. Noting that 5 < ks < 7,
by passive loads.
130
coefficients can be considerably smaller than unity when the network supplies
is that the negative sequence currents which arise as a result of the voltage
cients, and influence coefficients for interconnected network environments have been
developed. These have been verified using unbalanced load flow analysis. In summary:
• It has been demonstrated that the proposed new formulation for assessing the
for load bases which are dominated by passive elements, compared to the IEC
method.
of 0.6 to 1.1 and 0.5 to 1.4 respectively depending on the system and load
been undertaken employing a three-bus MV test system and also the IEEE
6.1 Introduction
The IEC approach of managing continuous power quality disturbances (e.g. har-
monics, flicker and voltage unbalance) through the allocation of emission limits to
set planning levels, which has been identified with regard to harmonics and flicker1 ,
is an anticipated problem with the new IEC/TR 61000-3-13 voltage unbalance al-
location aproach as well. This chapter examines the IEC/TR 61000-3-13 procedure
employing a simple three-bus HV test system in Section 6.2. The principles of the
constraint bus voltage (CBV) method, that was discussed in Section 2.9 of Chap-
ter 2 as an alternative approach for harmonics and flicker allocation, are introduced
to voltage unbalance so that a robust allocation technique which closely aligns with
1
Refer to Section 2.9 of Chapter 2.
131
132
IEC/TR 61000-3-13 is developed in Section 6.3 of this chapter. Section 6.4 examines
this revised allocation technique using the above mentioned test system. A summary
Consider the three-bus HV system (60Hz, 66kV, three-wire) shown in Fig. 6.1. The
system supplies constant power loads at the HV level. Considered operating scenario
and resulting positive sequence system conditions2 are also indicated in Fig. 6.1.
Lengths of the lines which are taken as identical in construction (including the phase
positioning) are shown alongside the lines. The relevant admittance data (per km)
(0.1040 + j0.1779)Skm
Chapters 3 - 5.
• Derivation of the busbar emission levels, which result in when all individual
installations are injecting at their allocated limits, using the general summa-
tion law.
2
Nodal voltages and line currents, which are obtained using load flow analysis.
3
Refer to Appendix H for further details.
133
HV busbar 2
(0.992pu, -10.840)
20MVA
0.95 lagging pf
50km
Upstream EHV 112A
(230kV) system HV busbar 1
(1.100pu, 00) (1.026pu, -7.980)
40km
64A
20km
151A
20MVA
0.95 lagging pf
10MVA
0.95 lagging pf
HV busbar 3
(1.007pu, -9.510)
Figure 6.1: Three-bus HV test system considered for examining the IEC/TR 61000-
3-13 approach
134
The procedure of the calculation of the emission limits to the three aggregated loads
The voltage at the upstream system of the HV network under evaluation is taken as
balanced. That is, the upstream contribution to voltage unbalance in the considered
HV system is zero, resulting in a global emission allowance Ug/hv which is equal to the
Referring to (2.10) and (2.11), the derivation of the busbar emission allowances Ug/hv:x
requires the initial evaluation of influence coefficients. A method for estimating influ-
ence coefficients was proposed in Section 5.3 of Chapter 5 as given by (5.37). Table 6.1
gives these influence coefficients for the considered test system, which are derived us-
ing the proposed method4 . Fig. 6.2 gives a comparison of these values with those
obtained using unbalanced load flow analysis. Arising from these values, an inter-
esting fact to note is that influence coefficients do not essentially hold the reciprocal
relationship ki−x = kx−i implying that the propagation from busbar i to busbar x is
The total apparent power Shv:x supplied by any busbar x, the total available
apparent power Shv:x−total of the entire sub-system as seen at the busbar x which is
4
As the test system supplies constant power loads at all three busbars (i.e. km:x = 0 implying
0
that Y−−:x−im ≈ 0 for x = 1, 2, 3), the admittances Y++:xz ≈ Y++:xz not only for x 6= z but also for
x = z.
135
Table 6.1: Influence coefficients for the test system shown in Fig. 6.1
k1−2 k1−3 k2−1 k2−3 k3−1 k3−2
1 1 0.57 0.71 0.69 0.86
1.2
Load flow results
1.0 Methodology
0.8
Value
0.6
0.4
0.2
0.0
k1-2 k1-3 k2-1 k2-3 k3-1 k3-2
Influence coefficient
Figure 6.2: A comparison of the influence coefficients for the test system derived using
the proposed method: (5.37), and unbalanced load flow analysis
Table 6.2: Shv:x , Shv:x−total and Ug/hv:x for the test system shown in Fig. 6.1
Busbar (x) Shv:x (MVA) Shv:x−total (MVA) Ug/hv:x (%)
1 20 38.2 0.88
2 20 48.6 0.74
3 10 44.2 0.48
136
derived using (2.10), and the busbar emission allowance Ug/hv:x at the busbar x that is
calculated using (2.11) with a summation law exponent of 1.45 are given in Table 6.2
for each of the three busbars. Note that, as the contributions from the neighboring
busbars 1 and 3 to voltage unbalance at busbar 2 (k1−2 = 1 and k3−2 = 0.86) are seen
to be greater than that at busbar 1 (k2−1 = 0.57 and k3−1 = 0.69), the allocation
approach allows a lower level of emission for busbar 2 than that for busbar 1 although
Two cases are considered for examining the IEC/TR 61000-3-13 approach:
• Case 1 - all HV lines are ideally transposed. That is, the contribution from
K 0 uex = 06 or Kuex = 17 for all busbars. This leads to voltage unbalance allo-
[2, 3], where the total busbar allowance Ug/hv:x can be allocated to installations.
• Case 2 - the line between busbars 1 and 3 is ideally transposed, and the other
two lines are untransposed. That is, system inherent asymmetries make some
contribution to the global emission levels resulting in a K 0 uex > 0 or Kuex < 1
for some or all busbars, implying that only a fraction of the busbar emission
in Section 4.3 of Chapter 4 as given by (4.16). These emissions for the test
5
Which is the indicative value given in IEC/TR 61000-3-13.
6
K 0 uex is the factor which accounts for the emission arising as a result of system inherent asym-
metries.
7
Kuex is the factor which represents the fraction of the busbar emission allowance that can be
allocated to installations.
137
system which are derived using the proposed method8 , the K 0 uex factors which
are calculated using (2.14) with the above busbar allowances Ug/hv:x (Table 6.2),
and the Kuex factors (= 1 − K 0 uex ) are given in Table 6.3. Fig. 6.3 gives a
comparison of these K 0 uex values with those obtained using unbalanced load
flow analysis.
The busbar emission limits Ehv:x 9 or the emission limits to the three aggregated
loads, which are derived using (2.15) with the above busbar allowances Ug/hv:x (Table
6.2) and Kuex factors (unity for Case 1, and as given in Table 6.3 for Case 2), are
given in Table 6.4 for Cases 1 and 2. Note that, in Case 2, only 70% of the busbar 2
allowance can be allocated to the connected load as the emission arising as a result
lines
Table 6.3: Ug/hv:x , K 0 uex and Kuex for Case 2 of the test system shown in Fig. 6.1
lines
Busbar (x) Ug/hv:x (%) K 0 uex Kuex
1 0 0 1
2 0.39 0.41 0.59
3 0.05 0.04 0.96
Table 6.4: Ehv:x according to IEC/TR 61000-3-13 for the test system shown in Fig. 6.1
Busbar (x) Ehv:x for Case 1 (%) Ehv:x for Case 2 (%)
1 0.88 0.88
2 0.74 0.51
3 0.48 0.47
8
As the test system supplies constant power loads at all three busbars (i.e. km:x = 0 implying
0 0
that Y−−:x−im ≈ 0 for x = 1, 2, 3), the admittances Y++:xy ≈ Y++:xy and Y−+:xy ≈ Y−+:xy not only
for x 6= y but also for x = y.
9
i.e. the combined emission limit of a load of which the agreed apparent power is equal to the
total apparent power Shv:x supplied by the busbar.
138
0.5
Load flow results
Methodology
0.4
0.3
K'uex
0.2
0.1
0.0
1 2 3
Busbar (x)
Figure 6.3: A comparison of the K 0 uex factors for the test system derived using the
proposed method: (4.16), and unbalanced load flow analysis
the imposed limit while making some contributions to voltage unbalance at other
limit and the influence coefficient between the two busbars. Thus, when all instal-
lations supplied by the system are injecting at their individual limits, the resulting
reult
global emission level Ug/s:x at any busbar x of the sub-system can be expressed using
the general summation law, also taking the global emission which arises as a result
reult
(k1−x Es:1 )α + (k2−x Es:2 )α + ... + (Es:x )α + ... + (kn−x Es:n )α
Ug/s:x =
lines α 1/α
+ (Ug/s:x ) (6.1)
10
Refer to the definition of the influence coefficient given in Section 2.8.3 of Chapter 2.
139
The busbar emission levels for the test system, which result in when the three
loads are injecting at the above allocated limits (Table 6.4), are derived using (6.1)
reult
Table 6.5: Ug/hv:x arising as a result of the IEC/TR 61000-3-13 allocation procedure
for the test system shown in Fig. 6.1
reult reult
Busbar (x) Ug/hv:x for Case 1 (%) Ug/hv:x for Case 2 (%)
1 1.24 1.15
2 1.52 1.51
3 1.40 1.30
Note that although the resulting emission levels at busbars 1 and 3 are below the set
planning level of 1.4%, this at busbar 2 exceeds the planning level by approximately
8% in each of the cases. This indicates, as anticipated, that the IEC allocation policy
is unlikely to be robust enough to comply with one of the key allocation objectives not
only in the cases of harmonics and flicker but also with regard to voltage unbalance.
The CBV allocation method which has been suggested for harmonics and flicker11
cannot be applied in its present form to voltage unbalance, as the case of voltage
system inherent asymmetries. Due to the presence of this additional emission, the
total busbar emission allowance Ug/s:x cannot be allocated to installations. That is,
in the case of voltage unbalance, the busbar emission limit Es:x 12 is usually smaller
11
Refer to Section 2.9.
12
Es:x of any busbar x is the emission limit of the load of power Ss:x .
140
than the busbar emission allowance Ug/s:x 13 , whereas Es:x = Ug/s:x for the cases of
harmonics and flicker. Thus, appropriate revisions addressing this issue are required
Closely following the IEC/TR 61000-3-13 approach, in which the busbar emis-
bar emission limit under the CBV harmonics/flicker allocation policy is extended to
p
α
Ug/s:x = ka Ss:x (6.2)
3-1315 , the emission which arises as a result of system inherent asymmetries can be
accounted for by using the factor Kuex (= 1 − K 0 uex )16 in determining the busbar
p
α
p
α
Es:x = Kuex Ug/s:x = ka Kuex Ss:x (6.3)
The factor Kuex can be expressed17 for the new allocation policy as a function of the
lines
!α lines α
Ug/s:x (Ug/s:x )
Kuex = 1 − =1− (6.4)
Ug/s:x kaα Ss:x
13
Ug/s:x for any busbar x accounts for the emissions which arise at the busbar as a result of both
the load of power Ss:x and line asymmetries.
14
See (2.11).
15
See (2.15).
16
The factors Kuex and K 0 uex√are defined by (2.13) and (2.14) respectively.
17
i.e. substituting Ug/s:x = ka α Ss:x .
141
q
lines α
Es:x == α
kaα Ss:x − (Ug/s:x ) (6.5)
at network busbars do not exceed the set planning level, the allocation constant ka
result
Ug/s:x ≤ Ug/s for every busbar x (6.6)
result
where, the resulting emission level Ug/s:x at any busbar x when all installations
are injecting at their individual limits can be written as given by (6.1) which can
result
kaα k1−x
α α α
Ug/s:x = S1 + k2−x S2 + ... + Sx + ... + kn−x Sn −
1/α
lines α lines α lines α
(Ug/s:1 ) + (Ug/s:2 ) + ... + (Ug/s:n ) (6.7)
vu n
lines α
X
α
u (Ug/s ) + Ug/s:x
u
u
α i=1,i6=x
ka = min u (6.8)
u
n
u X
α
t Ss:x + ki−x Ss:i
i=1,i6=x
the emission limit to any customer installation j of which the MVA demand is Ss:x−j
p
α
Es:x−j = ka Kuex Ss:x−j (6.9)
where the allocation constant ka is determined using (6.8), and the Kuex factor is
tion Technique
The same three-bus HV test system shown in Fig. 6.1 is employed for examining the
The calculation procedure, under the revised allocation policy, of the emission limits
to the three aggregated loads supplied by the test system is described in the follow-
ing steps:
Allocation Constant ka
Table 6.6 gives the values18 , corresponding to each of the three busbars, of the right
hand side (RHS) of (6.8) in relation to the two cases listed above. Choosing the
smallest value of the RHS of (6.8), which corresponds to busbar 2 in each of the
cases, the allocation constant ka for the two cases can be identified as given in Table
6.7. That is, the allocation constant is chosen such that the resulting emission level
Table 6.6: Values of the RHS of (6.8) in relation to the test system shown in Fig. 6.1
Busbar (x) RHS of (6.8) RHS of (6.8)
for Case 1 for Case 2
1 0.110 0.124
2 0.088 0.089
3 0.096 0.108
The Kuex factors calculated20 using (6.4), and the busbar emission limits Ehv:x or
the emission limits to the three aggregated loads derived21 from (6.9) for the two
different cases are given in Table 6.8. Figs. 6.4: I - II provides a comparison of the
individual emission limits derived according to IEC/TR 61000-3-13 and the revised
allocation method for Cases 1 and 2 respectively. This illustrates that the allocated
limits under IEC/TR 61000-3-13 are greater, specially at busbar 1, than that under
Table 6.8: Kuex and Ehv:x according to the revised allocation method for the test
system shown in Fig. 6.1
Busbar Kuex for Kuex for Ehv:x for Ehv:x for
(x) Case 1 Case 2 Case 1 (%) Case 2 (%)
1 1 1 0.75 0.76
2 1 0.60 0.75 0.53
3 1 0.96 0.46 0.45
20
α = 1.4, allocation constant ka as given in Table 6.7, apparent power values Shv:x as given in
lines
Table 6.2, and emissions Ug/hv:x = 0 for Case 1 and as given in Table 6.3 for Case 2 are used in the
calculation.
21
α = 1.4, allocation constant ka as given in Table 6.7, apparent power values Shv:x as given in
Table 6.2, and the Kuex factors as given in column 1 of this table are used in the calculation.
144
Ehv:x (%)
Ehv:x (%)
0.6 0.6
0.4 0.4
0.2 0.2
0.0 0.0
1 2 3 1 2 3
Busbar (x) Busbar (x)
Figure 6.4: Comparison of the busbar emission limits Ehv:x derived according to
IEC/TR 61000-3-13 and the revised method for the test system: I - for Case 1,
II - for Case 2
The busbar emission levels which result in when the three loads are injecting at the
above allocated limits (Table 6.8) are derived22 using (6.1) for the two cases, and given
in Table 6.9. Figs. 6.5: I - II provides a comparison of the resulting busbar emission
levels under IEC/TR 61000-3-13 and the revised allocation procedure for Cases 1
and 2 respectively. This illustrates that the resulting busbar emission levels under
IEC/TR 61000-3-13 are greater, at all three busbars, than that under the revised
method.
reult
Table 6.9: Ug/hv:x arising as a result of the revised allocation procedure for the test
system shown in Fig. 6.1
reult reult
Busbar (x) Ug/hv:x for Case 1 (%) Ug/hv:x for Case 2 (%)
1 1.12 1.04
2 1.40 1.40
3 1.28 1.18
22 lines
α = 1.4, influence coefficients as given in Table 6.1, and emissions Ug/hv:x = 0 for Case 1 and
as given in Table 6.3 for Case 2 are used in the calculation.
145
/ hv:x (%)
1.2 1.2
U glines
U glines
0.8 0.8
0.4 0.4
0.0
0.0
1 2 3
1 2 3
Busbar (x) Busbar (x)
reult
Figure 6.5: Comparison of the resulting emission levels Ug/hv:x derived according to
IEC/TR 61000-3-13 and the revised method for the test system: I - for Case 1,
II - for Case 2
As expected, the revised allocation technique restricts the resulting emission level
at the constrained busbar (e.g. busbar 2 of the test system) to the allowed global
emission level (= 1.4% for the test case) while maintaining the emission levels at other
busbars (e.g. busbars 1 and 3 of the test system) below the set limit. That is, this
proposed alternative technique allows a robust allocation in the sense that it satisfies
• The allocation constant ka determined using (6.8) ensures that the resulting
emission levels at all network busbars are maintained at or below the set plan-
ning level.
• The revised allocation policy given by (6.9) ensures that customer installations
busbars) receive identical emission limits, and also that larger customer instal-
lations (in MVA demand terms) to receive larger emission levels than smaller
installations.
23
See Section 2.9.
146
• In determining the allocation constant ka using (6.8), the resulting emission level
at a busbar is constrained at the allowed global emission level. This implies that
This chapter has examined the recently introduced IEC/TR 61000-3-13 voltage unbal-
ance allocation approach. It was seen, as in the case of the counterpart IEC harmonics
and flicker allocation methods, the prescribed voltage unbalance allocation procedure
also leads to planning levels are being exceeded even when no customer exceeds the
allocated limit.
Closely aligning with the IEC/TR 61000-3-13 guidelines, a revised voltage un-
whereby the emission levels at network busbars are explicitly forced to be at or below
the set planning levels when all loads inject their limits derived under the new ap-
proach, has been proposed. This revised allocation technique was seen to restrict the
resulting emission level at the constrained busbar to the allowed global emission level
while maintaining the emission levels at other busbars below the set limit. Further,
this proposed alternative technique allows a robust allocation in the sense that it
Unbalance in Interconnected
Power Systems
7.1 Introduction
works has continued to be a power quality issue of concern primarily due to difficul-
referred to as ‘study system’ hereinafter) shown in Fig. 7.1, which is owned and
mission system at S1 (bulk supply point: BSP) where the level of voltage unbalance
has been measured to be negligible. The system is divided into three sub-parts: up-
stream (US), central part (CP) and downstream (DS) for convenience in presenting.
Some of the sub-transmission lines are longer than 50km, and are not systematically
147
148
A
C B
D F
ZS2
ZS3
E
ZS4
I G
N
ZS6
PV generator :
ZS5 operates
Local only in limited time periods
generator:
J K operates only in limited time periods
L
Voltage regulators
ZS8 Loads
Capacitor bank s
ZS9
1.6
0.8
0.4
0.0
S2 S3 S4 S5 S6 S7 S8 S9
Busbar
Figure 7.2: Measured nodal VUF values for the study system
Victorian electricity distribution code2 [73], the measured voltage unbalance levels3
during the peak demand periods at S6, S7 and S8 have been seen to exceed the limit
in addition to significant levels (0.6% - 0.8%) at the upstream busbars S2 and S4.
These measured values corresponding to a selected time stamp4 that lies within the
system peak are illustrated in Fig. 7.2. Although effort has been put to address
have been noted indicating that asymmetries associated with the sub-transmission
lines can play a vital role in leading up to these excessive voltage unbalance levels.
• To carry out deterministic studies5 which are required to develop an insight into
ance levels, and transposition options that can effectively reduce the emission
allocation methodologies.
Sections 7.2 and 7.3 present deterministic studies and outcomes, together with
the verification of theoretical findings, in relation to line and load asymmetries re-
spectively. The overall system behaviour or the combined influence of line and load
The impact of the interaction of all asymmetrical lines of the study system on the
problem of voltage unbalance is established in terms of the VUF values at the various
busbars using unbalanced load flow analysis. This is accomplished by synthesising the
system operation6 , however under balanced loading7 conditions. Fig. 7.3 illustrates
5
These studies are supported by the developed unbalanced load flow program which is described
in Appendix O.
6
As revealed by measurements, voltage at the BSP is taken as balanced. This eliminates the
contribution made by the upstream EHV system to voltage unbalance in the study system.
7
Constant power loads are assumed.
151
these results in comparison to the measured voltage unbalance levels8 (Fig. 7.2)
with respect to the selected time stamp. These results demonstrate that the line
located in the central part (S6 and S7 - 1%) and in the downstream part (S8 and S9 -
1.4%) of the system, in addition to the considerable levels at the upstream busbars (S2
and S4 - 0.5%). This indicates the importance of proper line transposition practices
practice. However, the transposition of each single line is not economically viable
and/or practically feasible, and thus further knowledge is required in order to assist
2.0
Line asymmetries
Measured values
1.6
VUF (%)
1.2
0.8
0.4
0.0
S2 S3 S4 S5 S6 S7 S8 S9
Busbar
Figure 7.3: Nodal VUF values (load flow results) which arise as a result of the line
asymmetries, in comparison to the measured values
8
Note that these measured levels arise as a result of the interaction of both the line and load
asymmetries which exist in the actual system.
152
line9 which is energised by balanced supply voltages, as evident from Chapters 3 - 410 ,
t
the negative sequence voltage V−:rec at the receiving end busbar of the line (t) can be
t
V−:rec ≈ −Z−+:t I+:t (7.1)
where,
t
That is, the magnitude of V−:rec is given by the term |Z−+:t I+:t |. Thus, for a given
line, as illustrated in Fig. 7.4 which shows the results obtained using unbalanced
load flow analysis for each of the lines12 of the study system, the variation of |V−:rec
t
|
with |I+:t | is approximately linear with a gradient that is being equal to |Z−+:t |. In
t
addition, the phase angle θV−:rec
t of the negative sequence voltage V−:rec : θV−:rec
t ≈
1800 + θZ−+:t + θI+:t 13 , where θZ−+:t , θI+:t are phase angles of the impedance Z−+:t
and the current I+:t respectively. Fig. 7.5 illustrates the variation of θV−:rec
t with |I+:t |
obtained using unbalanced load flow analysis for each of the lines, justifying this re-
9
i.e. without introducing any effect of interactions which exist in interconnected network envi-
ronments, and of other sources of unbalance.
10
The impact of the negative sequence current I−:t/t in the line arising as a result of the line
itself (or of the term Z++:t I−:t/t where Z++:t is the positive sequence impedance of the line) on the
t
negative sequence voltage V−:rec is negligible when the line supplies passive loads.
11
The study system is three-wired. Thus, the impact of zero sequence variables on the negative
t
sequence voltage V−:rec can be ignored.
12
Descriptive data (lengths and impedances) of the lines is given in Appendix N (Table N.7).
13
Under high load power factor conditions, this can be further simplified as: θV−:rec
t ≈ 1800 +θZ−+:t .
153
constant for various |I+:t | values, where this distinct angle is determined by the angle
600 A
B
C
500
D
E
400 F
G
|V -:rec| (V)
H
300
I
t
J
200 K
L
M
100
N
0
0 50 100 150 200 250 300
|I+:t| (A)
t
Figure 7.4: Variation of |V−:rec | with |I+:t | for the individual lines
In summary, the term −Z−+:t I+:t (which is a vector) governs the voltage unbal-
|Z−+:t | which is the principal intrinsic parameter behind the voltage unbalance emis-
sion can be used as a measure for assessing the degree of asymmetry associated with
a line. Employing this, a rank for the lines of the study system can be assigned as
200
A
150 B
C
100
D
E
(deg .)
50
F
0
G
−:rec
-50 H
θV t
I
-100 J
-150
K
L
-200 M
25 75 125 175 225 275
N
|I+:t| (A)
Table 7.1: Ranking of the sub-transmission lines based on the associated degree of
asymmetry
Line/s |Z−+:t | (Ω) Rank
M ∼2.0 Extraordinarily High
F, I ∼1.3 Very high
A, D ∼0.65 High
B, N ∼0.5 Moderate
J, L, E, C ∼0.3 Low
G, K, H ∼0.1 Very low
155
The voltage unbalance behaviour exhibited by each of the lines, when operating in the
vector16 (referred to as ‘voltage unbalance emission vector’ [1]) using unbalanced load
This leaves the line under observation as the only source of unbalance that exists in
the system.
A line under observation (t) introduces voltage unbalance at various busbars where:
t
• The negative sequence voltage V−:rec−t at its receiving end busbar has to satisfy
t t
V−:rec−t = V−:send−t − Z−+:t I+:t − Z++:t I−:t/t (7.2)
t
where, V−:send−t is the negative sequence voltage introduced by the line under
other ideally transposed line (tany ) has to satisfy the following relationship17 :
t t
V−:rec−tany
= V−:send−tany
− Z++:tany I−:t/tany (7.3)
where,
t
V−:send−tany
- negative sequence voltage introduced by the line under observation
at the sending end busbar of any other ideally transposed line tany
Z++:tany - positive sequence impedance of any other ideally transposed line tany
I−:t/tany - negative sequence current in any other ideally transposed line tany ,
10
As evident from Chapters 3 - 4 (see footnote above), the influence of the terms
Z++:t I−:t/t in (7.2) and Z++:tany I−:t/tany in (7.3) on the negative sequence voltages
is insignificant compared to that of the term Z−+:t I+:t . That is, as in the case of a
various busbars of an interconnected network is also governed by the term −Z−+:t I+:t ,
Fig. 7.6 illustrates the VUF values at the various busbars (S2 - S9 of Fig. 7.1)
corresponding to the selected time stamp obtained by applying each of the lines
(one at a time) as a line under observation. Table 7.2 gives further details of the
lines including commentaries on the values of |Z−+:t | (columns 2), |I+:t | (columns 3)
and |Z−+:t I+:t | (columns 4), and also on the location in the network (column 5) by
assigning into the three sub-parts: US, CP and DS (Fig. 7.1). Commentary on the
location also indicates whether or not a line is in the direct path connecting the BSP
17
Note that the negative-positive coupling impedance Z−+:tany of any other line is zero, as it is
taken as ideally transposed.
18
Further explanation on this influence of the location of a line is given in Appendix N.
157
0.8 S2
S3
0.7
S4
0.6 S5
0.5
S6
VUF (%)
S7
0.4 S8
0.3 S9
0.2
0.1
0.0
A B C D E F G H I J K L M N
Line under observation
Figure 7.6: Nodal VUF values arising as a result of the individual lines
and CP and/or DS of the network. Depending on the emission levels which arise as a
result of the individual lines at the various busbars, a rank19 is assigned to the lines as
given in column 6. Referring to the entries in Table 7.2, the following can be derived:
• Line M which carries a relatively low level of positive sequence current is not
line of the network. Line J which carries the highest level of positive sequence
low level of asymmetry. This indicates, as implied by (7.2) and (7.3), that the
product |Z−+:t I+:t | is the term of concern when assessing the level of emission
• In contrary to the above, lines B and N introduce only a low level of emission
although their |Z−+:t I+:t | is equal to that of line J, a behaviour which can be
attributed to the location of these lines in the network20 . Lines B and N are not
19
Taking the highest emission level, i.e. the tallest bar of the group of bars of a line under
observation (Fig. 7.6), as the reference.
20
See Appendix N.
158
in the path connecting the BSP, CP and/or DS, and thus their contributions to
of line J.
200
A
B
Phase angle (deg.)
100 C
D
F
0
I
J
-100 L
M
N
-200
S2 S4 S6 S7 S8 S9
Busbar
Figure 7.7: Phase angles of the nodal negative sequence voltages introduced by the
individual lines
Fig. 7.7 illustrates the phase angles of the negative sequence voltages at S2, S4 and
that each of the above individual lines leads to a unique and nearly constant phase
angle across all busbars. Comparison of Figs. 7.5 (behaviour as standalone lines)
the similarity of the voltage unbalance behaviour of a line as a standalone line and
Table 7.2: Parameters, operating features and emission levels of the individual lines
connected Network
haviour which is vectorial in nature. This behaviour, in a global sense, can be ascer-
• The magnitude, as summarised in Table 7.2 - column 6 for the lines of the study
• The phase angle, as illustrated in Fig. 7.7 for the lines of the study system, can
Asymmetrical Lines
The behaviour of negative sequence variables is known to be linear [1]. That is, a
resultant negative sequence voltage at a busbar, which arises as a result of the inter-
I C
N
M
D
B L
J
A
F
Figure 7.8: Global emission vectors of the individual lines (drawn approximately to
a scale)
Fig. 7.8 for the lines of study system, establishes a basis that provides a comprehensive
understanding of the manner in which various asymmetrical lines interact with each
other to form the resultant influence. This basis can be used to derive the following:
• The resultant influence of the interaction of all lines in terms of a single vector
This, for the study system, is illustrated by the vector Rlines in Fig. 7.9.
utors as the respective global emission vectors lie in close proximity to the
slightly away from the resultant vector, it, being the line which introduces
the highest level of emission on its own, can also make a significant contri-
162
I C
N
M
D
B L
J
A
F
Rlines
Figure 7.9: Resultant influence of the interaction of all asymmetrical lines (drawn
approximately to a scale)
163
bution. The phase deviation close to 900 of the vector of line J with respect
of the vector of line D with respect to the resultant vector suggests that it
Fig. 7.10 illustrates the contributions, quantified using (7.4)27 while em-
ploying the results (presented in Figs. 7.6 and 7.7) obtained using unbal-
anced load flow analysis, made by each of the lines to the resultant voltage
unbalance levels (presented in Fig. 7.3) at the various busbars (S2, S4 and
S6 - S928 ).
t
|V−:Si |cos(θV−:Si
t − θV−:Si
lines )
Ct/Si = lines
× 100% (7.4)
|V−:Si |
where,
lines
V−:Si respectively
50
S2
40 S4
S6
30 S7
S8
Contribution (%)
20 S9
10
-10 S8
A B S6
C D Busbar
E F G H S2
I J K L
Line M N
Figure 7.10: Nodal contributions made by the individual lines to the resultant voltage
unbalance levels
line asymmetries.
Referring to Fig. 7.11 (I), which is deduced using Fig. 7.8 by representing
I I I
D D D
B+C+L+M+N B+C+L+M+N B+C+L+M+N
J J J
F A A
Resultant vector after
transposing lines F and A
Resultant vector after
transposing line F
Rlines
(I) (II) (III)
Figure 7.11: (I) Deduced from Fig. 7.8 (II) Effect of the transposition of line F only
(III) Effect of the transposition of lines A and F together (drawn approximately to a
scale)
out of which the vectors B+C+L+M+N and I lie in close proximity. This
is the case with the vectors D and J as well. The phase difference close
to 900 between these two groups (i.e. B+C+L+M+N and I, and D and
The effects of the transposition of line F only and lines A and F together
are illustrated in Figs. 7.11 (II) and (III) respectively. These demonstrate
Fig. 7.12 illustrates the effects, in terms of the residual VUF values at the
166
various busbars obtained using unbalanced load flow analysis29 , of the two
line transposition options identified above. This also shows the voltage
cantly affected by the line asymmetries as noted in Fig. 7.3), whereas this
0.8
0.4
0.0
S2 S3 S4 S5 S6 S7 S8 S9
Busbar
Figure 7.12: Effects, obtained using unbalanced load flow analysis, of the transposi-
tion of line F only, and lines A and F together
29
Operating scenario corresponds to the selected time stamp.
167
It has been noted, based on measurements, that the loads supplied by the study
system exhibit unbalance with regard to their active power components, whereas the
reactive power components have been seen to be reasonably balanced30 . Table 7.3
gives the distribution of the active (Pa , Pb and Pc ) and reactive (Qa , Qb and Qc )
power across the three phases at each of the load busbars, which corresponds to the
selected time stamp. It also gives the degree of asymmetry associated with the three
active power components, in terms of the standard deviation (µP ), at each of the
busbars.
Table 7.3: Distribution of the active and reactive power across the three phases at
each of the load busbars of the study system
Load busbar S2 S3 S4 S7 S8 S9
Pa (MW) 6.32 6.24 3.87 11.3 2.04 0.57
Pb (MW) 5.87 6.08 3.43 11.3 1.89 0.55
Pc (MW) 5.87 5.96 3.37 11.38 2.04 0.54
Qa ≈ Qb ≈ Qc 0.12 2.37 1.705 4.84 0.63 0.08
(MVAr)
µP 0.26 0.14 0.27 0.05 0.09 0.02
Similar to the case of the line asymmetries presented in Section 7.2.1, the impact
of the interaction of all unbalanced loads of the study system on the problem of volt-
30
This can be considered as a general case for higher voltage systems which are usually associated
with loads with a near unity power factor. That is, in comparison to the level of active power,
systems supply only a low level of reactive power at a load point, of which the level of unbalance
can be considered to be insignificant compared to that of the reactive power.
168
age unbalance is established in terms of the VUF values at the various busbars using
unbalanced load flow analysis. This is accomplished by synthesising the system oper-
ation31 while parameterising the loads as given in Table 7.3, and the lines assuming
they are ideally transposed. Fig. 7.13 illustrates these results, in comparison to the
emission levels arising as a result of the line asymmetries (reproduction of Fig. 7.3).
This reveals that the emission which arises due to the loads themselves is also consid-
erable (highest VUF: 1% at S9), although the line asymmetries are seen to dominate
the problem.
1.5
Load asymmetries
Line asymmetries
1.2
VUF (%)
0.9
0.6
0.3
0.0
S2 S3 S4 S5 S6 S7 S8 S9
Busbar
Figure 7.13: Nodal VUF values which arise as a result of the load asymmetries, in
comparison to that of the line asymmetries
31
Which corresponds to the selected time stamp.
169
the individual lines as elements in the interconnected network32 , this for the indi-
vidual loads is obtained using unbalanced load flow analysis by considering a single
fined in Table 7.3 while parameterising all other loads and all lines to represent a
balanced behaviour.
Fig. 7.14 illustrates the VUF values which arise as a result of each of the loads
as a load under observation at the various busbars. These results are summarised
in Table 7.4 by giving a rank to each of the loads based on its emission level in
a global sense33 (column 5). For brevity, the information given in Table 7.3 are
the real power components (column 2), and the three-phase loading level (column 3).
A commentary on the location (i.e. US, CP, DS) of each of the loads in the system
is also given in column 4. Referring to the entries in Table 7.4, the following can
be derived:
• Three-phase loading level is not a factor which governs the level of emission
factor that needs consideration when assessing the level of emission introduced
by the load35 .
32
See Section 7.2.3.
33
Taking the highest emission level as the reference, as in the case of the line asymmetries.
34
e.g. compare the emission levels introduced by the loads at S3 and S8 referring to their three-
phase loading levels (column 3).
35
e.g. compare the emission levels introduced by the loads at S9 and S8 referring to their µP .
170
• In contrary to the above, the load supplied by S8 (µP = 0.09) located in the
downstream part of the system is seen to introduce the highest level of emission,
the loads supplied by S2 and S4 (µP ≈ 0.3) located in the upstream part. It
is therefore evident that, for a similar degree of asymmetry, the loads supplied
by the busbars located in the downstream part of the system tend to introduce
0.6 S2
S3
0.5 S4
S5
0.4
S6
VUF (%)
S7
0.3
S8
S9
0.2
0.1
0.0
S2 S3 S4 S7 S8 S9
Load under observation
Figure 7.14: Nodal VUF values which arise as a result of the individual loads
Fig. 7.15 illustrates the phase angles of the negative sequence voltages which arise
as a result of the individual loads supplied by S2, S4, S7 - S936 at the various busbars
(S2 - S9). This demonstrates that, as in the case of the line asymmetries (Fig. 7.7),
the above individual loads yield a unique and nearly constant phase angle across all
the busbars.
36
Which introduce considerable emission levels as noted in Table 7.4.
171
Table 7.4: Operating features and emission levels of the individual loads of the study
system
250
200
Phase angle (deg.)
Load at S2
150 Load at S4
Load at S7
100 Load at S8
Load at S9
50
0
S2 S3 S4 S5 S6 S7 S8 S9
Busbar
Figure 7.15: Phase angles of the nodal negative sequence voltages introduced by the
individual loads
172
Arising as a result of a load under observation (LSi ), the negative sequence volt-
LSi
age V−:rec−tany
at the receiving end busbar of any line tany 37 (in other words, at any
LSi LSi
V−:rec−tany
= V−:send−tany
− Z++:tany I−:LSi /tany (7.5)
where,
LSi
V−:send−tany
- negative sequence voltage introduced by the load under observation at
I−:LSi /tany - negative sequence current in the line tany , which arises as a result of the
Equation (7.5) provides an explanation to the results presented in Figs. 7.15 and 7.16:
• The negative sequence currents I−:LSi /tany flow through the lines (i.e. through
the impedances Z++:tany ) are drawn by the load under observation. The level
of the negative sequence current drawn by this load is governed by the degree
of its asymmetry regardless of the three-phase loading level. That is, as seen
in Fig. 7.15, the principal intrinsic parameter that determines the negative se-
LSi
quence voltage V−:rec−tany
is the degree of asymmetry associated with the load
under observation. The location of this load in the system determines the part
of the network in which the negative sequence currents I−:LSi /tany flow38 , making
LSi
the negative sequence voltage V−:rec−tany
dependant secondarily on the location
of the load under observation as indicated by the results presented in Fig. 7.15.
37
Note that lines are taken as transposed for this case of observing the voltage unbalance behaviour
of individual loads. Thus, the impedance Z−+:tany = 0.
38
e.g. when the load is supplied by a upstream busbar, the negative sequence currents I−:LSi /tany
mainly flow through the lines located in the upstream part of the network.
173
• The phase angles of the negative sequence voltages which arise as a result of the
load under observation at the various busbars should be associated with the term
−Z++:tany I−:LSi /tany . The phase angle of the impedance Z++:tany depends only
on the X/R ratio of the line which has been noted to be nearly identical39 for
all lines of the study network. The phase angle of the negative sequence current
I−:LSi /tany is governed by the order of the distribution of the three active power
components of the load under observation across the three phases. Thus, the
phase angles of the negative sequence voltages at the various busbars remain
similar and unique for a particular load under observation as seen in Fig. 7.15.
Load at S8
Load at S7
Load at S2
Load at S9
Load at S4
Figure 7.16: Global emission vectors of the individual loads (drawn approximately to
a scale)
39
This can be considered as a general case.
174
vector of which:
• The magnitude, as summarised in Table 7.4 - column 5 for the loads of the study
associated with the load, and the location of the load in the network.
• The phase angle, as illustrated in Fig. 7.15 for the loads of the study system,
power of the load under observation across the three phases, and the X/R ratio
The global emission vectors of the individual loads41 S2, S4 and S7 - S9 of the
the global emission vectors of individual loads can be integrated to establish a basis
balanced loads interact with each other to form the resultant influence. Fig. 7.17
illustrates this resultant influence for the study system using a single vector (Rloads )
which is obtained by the summation of the individual global emission vectors shown
in Fig. 7.16.
40
See Section 7.2.4.
41
Which introduce considerable emission levels.
42
See Section 7.2.5.
175
Rloads Load at S8
Load at S7
Load at S2
Load at S9
Load at S4
Figure 7.17: Resultant influence of the interaction of all unbalanced loads (drawn
approximately to a scale)
176
Asymmetries
When both the line and load asymmetries exist concurrently in the study system43 ,
their impact on the problem of voltage unbalance is established in terms of the VUF
values at the various busbars using unbalanced load flow analysis. This is accom-
plished by synthesising the system operation44 while parameterising the loads as given
in Table 7.3, and the lines based on their actual construction. Fig. 7.18 illustrates
these results, in comparison to the emission levels which arise as a result of the line
asymmetries alone (reproduction of Fig. 7.3) and the load asymmetries alone (re-
Fig. 7.2) are also shown in Fig. 7.18. This reveals that the interaction of the various
line and load asymmetries causes voltage unbalance levels up to 1.8% at S8 and S9
System
Employing the linearity of negative sequence variables, the global emission vectors
of individual sources of unbalance (i.e. individual lines and loads) can be integrated
0.8
0.4
0.0
S2 S3 S4 S5 S6 S7 S8 S9
Busbar
Figure 7.18: Nodal VUF values which arise as a result of both the line and load asym-
metries, in comparison to that of the line asymmetries alone, and the load asymmetries
alone, and also to the measured values
which various untransposed lines and unbalanced loads interact with each other to
form the overall influence. Fig. 7.19 illustrates this overall influence for the study
system using a single vector (Rsystem ) which is obtained by the summation of the
individual global emission vectors shown in Figs. 7.9 (for lines) and 7.16 (for loads).
Fig. 7.19 also shows the vectors Rlines (Fig. 7.10) and Rloads (Fig. 7.17) where
This can be used to make the following assessments on the problem of voltage
unbalance, which are also confirmed using unbalanced load flow analysis:
• Contributions made by the line and load asymmetries to the overall voltage
unbalance levels.
Referring to the vectors Rsystem , Rlines and Rloads , the component of Rlines
which is in-phase with Rsystem accounts approximately for 60% of the mag-
178
Rload
Rsystem
Rlines
Figure 7.19: Resultant influence of the interaction of all lines and loads (drawn ap-
proximately to a scale)
179
the problem, whereas the load asymmetries play only a secondary role.
obtained using unbalanced load flow analysis, made by the line and load
at the various busbars (S2, S4 and S6 - S946 ). Fig. 7.20 clearly demon-
strates that, as identified above using the proposed technique, the line
80
Line asymmetries
Load asymmetries
60
Contribution (%)
40
20
0
S2 S4 S6 S7 S8 S9
Busbar
Figure 7.20: Nodal contributions made by the line and load asymmetries to the overall
voltage unbalance levels
45
Based on the approach given by (7.4).
46
Significantly affected busbars.
180
unbalance levels.
Observation of the global emission vectors of the individual lines and loads
illustrated in Figs. 7.9 and 7.16 respectively together with the vector
Rsystem (Fig. 7.19) suggests that, among all sources of unbalance, lines
F and I, these being represented by the largest and the closest vectors to
unbalance levels. In addition, the vectors of line A and the loads sup-
plied by S2 and S4, having relatively large magnitudes and being closer
major contributors (i.e. lines F and I). The phase deviation close to 900
of the vector of the load supplied by S8 with respect to Rsystem can make
and the load supplied by S7 with respect to Rsystem suggests that they can
(presented in Figs. 7.6, 7.7, 7.14 and 7.15) obtained using unbalanced load
Further, the assessments made on the role of line A and the loads supplied
Fig. 7.21.
S2
50
S4
S6
40
S7
S8
30
S9
Contribution (%)
20
10
-10
-20
Load at S2
Load at S3
Load at S4
Load at S7
Load at S8
S7
Load at S9
B
A
C
D
E
F
S2 Busbar
G
H
I
J
K
L
M
N
Source of unbalance
Figure 7.21: Nodal contributions made by the individual sources of unbalance to the
overall voltage unbalance levels
This chapter has established theoretical bases to broaden the understanding of the
terministic studies were carried out with the view to develop insights into the in-
fluences made by line and load asymmetries on the problem in a systematic man-
ner considering each of the asymmetrical elements. The following can be drawn
• The voltage unbalance behaviour (in terms of the level of voltage unbalance
secondarily dependant on the location of the line in the network. This be-
by referring to the product |Z−+:t I+:t | of the line and the location of the line in
the network, and the phase angle can be approximately derived using the term
−Z−+:t I+:t .
a global emission vector. Of this global emission vector, the magnitude can
with the load and the location of the load in the network. The phase angle
power of the load under observation across the three phases, and the X/R ratio
lines and loads) can be integrated to establish a basis that provides a compre-
unbalanced loads interact with each other to form the overall system behaviour.
That is, the overall influence of various sources of unbalance that exist in an in-
This basis can be used to assess the interconnected networks on the problem of
corrective options. This proposed technique was applied to the study system
to identify the major contributors to voltage unbalance levels and the effective
line transposition options, which were also confirmed using unbalanced load
flow analysis.
Chapter 8
8.1 Conclusions
This thesis has focused on making contributions for further development of the re-
cently released Technical Report IEC/TR 61000-3-13 which provides guiding prin-
ciples for coordinating voltage unbalance between various voltage levels of a power
aspects of this voltage unbalance allocation procedure such as the global emission
ance and associated deficiencies have been addressed. Furthermore, theoretical bases
that exist in interconnected networks on voltage unbalance, which would provide ad-
184
185
As an essential tool required for specific applications of the work presented in this
thesis, an unbalanced load flow algorithm based on the phase coordinate reference
frame was developed. The component level load flow constraints and the three-phase
modelling of system components were incorporated. The major task of this work
was to develop models for the representation of three-phase induction motors, which
Preliminary studies carried out on the global emission which arises as a result
of system inherent asymmetries revealed that this emission has some degree of load
of an asymmetrical radial line on the global emission does not consider this load
dependency, and thus it can be applied only when the line supplies passive loads. This
IEC approach was noted to be conservative when the line supplies large proportions
of three-phase induction motors. In this case of motor loads, the negative sequence
global emission, whereas it was seen to be insignificant for passive loads. The degree
of this influence was seen to be dependant primarily on the motor proportion, and
secondarily on the system and motor characteristics. In essence, the global emission
motors, were seen to arise not only as a result of the local HV lines but also as a
ing the global emission which arises due to line asymmetries at nodal level, taking
the load dependency of this emission into account were proposed. These were verified
using unbalanced load flow analysis in relation to simple test systems. These test
increases the emission, compared to emission levels when only passive loads exist, at
186
the busbar which is directly connected to the upstream system. At all other network
busbars, induction motor loads were seen to reduce the emission levels which arise due
to the local line asymmetries, compared to that when only passive loads exist. In HV
power system, in the presence of motor loads, the influence of the downstream MV
line asymmetries was noted to either decrease or increase the resultant emission lev-
els with respect to the local emission levels depending on the impedance/admittance
characteristics of the downstream lines relative to the local HV lines. The proposed
approach was further verified using the IEEE 14-bus test system taking it as an HV
The propagation of voltage unbalance from higher voltage to lower voltage systems
in terms of transfer coefficients was initially examined in the presence of four basic
load types (i.e. constant impedance, constant current, constant power and three-
• It was seen that the value of unity which has been assumed in IEC/TR 61000-
prevailing constant power loads. Transfer coefficients for passive loads were
τ
1
seen to be scaled up by the factor 1 + j ksc−s ∠θpf :s relative to unity, where
loads respectively1 . Noting that the factor ksc−s can be a value in the range
all load types with a high degree of accuracy. That is, voltage unbalance can get
1
ksc−s - ratio between the short-circuit capacity (in MVA) at the busbar S under evaluation and
the total load (in MVA) supplied by the busbar S, and θpf :s - pf angle (− and + for lagging and
leading conditions respectively) at the busbar S.
187
higher voltage to lower voltage levels. As the values of ksc−s for MV systems
are usually smaller than those for LV systems, this amplification in the case of
from higher voltage to lower voltage levels in the presence of induction motor
loads. This attenuation, relative to unity, can be given by the scaling factors
1 01
„ « and 2 13 in the cases of the MV to LV and HV
1+ k ks
“ ”
41+ 1 ks
sc−lv ksc−mv
@
ks
A5
1+
ksc−lv
agg
to MV propagation respectively2 . As ksc−mv for MV systems < ksc−lv for LV
5 < ks < 7, the degree of this reduction of the transfer coefficients relative to
MV transfer coefficients were developed. These were verified using unbalanced load
flow analysis for various scenarios. It was demonstrated that, compared to the IEC
method, the proposed new formulation for assessing the MV to LV transfer coefficient
gives a more accurate estimation, particularly for load bases which are dominated by
vary in the ranges of 0.6 to 1.1 and 0.5 to 1.4 respectively depending on the system
2
Motor loads are considered to be supplied at the LV level. ks - ratio between the positive and
negative sequence impedances of the aggregated motor load supplied by the LV system, ksc−lvagg -
ratio between the short-circuit capacity (in MVA) at, and the total load (in MVA) supplied by, the
aggregated LV busbar.
188
Preliminary studies carried out on the propagation of voltage unbalance from one
revealed that, for a simple two-bus radial sub-system, the influence coefficients can
influence coefficients were seen to be considerably smaller than unity (in other words,
than that in the case of passive loads) for induction motor loads. The conclusion of
this observation is that the negative sequence currents which arise as a result of the
on influence coefficients for motor loads, whereas it is insignificant for passive loads.
Employing this, a systematic method for evaluating influence coefficients for inter-
connected network environments was developed. This was verified using unbalanced
load flow analysis in relation to a simple test system, and also to the IEEE 14-bus
employing a simple test system with and without the inclusion of the influence of
system inherent asymmetries. For both of the cases stated above, the IEC/TR 61000-
3-13 procedure was noted to lead to situations where the set planning level are being
exceeded even when no customer exceeds the allocated limit. Closely aligning with the
proposed. In this CBV allocation technique, emission levels at network busbars are
or below the set planning levels when all loads inject their limits derived under the new
approach. The issue of the emission arising as a result of system inherent asymmetries
involved with the case of voltage unbalance was taken into account according to
IEC/TR 61000-3-13 using the factor Kue. This revised allocation technique was
189
examined employing the above mentioned test system, and it was seen to restrict the
resulting emission level at all network busbars at or below the set limit. Further, this
proposed alternative technique was noted to be able to satisfy all four key allocation
objectives.
With the view of developing theoretical bases which describe the voltage unbal-
work has been monitored to experience voltage unbalance levels above the applicable
vector’, and extensively analysed. The voltage unbalance behaviour (in terms of
the magnitude and the phase angle) exhibited by an asymmetrical line at various
term −Z−+:t I+:t as in the case of a standalone line, and secondarily dependant
on the location of this line in the network. This behaviour, in a global sense, was
of the line and the location of this line in the network. The phase angle of this
seen, as for an asymmetrical line, that the voltage unbalance behaviour exhib-
190
using a global emission vector. Of this global emission vector, the magnitude
ated with the load and the location of this load in the network, and the phase
of power of the load under observation across the three phases and the X/R
• It was seen that the voltage unbalance behaviour exhibited by individual lines
and loads can be ascertained, in a global sense, in terms of the global emission
global emission vectors were integrated to establish a basis that provides a com-
unbalanced loads interact with each other to establish the overall system be-
haviour. That is, the overall influence of various sources of unbalance that exist
3-13. This proposed technique was applied to the study system to identify the
major contributors to voltage unbalance levels and the effective line transposi-
tion options, which were also confirmed using unbalanced load flow analysis.
191
• In the study system, the line asymmetries were seen to contribute approxi-
mately by 60% - 70% to the overall voltage unbalance levels, whereas the con-
tribution made by the load asymmetries was only 25% - 30%. This indicates
practice.
Australian standard in the future, provides valuable techniques for managing voltage
Zealand harmonics (AS/NZS 61000-3-6 [75]) and flicker (AS/NZS 61000-3-6 [76])
standards, which are essentially based on the respective IEC technical reports, amend-
ments has been shown to require in order to allow accurate and straightforward
missioned the writing of the handbook HB-264-2003 [4] which gives more prescriptive
procedures for the use of these harmonics and flicker standards. Similarly, in adopting
require in order to assist the application of such a standard to complex systems such
process would be the development of software modules which are able to produce
various factors (e.g. Kue factor) and coefficients (e.g. transfer and influence coeffi-
cients) directly from network databases. Such software modules would also benefit
Although the Technical Report IEC/TR 61000-3-13 provides the guidance for eval-
uating customer emission limits, it does not discuss the assessment of the compliance
against these limits. Development of systematic techniques to carry out this would sig-
nificantly assist in further developing IEC/TR 61000313 such that a complete voltage
ing group which is responsible for updating IEC/TR 61000-3-13. The work presented
such techniques, yet a substantial amount of further work is required for completing
the task.
Variation in network topologies and loads makes the setting of prescriptive generic
the planning levels proposed in IEC/TR 61000-3-13 are given as indicative values. The
significant differences between networks mean that there are likely to be situations
in which it may suit utilities to choose alternative planning levels. Guidance on how
The Technical Report IEC/TR 61000-3-13, as well as this thesis, has focused
third and final stage, or as to how a customer who would fail to comply with the
sented in Chapter 7, extended work such as the carrying out suitable field measure-
ments and the examination of the proposed techniques in relation to other networks
193
initial step towards a broad research area, requires more deterministic, analytical and
theoretical work such that the techniques are further developed to be more definitive.
As an example, Chapter 7 only covers a relative assessment of the influence and the
Commission, 2008.
[6] M.B. Hughes. Revenue metering error caused by induced voltage from adjacent
1992.
[7] M. Tavakoli Bina and E. Pasha Javid. A critical overview on zero sequence
[8] J.E. Parton and Y.K. Chant. The three-limbed phase transformer with controlled
zero sequence effect. IEEE Trans. on Power Apparatus and Systems, 90(5):2019–
[10] NRS 048-2: Electricity supply - quality of supply, part 2 - voltage characteristics,
[11] National Electricity Code Australia, Version 1.0 - Amendment 9.0. Technical
[14] CIGRE/CIRED Joint Working Group C4.103. Assessment of emission limits for
[15] UIE guide to quality of electrical supply for industrial installations - part 1: gen-
[16] UIE guide to quality of electrical supply for industrial installations - part 4:
[17] A.K. Singh, G.K. Singh, and R. Mitra. Some observations on definitions of
[18] P. Pillay and M. Manyage. Definitions of voltage unbalance. IEEE Power Engi-
[20] NEMA MG1: Motors and generators. Technical report, National Electricity
[21] IEEE 112: Standard test procedures for polyphase induction motors and gener-
[22] IEEE 100: standard dictionary of electrical and electronics terms. Technical
[23] IEEE 1159: IEEE recommended practice for monitoring electric power quality.
[24] M.H.J. Bollen. Definitions of voltage unbalance. IEEE Power Engineering Re-
[25] Math H.J. Bollen and Irene Gu. Signal Processing of Power Quality Disturbances.
[26] Tsai-Hsiang Chen. Criteria to estimate the voltage unbalances due to high-speed
[27] Tsai-Hsiang Chen and Hung-Yuan Kuo. Analysis on the voltage unbalances
ergy Management and Power Delivery (EMPD ’95), volume 2, pages 657–661,
November 1995.
[28] M. Cao, P.P. Biringer, and V. Bulat. Eddy current losses and unbalance in
[29] Eric T.B. Gross and S.W. Nelson. Electromagnetic unbalance of untransposed
893, 1955.
[30] R.H. Brierley, A.S. Morched, and T.E. Grainger. Compact right-of-ways with
1991.
[31] Z. Emin and D.S. Crisford. Negative phase-sequence voltages on E&W trans-
[32] A. Ametani, D. Van Dommelen, and I. Utsumi. Study of super-bundle and low
[33] W.R. Bullard, H.L. Lowe, and H.W. Wahlquist. Calculation of unbalanced volt-
[34] Paulo Vinicius Santos Valois, Carlos Mhrcio Vieira Than, Nelson Kagan, and
[35] E.T.B. Gross and Wing Chin. Electrostatic unbalance of untransposed single cir-
cuit lines. AIEE Trans. on Power Apparatus and Systems, 87(1):24–34, January
1968.
[36] R. Gutman and L.B. Wagenaar. EHV transformer bank unbalance: practical
1996.
[39] Gafford W.C. Duesterhoeft and C.C. Mosher. Heating of induction motors on
[40] C.Y. Lee. Effects of unbalanced voltage on the operation performance of a three-
1999.
Australia, 1997.
[42] D.P. Manjure and E.B. Makram. Impact of unbalance on power system harmon-
[43] A. von Jouanne and B. Banerjee. Assessment of voltage unbalance. IEEE Trans.
[44] Math H.J. Bollen. Understanding Power Quality Problems - Voltage Sags and
[45] L. Moran, P.D. Ziogas, and G. Joos. Design aspects of synchronous PWM
[46] V.O. Zambrano, E.B. Makram, and R.G. Harley. Stability of a synchronous ma-
[47] E.B. Makram, V.O. Zambrano, and R.G. Harley. Stability of a synchronous
[48] Edited by Angelo Baggini. Handbook of Power Quality. John Wiley and Sons,
2008.
[50] T.H. Chen. Evaluation of line loss under load unbalance using the complex un-
[51] J. Kuang and S.A. Boggs. Pipe-type cable losses for balanced and unbalanced
[52] A.E. Emanuel. On the definition of power factor and apparent power in unbal-
anced polyphase circuits with sinusoidal voltage and currents. IEEE Trans. on
[54] J.S. Wu, K.L. Tomsovic, and C.S. Chen. A heuristic search approach to feeder
[55] M.W. Siti, D.V. Nicolae, A.A. Jimoh, and A. Ukil. Reconfiguration and load
[56] Tsai-Hsiang Chen and Jeng-Tyan Cherng. Optimal phase arrangement of dis-
provement and loss reduction using a genetic algorithm. IEEE Trans. on Power
[57] Paul M. Anderson. Analysis of Faulted Power Systems. John Wiley and Sons,
1995.
[59] Jen-Hung Chen, Wei-Jen Lee, and Mo-Shing Chen. Using a static Var compen-
[60] Kuang Li, Jinjun Liu, Biao Wei, and Zhaoan Wang. Comparison of two control
Power Electronics and Motion Control (IPEMC 2004), volume 3, pages 1213–
[61] Tong XiangQian, Xi Keqing, Shen Ming, and Ma Xianhong. Reactive power
[62] A. Campos, G. Joos, P.D. Ziogas, and J.F. Lindsay. Analysis and design of a
[63] A. Campos, G. Joos, P. Ziogas, and J. Lindsay. Analysis and design of a series
[64] V.B. Bhavaraju and P.N. Enjeti. An active line conditioner to balance voltages
March-April 1992.
[65] Hong-Ju Jung, In-Young Suh, Byung-Seob Kim, Rae-Young Kim, See-Young
Choi, and Jong-Hwhan Song. A study on DVR control for unbalanced voltage
[66] Gong Maozhong, Liu Hankui, Gu Hanjun, and Xu Dianguo. Active voltage
[67] D. Graovac, V. Katic, and A. Rufer. Power quality problems compensation with
Afonso. A combined series active filter and passive filters for harmonics, un-
Environment. Springer.
[72] A. Robert and J. Marquet (on behalf of Working Group 36.05). Assessing volt-
age quality with relation to harmonics, flicker and unbalance. In Proc. CIGRE
[74] Robert Koch, Germain Beaulieu, Luc Berthet, and Mark Halpin. Interna-
tional survey of unbalance levels in LV, MV, HV and EHV power systems:
203
[77] J. Duncan Glover and Gareth Digby. Software Manual - Power System Analysis
[78] R.C. Dugan. Induction machine modeling for distribution system analysis -
[79] Paul C. Krause, Oleg Wasynczuk, and Scott D. Sudhoff. Analysis of Electric
Machinery and Drive Systems. John Wiley and Sons Inc., 2002.
[80] R.C. Dugan, M.F. McGranaghan, and H.W. Beaty. Electrical Power Systems
Derivation of (3.5)
Equation (3.4) in relation to the considered radial MV-LV network shown in Fig. 3.2
t
|V−:g/mv:rec | = |Z−+:t I+:t + Z++:t I−:t/t | (A.1)
When the loads Smv:rec−local and Smv:rec−ds represent constant impedance loads, the
V+:send
I+:t = (A.2)
Z++:send
I−:t/t = Y−+:send V+:send (A.3)
where,
V+:send - positive sequence voltage at the sending end busbar (labelled ‘send’)
1
Note the voltage at the sending end busbar is balanced.
204
205
−Z−+:xr
Y−+:xr ≈ (A.4)
Z++:xr Z−−:xr
seen at the busbar xr can be expressed using various admittance elements seen at the
busbar as:
−Y−+:xr
Z−+:xr ≈ (A.5)
Y++:xr Y−−:xr
where,
For the considered system where the loads are balanced (i.e. decoupled sequence
impedances) and the MV line is untransposed: Z−+:send = Z−+:t . Thus, the admit-
−Z−+:t
Y−+:send ≈ (A.6)
Z++:send Z−−:send
t
Z++:t
|V−:g/mv:rec | ≈ Z−+:t I+:t 1 −
(A.7)
Z−−:send
2
Through the inversion of the impedance matrix and the simplification.
206
t
Z −−:rec
|V−:g/mv:rec | ≈ Z−+:t I+:t (A.8)
Z−−:send
Appendix B
(Fig. 3.2)
This test system is designed mainly based on the data given in [78].
• MV line -
phase positioning1 : a b c
– Conductor data:
resistance = 0.19014Ω/km
1
This shows the considered arrangement of the three phase conductors (a, b and c) of the hori-
zontal tower.
207
208
(0.0349∠300 )
(0.1839∠820 )
= 5%, and secondary tap setting = 1.03pu. This gives a value of 19 for the
factor ksc−lvagg .
– Other relevant details: input power = 50kVA at 0.9 lagging pf and ks = 6.7
power of 50kVA at 0.9 lagging pf which consist of equal shares constant power
Derivation of (3.14)
Equation (3.13) in relation to the considered radial MV-LV network shown in Fig. 3.2
t
Z−−:rec−im
|V−:g/mv:rec | ≈ Z−+:t I+:t
(C.1)
Z−−:send−im
Noting Z−−:send−im = Z−−:rec−im + Z++:t and the inductive nature of the associated
t 1
|V−:g/mv:rec | ≈ |Z−+:t I+:t |
Z++:t Z++:rec
(C.2)
1 + Z++:rec Z−−:rec−im
where, Z++:rec - downstream equivalent positive sequence impedance seen at the re-
ceiving end busbar of the MV line. The ratio ZZ++:rec
++:t
can be expressed as:
Z++:t V Rt
Z++:rec ≈ 1 − V Rt (C.3)
1
e.g. the impedance Z−−:rec−im is made up of the negative sequence impedances of induction
motors and the MV-LV transformer.
209
210
! !
Z++:rec 1 1
Z−−:rec−im = (C.4)
(nml )2 |Z−−:im +Z++:ml−lv | |V+:rec |2
|V+:rec |2 |Z++:rec |
where,
|V+:rec |2
The term |Z++:rec |
represents the total MVA load (i.e. Smv:rec−local + Smv:rec−ds ) sup-
plied by the MV busbar. Using the relationships |V+:rec | = nml |V+:lv | and |Z−−:im +
Z++:rec 1 1
Z−−:rec−im ≈ (Smv:rec−local + Smv:rec−ds ) Z−−:im Z++:im |Z++:lv |
|Z++:ml−lv |
Z++:im Z++:lv |V+:lv |2 + |V+:lv |2
(C.5)
where,
Noting that:
Z−−:im 1 Z++:im 1 |Z++:lv | 1
Z++:im = ks
, Z++:lv = km
, |V+:lv |2
= Smv:rec−ds
Z++:rec klv
Z−−:rec−im ≈ (C.6)
1 1
k s km
+ ksc−lv
agg
t 1
|V−:g/mv:rec | ≈ |Z−+:t I+:t | (C.7)
V Rt klv
1+ 1−V Rt 1
+k 1
ks km sc−lvagg
Appendix D
The admittance Y−−:x−im can be generally written noting the inductive nature of the
Z++:x
Y−−:x−im ≈ −j |Y++:x |
(D.1)
Z−−:x−im
where, Y++:x , Z++:x - downstream equivalent positive sequence admittance and imped-
ance respectively seen at the busbar x. Noting that the system representation of the
MV busbar x shown in Fig. 3.6 and that of the receiving end busbar of the MV line
considered in Fig. 3.2 are similar, the ratio ZZ−−:x
++:x
can be given according to (C.6) as:
Z++:x klv:x
Z−−:x ≈ (D.2)
1
ks:x km:x
+ ksc−lv1
agg :x
The absolute value of the admittance Y++:x can be expressed in terms of the system
√
3 |I+:x |
|Y++:x | ≈ (D.3)
Vn−mv
212
213
The matrix equation (3.25) which gives the proposed methodology for MV networks
0
lines
[V−:g/mv ]n×1 ≈ −[Y++ ]−1
n×n [Y−+ ]n×n [V+ ]n×1 (E.1)
where,
0
Y++:xy ≈ Y++:xy + Y−−:x−im for x = y
0
Y++:xy = Y++:xy for x 6= y
214
215
0
1.05∠−4.07
12.47
0.98∠−6.74 ×
[V+ ] = 0 √ kV (E.2)
3
1.01∠−5.570
Noting that the positive sequence admittance per km of the lines = 1.0098−j2.0630S
and also that the impedance Y++:11 has to be incorporated with the HV-MV trans-
0.3722 − j1.3119 −0.1010 + j0.2063 −0.2020 + j0.4126
[Y++ ] =
−0.1010 + j0.2063 0.2020 − j0.4126 −0.1010 + j0.2063
S
(E.3)
−0.2020 + j0.4126 −0.1010 + j0.2063 0.3029 − j0.6189
Noting that the negative-positive sequence coupling admittance per km of the lines
0.0078 + j0.0546 −0.0026 − j0.0182 −0.0052 − j0.0364
[Y−+ ] =
−0.0026 − j0.0182 0.0052 + j0.0364 −0.0026 − j0.0182
S
(E.4)
−0.0052 − j0.0364 −0.0026 − j0.0182 0.0078 + j0.0546
As busbars 1 and 3 supply passive loads (i.e. km:1 = 0, km:3 = 0), the admittances
• Case 1 - busbar 2 supplies passive loads or km:2 = 0. That is, the admittance
0
Y−−:2−im ≈ 0 implying that [Y++ ] ≈ [Y++ ]. Then, the substitution of (E.2),
0
lines
[V−:g/mv ]=
47.38 V
(E.5)
26.64
0
lines
[Ug/mv ]=
0.67 %
(E.6)
0.37
(3.24) gives the admittance Y−−:2−im for klv:2 = 1, ksc−lvagg :2 = 18, ks:2 = 6.7,
0
Then, the matrix [Y++ ] can be obtained as:
0 0 0
0
[Y++ ] ≈ [Y++ ] +
0 −j0.1316 0 S
(E.8)
0 0 0
5.94
lines
[V−:g/mv ]=
31.46 V
(E.9)
17.38
217
0.08
lines
[Ug/mv ]=
0.45 %
(E.10)
0.24
Appendix F
Derivation of (4.7)
t
Equation (4.5) which gives the negative sequence voltage V−:g/hv:rec caused by the
local HV line t of the considered radial HV-MV-LV network shown in Fig. 4.2 is
t
V−:g/hv:rec = −(Z−+:t I+:t + Z++:t I−:t/t ) (F.1)
Similar to (A.8) in the case of the radial MV-LV network considered in Fig. 3.2, (F.1)
can be written noting that the local HV line is the element which gives rise to the
negative-positive sequence coupling impedance seen at the sending end busbar of the
HV line as1 :
t Z++:t
V−:g/hv:rec = −Z−+:t I+:t 1 − (F.2)
Z−−:send−im
Z++:t
Due to the inductive nature of the impedances Z++:t and Z−−:send−IM , Z−−:send−im
≈
Z++:t
Z−−:send−im = µ. Thus, (F.2) can be rewritten as:
t
V−:g/hv:rec = −Z−+:t I+:t (1 − µ) (F.3)
1
The downstream MV line is treated as balanced in this case.
218
219
td
Equation (4.6) which gives the negative sequence voltage V−:g/hv:rec caused by the
td
V−:g/hv:rec = −Z++:t I−:td /t (F.4)
The negative sequence current I−:td /t = Y−+:send V+:send where, according to (A.4),
−Z−+:send 2
Y−+:send = Z++:send Z−−:send−im
. Noting that the impedance Z−+:send arises as a result
of the downstream MV line in this case3 , it is not simply equal to the impedance
where, nhm - operating turns ratio of the HV-MV coupling transformer. Then, (F.4)
td 2 Z++:t V+:send
V−:g/hv:rec ≈ (nhm ) ktd kn Z−+:td (F.6)
Z−−:send−im Z++:send
Noting that:
td
V−:g/hv:rec ≈ (µktd kn σ)Z−+:t I+:t = ζZ−+:t I+:t (F.7)
2
New notations which have been used here are as defined in Appendix A while using the subscript
‘send’ with reference to the sending end busbar of the local HV line.
3
The local HV line is treated as balanced in this case.
4
A uniform pf across the network is assumed.
220
t+td
Summation of (F.3) and (F.7) gives the negative sequence voltage V−:g/hv:rec which
t+td
|V−:g/hv:rec | ≈ |Z−+:t I+:t (1 − µ − ζ)| (F.8)
Appendix G
Derivation of (4.9)
New notations used in this appendix, unless defined here or in Chapter 4 (Section 4.2),
are as per Appendix C while using the subscripts ‘send’ and ‘rec’ with reference to the
sending and receiving end busbars respectively of the local HV line of the considered
Z++:t
µ = (G.1)
Z−−:send−im
1
µ≈
Z−−:rec−im Z++:rec (G.2)
1 + Z++:rec Z++:t
The ratio ZZ++:rec can be expressed in terms of the voltage regulation of the line t as:
++:t
Z++:rec 1 − V Rt
Z++:t ≈ V Rt (G.3)
221
222
Noting that Z−−:rec−im = (nml nhm )2 (Z−−:im +Z++:mlr −lv +Z++:td −lv +Z++:hm−lv ) when
motor loads are supplied only by the MV line at the busbar LVr , and the inductive
Z−−:rec−im
nature of the associated impedances, the ratio Z++:rec can be given as:
!
2
≈ |V+:rec | |Z−−:im | |Z++:mlr −lv + Z++:td −lv + Z++:hm−lv |
Z−−:rec−im
|V+:rec |2
+ |V+:rec |2
Z++:rec |Z++:rec |
(nml nhm )2 (nml nhm )2
(G.4)
|V+:rec |2
The term |Z++:rec |
represents the total load Srec (in MVA) supplied by the HV busbar
nml nhm |V+:lv |
rec. Further, |V+:rec | ≈ (1−V Rtd )
. Substitution of these and rearrangement of (G.4)
gives:
≈ Srec (1 − V Rt )2 |Z−−:im | + |Z++:mlr −lv + Z++:td −lv + Z++:hm−lv |
Z−−:rec−im
d
Z++:rec |V+:lv |2 |V+:lv |2
(G.5)
Noting that:
|Z−−:im | 1 1
|V+:lv | 2 = ksr kmr Slvr
(see (C.5) - (C.6))
agg
Srec 1
Slvragg
= klvr
Slvragg 1
Ssc−lvragg
= ksc−lvragg
Z−−:rec−im (1 − V Rtd )2
≈ 1 1
Z++:rec + (G.6)
klvr ksr kmr ksc−lvragg
223
1
µr ≈ „ « (G.7)
1 1
(1−V Rt )(1−V Rtd )2 ksr kmr
+k
sc−lvragg
1+ V Rt klvr
Appendix H
• System loads -
– Shv = 0
– Slvragg - represents passive loads and induction motors for the cases where
• HV line -
phase positioning1 :
a
b
c
1
This shows the considered arrangement of the phases a, b and c of the horizontal tower.
224
225
– Conductor data:
resistance = 0.105Ω/km
(0.0349∠300 )
(0.2061∠600 )
the considered operating scenario) 12.47kV, 3.2187km lines of which the details
• Number of 12.47kV, 3.2187km lines and the values of the factors ksc−lvragg and
2
Which are obtained using load flow analysis.
Appendix I
According to (A.4), the admittance Y−+:x which arises as a result of line asymmetries
that exist in the downstream system supplied by the busbar x can be written as1 :
−Z−+:x
Y−+:x ≈ ≈ j Z−+:x Y++:x |Y−−:x−im | (I.1)
Z++:x Z−−:x−im
Noting that:
• the impedance Z−+:send given by (F.5) in the case where the untransposed MV
line gives rise to unbalance in the considered radial HV-MV-LV network shown
in Fig. 4.2, is equal to the respective impedance Z−+:rec that is seen at the
receiving end busbar of the HV line, as the local HV line is treated as balanced
• the system representation of any HV busbar x shown in Fig. 4.5 and that of the
receiving end busbar of the radial HV-MV-LV network considered in Fig. 4.2
are similar,
1
Note that the impedance Z−−:x−im is inductive.
227
228
where, nhm:x - operating turns ratio of the HV-MV coupling transformer supplied by
the busbar x. Similar to (D.3), the admittance Y++:x can be expressed as:
√ !
3 |I+:x |
Y++:x ≈ ∠θpf :x (I.3)
Vn−hv
√ !
3 |I+:x |
|Y−+:x | ≈ ktd:x kn:x |Y−−:x−im Z−+:td:x −hv | (I.4)
Vn−hv
θY−+:x ≈ 900 + θZ−+:td:x + θpf :x (I.5)
Appendix J
The matrix equation (3.22) which gives the proposed methodology for HV networks
0
lines
[V−:g/hv ]n×1 ≈ −[Y++ ]−1
n×n [Y−+ ]n×n [V+ ]n×1 (J.1)
where,
0
Y++:xy ≈ Y++:xy + Y−−:x−im for x = y
0
Y−+:xy ≈ Y−+:xy + Y−+:x for x = y
0 xy
Y++:xy = Y++ for x 6= y
0 xy
Y−+:xy = Y−+ for x 6= y
229
230
The matrices [V+ ]1 , [Y++ ] and [Y−+ ] for the three-bus HV test system shown in
As busbars 1 and 3 supply passive loads (i.e. km:1 = 0, km:3 = 0), the admittances
• Case 1 - busbar 2 supplies passive loads or km:2 = 0. That is, the admittances
0 0
Y−−:2−im ≈ 0 and Y−+:2 ≈ 0 implying that [Y++ ] ≈ [Y++ ] and [Y−+ ] ≈ [Y−+ ]
respectively. Then, the substitution of (J.2), (J.3) and (J.4) in (J.1) gives:
0
lines
[V−:g/hv ]=
226.89 V
(J.5)
119.02
1
This is obtained using load flow analysis.
231
0
lines
[Ug/hv ]=
0.59 %
(J.6)
0.30
(4.13) gives the admittance Y−−:2−im for klvr:2 = 1, ksc−lvragg :2 = 3.6, ksr:2 = 6.7,
Equation (4.14) gives the admittance Y−+:2 for ktd:2 = 1, kn:x = 1, I+:2 = 208A,
−260 as:
0 0
Then, the matrices [Y++ ] and [Y−+ ] can be obtained as:
0 0 0
0
[Y++ ] ≈ [Y++ ] +
0 −j0.0154 0 S
(J.9)
0 0 0
0 0 0
0 × 10−3 S
[Y−+ ] ≈ [Y−+ ] +
0 −0.0077 + j0.1093 0 (J.10)
0 0 0
232
94.19
lines
[V−:g/hv ]=
73.52 V
(J.11)
5.51
0.23
lines
[Ug/hv ]=
0.19 %
(J.12)
0.01
Appendix K
base.
233
234
Table K.3: Generator and load bus data: three-phase MW and MVAr values
Bus Generation Load
number MW MVAr MW MVAr
1 (reference bus) 0 0 0 0
2 40 0 21.7 12.7
3 0 0 94.2 19.0
4 0 0 47.8 3.9
5 0 0 7.6 1.6
6 0 0 11.2 7.5
7 0 0 0 0
8 0 0 0 0
9 0 0 29.5 16.6
10 0 0 9.0 5.8
11 0 0 3.5 1.8
12 0 0 6.1 1.6
13 0 0 13.5 5.8
14 0 0 14.9 5.0
Table K.4: Transformer data: impedances and secondary tap settings (1st and 2nd
bus numbers refer to the primary and the secondary respectively)
Transformer Impedance Secondary tap
number (pu) setting
4-7 j0.20912 1.022
4-9 j0.55618 1.032
5-6 j.25202 1.073
7-8 j0.17615 1
7-9 j0.11001 1
235
Derivation of (5.18)
equal positive, negative and zero sequence impedances) is supplied by the LV sys-
tem shown in Fig. 5.2, where the busbars US and S represent MV and LV systems
V−:Umv /lv−mv 1
= (L.1)
V−:mv 1
1 + Z++:ml−lv Z++:L
+ 1
+ ... + 1
1
Z++:L2 Z++:Ln
where, Z++:Li - positive sequence impedance of any load Li (i = 1, 2, ..., n). Employing
(5.3), (L.1) can be expressed in terms of the system and load characteristics and the
V−:Umv /lv−mv 1
= (L.2)
V−:mv k L k L kLn
1 + j ksc−lv ∠θpf :L1 + j ksc−lv ∠θpf :L2 + ... + j ksc−lv ∠θpf :Ln
1 2
where,
kLi - ratio between the load Li (in MVA) and the total load (in MVA) supplied by
the LV system
237
238
V−:Umv /lv−mv 1
≈
V−:mv k L k L k Ln
1 + j ksc−lv ∠θpf :L1 1 + j ksc−lv ∠θpf :L2 ... 1 + j ksc−lv ∠θpf :Ln
1 2
(L.3)
Alternatively, the negative sequence voltage |V−:Umv /lv−mv | can be written in an ex-
panded form by decomposing the negative sequence current I−:Uus /tf (I−:Uus /tf =
I−:L1 + I−:L2 + ... + I−:Ln ) for the above considered case as:
n
X
|V−:Umv /lv−mv | = V−:mv − (Z++:ml−mv I−:Li ) (L.4)
i=1
where, I−:Li - negative sequence current (referred to MV) in the load Li , which
arises as a result of the MV unbalance. Noting that the influence of the term
V
Z++:tf −mv I−:Umv /tf on the ratio −:UVmv /lv−mv
−:mv
has been replaced by the factor (referred
1
to as ‘replacement factor’) ˛ ˛ for an aggregated constant impedance
˛1+j 1 ∠θpf :lv ˛
˛ ˛
˛ ksc−lv ˛
load4 , the comparison of (L.3) and (L.4) indicates, for the series of loads, that the in-
V
fluence of each of the Z++:ml−mv I−:Li components on −:UVmv /lv−mv
−:mv
has been replaced
1
by the factor ˛
k
˛ which involves an additional factor kLi . This observa-
˛1+j Li ∠θpf :L ˛
˛ ˛
˛ ksc−lv i˛
tion suggests that the impact of a share I−:Li of the total negative sequence current
I−:Uus /tf (or of a share Z++:ml−mv I−:Li of the total Z++:ml−mv I−:Uus /tf ) with a unique
behaviour, which is determined by the load type and its characteristics, on the prop-
agation of the negative sequence voltage from higher voltage to lower voltage systems
1
− and + for lagging and leading conditions respectively.
2
As kLi , kLj < 1 and 10 < ksc−lv < 25.
3 k L kL
i j
kL k L
i j
i.e. (ksc−lv )3 , (ksc−lv )4 etc.
4
Refer to (5.5) and (5.6).
239
terms where an individual term accounts for an unique I−:Li or a load element Li .
L2 ,..., Ln 5 on the propagation of the negative sequence voltage from higher voltage
form given by (L.3). An individual term of this product, which corresponds to any
load element Li , can be obtained by modifying the replacement factor for its load
1
type6 where ksc−lv is multiplied by the load proportion kLi . Applying this to a mix
of constant impedance (Z), constant current (I), constant power (P Q) and induction
motor (IM ) loads (i.e. Li = z, i, pq, im), the individual terms to form the product
1
Z ˛ ˛
˛1+j kz ∠θpf :z ˛
˛ ˛
˛ ksc−lv ˛
I 1
1
PQ ˛ ˛β
˛1+j kpq ∠θpf :pq ˛
˛ ˛
˛ ksc−lv ˛
1
IM „ «
1+ kkm ks
sc−lv
5
Li can represent any type (i.e. constant impedance, constant current, constant power and
induction motor loads) of load element.
6 1 ˛γ where γ = 1, 0, −2 ∼ −1 for constant impedance, constant current and
i.e. ˛˛ 1
∠θpf :lv ˛
˛
˛1+j k
sc−lv
1
constant power loads respectively, and “
ks
” for induction motor loads.
1+ k
sc−lv
Appendix M
The matrix equation (5.37) which gives the proposed methodology for evaluating influ-
ence coefficients [ki−x ] between any busbar i and other busbars x (x = 1, 2, ..., n x 6= i)
0 −1
[ki−x ](n−1)×1 ≈ [Y++:xz ](n−1)×(n−1) [Y++:xi ](n−1)×1 (M.1)
where,
0
Y++:xz ≈ Y++:xz + Y−−:x−im for x = z
0
Y++:xz = Y++:xz for x 6= z
For the purpose of evaluating the influence coefficients k1−2 and k1−3 between
240
241
system shown in Fig. 5.16, the matrices [Y++:xz ] and [Y++:xi ] are:
0.2020 − j0.4126 −0.1010 + j0.2063
[Y++:xz ] = S (M.2)
−0.1010 + j0.2063 0.3029 − j0.6189
−0.1010 + j0.2063
[Y++:xi ] = S (M.3)
−0.2020 + j0.4126
As busbar 3 supplies passive loads (i.e. km:3 = 0), the admittance Y−−:3−im ≈ 0.
• Case 1 - busbar 2 supplies passive loads or km:2 = 0. That is, the admittance
0
Y−−:2−im ≈ 0 implying that [Y++:xz ] ≈ [Y++:xz ]. Then, the substitution of (M.2)
(5.36) gives the admittance Y−−:2−im for klv:2 = 1, ksc−lvagg :2 = 18, ks:2 = 6.7,
0
Then, the matrix [Y++:xz ] can be obtained as:
0 0 0
[Y++:xz ] ≈ [Y++:xz ] + S (M.6)
0 −j0.1316
242
k1−2 0.75
= (M.7)
k1−3 0.91
Appendix N
66kV Sub-transmission
Data/Information
Note that the impedance values given in pu are based on a 100M V A base.
243
244
Table N.3: Generator and load bus data: three-phase MW and MVAr values
Busbar Generation Load
MW MVAr MW MVAr
S1 (reference bus) 0 0 0 0
S2 0 0 18.06 0.36
S3 0 0 18.27 7.11
S4 0 0 10.68 5.13
S5 13.50 0 0 0
S61 0 0 0.09 3.99
S7 0 0 33.99 14.52
S8 0 0 5.97 1.89
S9 0 0 1.65 0.24
1
This represent the operation (balanced) of a static Var compensator.
245
Table N.4: Voltage regulator data: impedances and secondary tap settings
Busbar Impedance Secondary tap
(pu) setting
S2 0.0014 + j0.0192 1.058
S7 0.0014 + j0.0192 1.111
Notations:
Table N.7: Lengths and impedances (Z−+ and Z−+ ) of the sub-transmission lines
Line Length Z−+ Z++
(km) (Ω) (Ω)
A 67.65 0.62∠840 24.09∠730
B 19.16 0.52∠300 8.36∠530
C 17.83 0.25∠3360 6.37∠720
D 71.49 0.74∠1780 24.81∠710
E 19.59 0.27∠650 9.63∠480
F 45.37 1.25∠300 22.81∠450
G 66.29 0.14∠340 32.50∠450
H 56.46 0.03∠470 28.80∠450
I 55.32 1.40∠300 18.98∠720
J 11.40 0.31∠1480 4.01∠720
K 15.57 0.08∠1210 5.47∠730
L 80.65 0.30∠1310 41.00∠450
M 83.20 1.91∠550 29.90∠640
N 21.16 0.45∠230 9.23∠590
• When a line (e.g. line J of the study system) is located in the downstream part
bar (e.g. consider S2 and S6 in the case of line J) has to satisfy the relationship
t
given by (7.3) where V−:send−tany
is approximately zero1 .
• Alternatively, a line (e.g. line A of the study system) which is located in the
upstream part of its network can introduce voltage unbalance at the upstream
busbars as well as the downstream busbars (e.g. consider the upstream busbar
S2 and the downstream busbar S7 in the case of line J) as the line is in a position
t
to take part of V−:send−tany
for all other lines.
• When a line (e.g. line B of the study system) is isolated from the rest of
the network, its influence at other main busbars, compared to that of a line
(e.g. line J of the study system) which is located in the main part of the network,
t
is somewhat reduced. Because, the negative sequence voltage V−:rec−tany
at any
other busbar (e.g. consider S2 in the case of line B) has to satisfy the relationship
(7.3) where there can exist several connections (e.g. lines E and A for S2) to
t t
the busbar resulting in several V−:send−tany
(e.g. V−:send−E ≈ Z−+:B I+:B , and
t
V−:send−A = 0 for S2) out of which at least one or more can be approximately
zero.
Voltages
lines A - N of the study system at the busbar S2. Table N.9 gives a comparison of
1
Because the voltage at the bulk supply is balanced, and the term Z++:tany I−:t/tany in (7.3) is
inconsiderable.
2
These, which are obtained using unbalanced load flow analysis, correspond to the selected time
stamp.
248
of the individual vectors given in Table N.8, and unbalance load flow analysis.
t
Table N.8: Negative sequence voltages V−:S2 caused by the individual lines A - N
at the busbar S2
t
Line (t) V−:S2 (V )
A 58.560∠−127.1760
B 14.691∠−120.6320
C 12.124∠151.7370
D 45.203∠−24.2310
E 4.401∠−139.6360
F 95.222∠−137.7970
G 0.950∠−134.2140
H 0.257∠−119.4290
I 54.234∠176.4530
J 14.402∠−47.7590
K 1.850∠−75.2630
L 1.409∠−68.1720
M 0.361∠−170.8340
N 13.425∠176.6590
lines
Table N.9: Resultant negative sequence voltage V−:S2 at the busbar S2
lines
V−:S2 (V )
Vector summation 207.735∠−133.9680
Unbalanced load flow analysis 205.703∠−134.3440
3
i.e. as a result of the interaction of all lines.
Appendix O
O.1 Introduction
Unbalanced load flow is an essential tool for analysing steady-state unbalanced power
system problems (e.g. voltage unbalance). Due to the need for careful representation
including the background information (Sections O.2 and O.3), and the representation
of power system components (Section O.4) employed for the formulation of load flow
equations. The key section, Section O.4.5, discusses existing models of three-phase
induction motors that have been used for unbalanced load flow studies, and proposes
two types of models which overcome the limitations associated with the existing
models.
249
250
References [1, 2, 3, 4, 5] propose two basic approaches for unbalanced load flow analy-
sis, which are based on the symmetrical component and the phase coordinate reference
frames respectively. Analysis of unbalanced power system problems has been tradi-
tionally based on the symmetrical component reference frame due to the advantages
of the availability of sequence impedances for power system components, and the
reference frame [1]. However, the use of the phase coordinate reference frame has been
facilitates the maintenance of the initial physical identity of the system with regard
to line parameters and variables such as nodal voltages and line currents [2, 4, 5].
The only drawback of this approach is that the size of the problem is significantly
large compared to that based on the symmetrical component reference frame [3].
ponent approach, an unbalanced load flow program that is based on the phase coor-
dinate reference frame is developed for specific applications of the work presented in
this thesis.
Flow Program
nections (e.g. single-phase and dual-phase loads) can be easily achieved. The
concept of specifying load flow constraints for each bus or each phase of a bus,
take various component connections into account. In view of the fact that the
power constraints such as the specified power generation and consumption are
that is referred to as ‘component level power flow constraints’ which allows the
three-phase induction motor responds differently to the positive and negative se-
quence applied voltages, whereas a typical residential load would exhibit nearly
identical behaviours with both the positive and negative sequence voltages.
The developed unbalanced load flow program incorporates the suggested concept
of the component level power flow constraints, and takes into account the second
The synchronous generator model used in the developed program is based on [4,
5], which takes the different machine responses for the positive, negative and zero
sequence current injections into account. As illustrated in Fig. O.1 and given by
(O.1), this is a positive sequence voltage source behind the generator admittance
a Eg
k b a2Eg m
[Yg]
c aEg
where,
[Eg ] = [Eg a2 Eg aEg ]t , matrix of the generator internal phase voltages (balanced)
¯ ¯
[Ikm:g ] = [Ikm:g−a Ikm:g−b Ikm:g−c ]t , matrix of the generator currents flow from side
k to side m
[Yg ] = T [Yg:seq ]T −1
Yg:0 0 0
[Yg:seq ] =
0 Yg:+ 0
0 0 Yg:−
1 1 1
T = 2
1 ¯a a¯
1 a a2
¯ ¯
Yg:0 , Yg:+ , Yg:− - zero, positive and negative sequence admittances respectively of
the generator
253
a = 1∠1200
¯
If the generator operates as a slack machine, the magnitude and the phase angle
where,
[T+ ] = 13 [1 a a2 ]
¯ ¯
|Vg:spec |, θg:spec - specified magnitude and phase angle respectively of the positive se-
If the generator operates as a PV machine1 , the three-phase active power and the
magnitude of the positive sequence voltage at the generator terminals are constrained:
where,
The exponential load model [6] which takes the voltage dependency of the active
and reactive power into account is employed. As illustrated in Fig. O.2, this is
load configurations.
λp
|Vk − Vm |
P = P0 (O.5)
V0
λq
|Vk − Vm |
Q = Q0 (O.6)
V0
where,
V0 - rated voltage
λp = λq = 0 for constant power loads, and different values can be chosen to represent
The active and reactive power consumed by this load are constrained:
c
Ikm:l (Vk − Vm ) = Pl:spec + jQl:spec (O.7)
where,
Pl:spec , Ql:spec - specified consumption of the active and reactive power respectively of
the load
255
k m
P + jQ
!
De
Zt:xy = Rearth +Rconductor +j k ln in Ω/m, when x = y, i.e. self impedance
Dxy
(O.8)
!
De
Zt:xy = Rearth + j k ln in Ω/m, when x 6= y, i.e. mutual impedance
Dxy
(O.9)
where,
Capacitor banks are represented as constant impedance elements which allow their
tion.
nected units. Each unit is modelled as an impedance in series with an ideal trans-
former having taps on the secondary. The equivalent circuit is shown in Fig. O.3
where [Avr:x ], [Bvr:x ] and [Cvr:x ] are 3 × 3 diagonal matrices of which the elements
i, j - represent the busbars at which the primary and the secondary respectively are
connected
tapxy - tap position, as a ratio of the primary voltage and the secondary voltage, of
any phase x (= a, b, c)
unbalanced load flow studies has been proposed [4, 5, 11, 13]. A simple model based
on the sequence equivalent circuits is given in [4, 5], where the positive sequence
257
Busbar i Busbar j
[Avr:x]
[Bvr:x] [Cvr:x]
the total real and reactive power drawn by the motor are equal to specified values. In
voltage source behind the known negative sequence impedance, and the voltage source
is solved to meet the power flow constraints stated above. The phase domain model
of the above, which is established by transforming the sequence elements into the
phase domain, is illustrated in Fig. O.4 and expressed by (O.13). The considered
Eim
a
k b a2Eim m
[Yim]
c aEim
where,
−1
[Yim ] = T [Yim:seq
]T
Yim:0 0 0
[Ysequence ] =
0 Yim:− 0
0 0 Yim:−
Yim:0 , Yim:− - zero and negative sequence admittances respectively of the induction mo-
tor
[Ikm:im ] = [Ikm:im−a Ikm:im−b Ikm:im−c ]t , matrix of the induction motor currents flow
Pim:spec , Qim:spec - specified motor input active and reactive power respectively
Reference [10] reports that although the active power drawn by a three-phase
induction motor is nearly independent of the supply voltage level until the point of
stalling, the reactive power is more sensitive to the voltage level. Fig. O.5, which
is reproduced using [10], illustrates the variation of the real (P) and reactive (Q)
power drawn by a typical induction motor with the voltage level justifying the above
statement4 . Hence, the power flow constraints used in the model proposed in [4, 5],
which controls the reactive power drawn by the motor at a specified value, does not
The subject of the modelling of three-phase induction motors for unbalanced load
flow studies has received increased attention recently by the IEEE Distribution System
4
The rated mechanical load and balanced supply voltages are assumed.
259
2.6
2.4 P
2.2 Q
2.0
1.8
P, Q (pu) 1.6
1.4
1.2
1.0
0.8
0.6
0.4
0.2
0.55 0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.95 1.00
Voltage (pu)
Figure O.5: Variation of the real (P) and reactive (Q) power with the supply voltage
level for a typical three-phase induction motor
Analysis Sub-committee with the view to provide reference test cases for developers
to validate the induction machine models under unbalanced conditions [11, 12, 13].
The model discussed in [11, 12] is also based on the sequence equivalent circuits,
and considers both the positive and negative sequence impedances as functions of
the motor slip which is taken as a new state variable to be determined such that the
input real power is equal to a specified value. The reactive power is then adjusted by
the load flow algorithm according to the existing supply voltage condition. This is a
valid representation only if the specified power corresponds to the actual operating
developed by the motor and the mechanical torque demanded by the load. However,
the nominal input real power corresponding to the rated motor speed has been used
as the specified power in [11, 12]. In other words, the rated mechanical load which
lead the motor to operate at the rated speed has been assumed. Although the real
voltage conditions until the point of stalling (see Fig. O.5), it is highly sensitive to
260
the changes in mechanical loading conditions. As an example, Figs. O.6 and O.7
illustrate the variation5 of the real (P) and reactive (Q) power and the motor speed
the motor is excited at the rated voltage (balanced). These demonstrate that the
changes in mechanical loading conditions have a high degree of influence on the real
and reactive power, whereas such changes have only a minor influence on the motor
speed. The rated mechanical loading condition assumed in the model discussed in
[11, 12] may not always arise in practice, and hence this model also has limitations
in generalised applications.
2.5
P kp:rated = 0.93
2.0 Q
P (MW), Q (MVAr)
1.5
1.0
0.5
0.0
kp 0.3 0.5 0.7 0.9 1.1
Loading
33 54 76 97 118
level (%)
Figure O.6: Variation of the real (P) and reactive (Q) power with kp (motor loading
levels corresponding to various kp is also given as a percentage to the rated output
power) for a 2250hp induction motor
In the remaining part of this section, two types of induction motor models which
reactive power components as shown in Fig. O.97 , together with a power flow
constraint.
1.2
1.0
Motor speed (pu)
0.8
0.6
kp:rated = 0.93
0.4
0.2
0.0
kp 0.3 0.5 0.7 0.9 1.1
Loading
33 54 76 97 118
level (%)
Figure O.7: Variation of the speed with kp (motor loading levels corresponding to
various kp is also given as a percentage to the rated output power) for a 2250hp
induction motor
Busbar k
a
b [Yim]
c
7
Where, Pim:a , Pim:b , Pim:c - active power drawn by the phases a, b and c respectively of the
motor, Qim:a , Qim:b , Qim:c - reactive power drawn by the phases a, b and c respectively of the motor.
262
Busbar k
a Pim:a + jQim:a
b Pim:b + jQim:b
c Pim:c + jQim:c
This is an extension to the model given in [11, 12], which is based on the sequence
equivalent circuits. The positive and negative sequence equivalent circuits of an in-
duction motor are shown in Fig. O.10. The positive (Zim:+ ) and negative (Zim:− ) se-
quence impedances, as functions of the motor speed, are given by (O.15) and (O.16)
1
Zim:+ = rst + jxst + 1 1 (O.15)
jxmag
+ rrt ωsyn
+jxrt
ωsyn −ωrt
1
Zim:− = rst + jxst + 1 1 (O.16)
jxmag
+ rrt ωsyn
+jxrt
ωsyn +ωrt
where,
rrt , xrt - rotor resistance and leakage reactance respectively (referred to the stator
side)
rrt ω syn
rst jxst ω syn − ω rt jxrt
jxmag
I
rrt ω syn
rst jxst ω syn + ω rt jxrt
jxmag
II
where,
V0:k , V+:k , V−:k - zero, positive and negative sequence voltages respectively at the
I0:im , I+:im , I−:im - zero, positive and negative sequence currents respectively drawn
by the motor
As the load flow method is based on the phase coordinate reference frame, the se-
quence domain equations (O.17) - (O.19) are transformed into the phase domain. Thus:
where,
Va:k , Vb:k , Vc:k - three phase voltages at the supply terminals of the motor
The motor speed ωrt associated with (O.17) - (O.19) is a new variable to be de-
complete the load flow formulation, which essentially has to be the power flow con-
straint under which the induction motor operates. The general situation where the
torque/power demanded by the rotating load is applied as the power flow constraint
instead of constraining the power drawn by the motor at the nominal values as in the
to the sum of the two torque components corresponding to the positive and negative
sequence voltage/current inputs. The positive (Tim:+ ) and negative (Tim:− ) sequence
torque components can be expressed in terms of the three phase voltages, motor
2 Zmag 2
(V + a V + aV )
1 a:k ¯ b:k ¯ c:k Zmag +Zst rrt
Tim:− =− Zmag Zst
(O.21)
3 + jxrt + ωrrt ω+ω ωsyn + ωrt
syn
Z +Z
mag st syn rt
where,
Zmag = jxmag
Considering pump systems, torque characteristics of the mechanical load (Tload ) can
be expressed as:
2
Tload = kl−1 + kl−2 ωrt (O.22)
266
where, kl−1 , kl−2 are constants for a given pump system. Then, the suggested power
where,
The functions f1 , f2 , f3 and f4 which give the complete load flow formulation for
a three-phase induction motor allow the load flow algorithm to determine the motor
speed and three stator phase voltages and currents thus naturally adjusting the input
The three-phase induction motor model proposed in this section is in line with the
exponential load model that has been used to represent induction motors in balanced
The real and reactive power drawn by each of the three phases of an induction
motor can be expressed using the stator phase voltages and the phase admittance
matrix as:
c
Pim:a + jQim:a Va:k 0 0 Va:k
P = 0 V [Yim ]c V (O.24)
im:b + jQim:b 0
b:k b:k
Pim:c + jQim:c 0 0 Vc:k Vc:k
267
where,
Yim:s Yim:m1 Yim:m2
[Yim ] =
Yim:m2 Yim:s Yim:m1
Yim:m1 Yim:m2 Yim:s
Referring to (O.24), the complex power per phase can be generally written as:
c
Pim:x + jQim:x = Yim:s |Vx:k |2 + Yim:m1
c c
Vx:k Vy:k c
+ Yim:m2 c
Vx:k Vz:k (O.25)
where, x, y, z - represent the three phases a, b and c. The real and reactive power
where,
θim:s , θim:m1 , θim:m2 - phase angles of Yim:s , Yim:m1 and Yim:m2 respectively
θx:k , θy:k , θz:k - phase angles of Vx:k , Vy:k and Vz:k respectively
268
For the case of balanced voltage angles8 (i.e. θx:k −θy:k = 1200 and θx:k −θz:k = −1200 ),
|Y
im:s |cos(−θim:s )
|V
x:k |
0
2
n
Pim:x−xx = Pim:x−xx n n n
(O.28)
|Yim:s | cos(−θim:s ) |Vx:k |
0 n
|Y
im:m1 |cos(−θim:m1 + 1200 ) |Vx:k ||Vy:k |
Pim:x−xy = Pim:x−xy n n n n
(O.29)
|Yim:m1 |cos(−θim:m1 + 1200 ) |Vx:k ||Vy:k |
0 n
|Y
im:m2 |cos(−θim:m2 − 1200 ) |Vx:k ||Vz:k |
Pim:x−xz = Pim:x−xz n n n n
(O.30)
|Yim:m2 |cos(−θim:m2 − 1200 ) |Vx:k ||Vz:k |
|Y
im:s |sin(−θim:s )
|V
x:k |
0
2
Qim:x−xx = Qnim:x−xx n n n
(O.31)
|Yim:s |sin(−θim:s ) |Vx:k |
0
|Y
im:m1 |sin(−θim:m1 + 1200 ) |Vx:k ||Vy:k |
Qim:x−xy = Qnim:x−xy n n n n
(O.32)
|Yim:m1 |sin(−θim:m1 + 1200 ) |Vx:k ||Vy:k |
0
|Y
im:m2 |sin(−θim:m2 − 1200 ) |Vx:k ||Vz:k |
Qim:x−xz = Qnim:x−xz n n n n
(O.33)
|Yim:m2 |sin(−θim:m2 − 1200 ) |Vx:k ||Vz:k |
where,
0
- refers to the condition of balanced voltage angles
Qnim:x−xx = |Vx:k
n 2 n
| |Yim:s n
|sin(−θim:s )
8
In most practical circumstances unbalance in phase voltages arise mainly due to the unbalance
in their magnitudes. Thus, this condition of balanced voltage angles can be considered as a general
case.
269
n
Qnim:x−xy = |Vx:k n
||Vy:k n
||Yim:m1 n
|sin(−θim:m1 + 1200 )
Qnim:x−xz = |Vx:k
n n
||Vz:k n
||Yim:m2 n
|sin(−θim:m2 − 1200 )
0 0
Referring to (O.28) - (O.33), the power components Pim:x−xx - Qim:x−xz can be written
0 n
Y |Vi:k ||Vj:k |
Sim:x−ij = Sim:x−ij (O.34)
Y n |V n ||V n |
i:k j:k
where,
0 0 0
Sim:x−ij - represents any Pim:x−ij or Qim:x−ij where i, j = x, y, z
n 0
Sim:x−ij - nominal value of Sim:x−ij (i.e. at the rated voltage and the rated motor
speed)
For the purpose of illustrating the characteristics of the various power and admit-
tance elements in (O.34), a 60Hz, 3hp, 220V induction motor with the parameters
n
given in Table O.1 [15] is used. Table O.2 gives the power components Px−xx - Qnx−xz
ωrt
respectively with n
ωrt
over a range of ωrt corresponding to 125% - 25% of the rated
Y
= ωrt
λsp:s−ij
(O.35)
Y n ωn rt
270
n
Table O.2: Power components Px−xx - Qnx−xz for the 3hp, 220V motor
P n (W) Qn (VAr)
x − xx 1834.94 2968.26
x − xy 1923.00 -2483.74
x − xz -2842.67 166.41
1.10
Approximation: y = x-3.3828
Actual variation
1.05
Yim:s cos( −θ im:s )
Yimn :s cos( −θ imn :s )
1.00
0.95
0.90
0.85
0.98 0.99 1.00 1.01 1.02 1.03 1.04 1.05
ω rt
ω rtn
which is a constant for a given motor. Hence, (O.34) can be written as:
0
ω λsp:s:ij |V ||V |
n rt i:k j:k
Sim:x−ij = Sim:x−ij n n n
(O.36)
ωrt |Vi:k ||Vj:k |
A speed range corresponding to 125% - 25% of the rated motor loading level is
usually sufficient to cover most practical circumstances, within which the accuracy
can be achieved if the speed range under consideration can be further narrowed. The
speed coefficients corresponding to the power components Px−xx - Qx−xz for a range
1.2
Approximation: y=x-5.5278
Actual variation
Yim:m 2 cos( −θ im:m 2 − 120 0 )
Yimn :m 2 cos( −θ imn :m 2 − 120 0 )
1.1
1.0
0.9
0.8
0.7
0.98 0.99 1.00 1.01 1.02 1.03 1.04 1.05
ω rt
ω rtn
The power components Px−xx - Qx−xz which represent the behaviour of an induc-
tion motor under both unbalanced voltage magnitudes and unbalanced voltage angles
272
Table O.3: Speed coefficients corresponding to the power components Px−xx - Qx−xz
for a range of induction motors
Motor specification
λsp:s−ij 3hp, 220V 50hp, 460V 500hp, 2.3kV 2250hp, 2.3kV
λsp:p−xx -3.3828 -5.9151 -40.658 -94.6090
λsp:p−xy -2.9136 -3.0044 -12.089 -19.8360
λsp:p−xz 2.0121 2.7340 13.9710 23.2230
λsp:q−xx -0.0848 -0.4081 -2.9381 -4.2029
λsp:q−xy 0.2243 0.6279 5.6215 8.3997
λsp:q−xz -5.5278 3.8814 10.469 11.9090
0 0
can be expressed in terms of the power components Px−xx - Qim:x−xz (which represent
0
Pim:x−xx = Pim:x−xx (O.37)
0 0
Pim:x−xy = Pim:x−xy cos(θxy:k ) − Qim:x−xy sin(θxy:k ) (O.38)
0 0
Pim:x−xz = Pim:x−xz cos(θxz:k ) − Qim:x−xy sin(θxz:k ) (O.39)
0
Qim:x−xx = Qim:x−xx (O.40)
0 0
Qim:x−xy = Qim:x−xy cos(θxy:k ) + Pim:x−xy sin(θxy:k ) (O.41)
0 0
Qim:x−xz = Qim:x−xz cos(θxz:k ) + Pim:x−xy sin(θxz:k ) (O.42)
where,
The power flow constraint considered in the impedance type model, which is given
where,
The total gross electromechanical power developed by the motor (i.e. Pim:+ + Pim:− )
can be expressed in terms of the input active power components (given by (O.26)) in
Based on the observation that the real power drawn by an induction motor is nearly
independent of the supply voltage condition until the point of stalling (Fig. O.5),
the motor efficiency ηim can be expressed in terms of the motor parameters and the
speed as:
Pgross at V n
ηim = (O.45)
Pin at V n
where,
Pgross at V n - total gross electromechanical power developed by the motor at the rated
voltage V n (balanced)
Z
2
V n Z mag
mag +Zst ωrt rrt
Pgross at V n = 3 Zmag Zst
rrt ωsync
Zmag +Zst +jxrt + ωsync −ωrt ωsync −ωrt
Pin at V n - motor input real power at the rated voltage V n (balanced)
n2
Pin at V n = 3 real Z|Vc |
im:+
The motor efficiency ηim , as a function of the motor speed ωrt , can be shown to
ple, Fig. O.13 illustrates the variation of ηim with ωrt obtained using (O.45) and the
274
approximation (O.46) for the 3hp, 220V motor, where the speed range considered
where, λef f −1 , λef f −2 - efficiency coefficients, which are constants for a given mo-
tor. Table O.4 gives these efficiency coefficients for the range of induction motors
considered earlier. Thus, the power flow constraint given by (O.43) can be modified
Collecting all related equations together, the complete load flow formulation rep-
n
The model requires the nominal power components Px−xx - Qnx−xz , speed coefficients
αsp:p−xx - αsp:q−xz , efficiency coefficients λef f −1 , λef f −2 and mechanical loading char-
involved with this model in comparison to the impedance type model. However, this
model makes the load flow formulation and the derivation of the jacobian matrix
much simpler, especially in the case where the condition of balanced voltage angles
can be applied.
275
97 Approximation: y = 95.776x0.9019
Actual variation
95
93
ηim (%)
91
89
87
85
0.93 0.94 0.95 0.96 0.97 0.98 0.99 1.00
ωrt (pu)
Figure O.13: Variation of ηim with ωrt for the 3hp, 220V motor
276
Cases 1 - 3 listed below demonstrate that the PQ type model behaves almost similar
assumptions10 .
Fig. O.14 illustrates the variation of the per phase input active and reactive
power with the motor loading level (given as a % to the rated motor load) for
the PQ type model, of the motor efficiency is being independent of the supply
voltage condition
Fig. O.15 illustrates the variation of the per phase input active and reactive
power components with the motor loading level (given as a % to the rated
motor load) for the 3hp, 220V motor excited at reduced (|V+:k | = 0.9pu) and
• Case 3 - demonstration of the validity of the PQ type model for large motors
Fig. O.16 illustrates the variation of the per phase input active and reactive
power components with the motor loading level (given as a % to the rated motor
load) for a 2250hp, 2.3kV motor [15] excited at reduced (|V+:k | = 0.9pu) and
600
400
200
0
20 40 60 80 100 120
Loading level ( %)
Figure O.14: Variation of the per phase input active and reactive power with the
motor loading level for the 3hp, 220V motor excited at the rated voltage (balanced)
1400
Pim:b - PQ type
Pim:c - impedance type
800 Pim:c - PQ type
Qim:a - impedance type
600 Qim:a - PQ type
Qim:b - impedance type
Qim:b - PQ type
400
Qim:c - impedance type
Qim:c - PQ type
200
0
20 40 60 80 100 120
Figure O.15: Variation of the per phase input active and reactive power components
with the motor loading level for the 3hp, 220V motor excited at reduced and unbal-
anced voltages
278
0.9
0.0
20 40 60 80 100 120
Loading level (%)
Figure O.16: Variation of the per phase input active and reactive power components
with the motor loading level for a 2250hp, 2.3kV motor excited at reduced and un-
balanced voltages
kp:rated = 0.87
1000
800
Pim:a (W)
0
0.15 0.4 0.65 0.9 1.15
kp
Figure O.17: Variation of Pim:a with kp for the existing and proposed induction motor
models
279
Figs. O.17 and O.18 illustrate the variation of the input real and reactive power
respectively for the phase a with kp of pump systems obtained employing the ex-
isting and proposed models in relation to the 3hp, 220V motor excited at reduced
the model discussed in [11, 12]14 behaves similar to the proposed models only when
the motor supplies the rated mechanical load15 . The model given in [4, 5]16 does not
represent the motor behaviour accurately even when the motor is loaded with the
450
Qim:a (VAr)
350
Figure O.18: Variation of Qim:a with kp for the existing and proposed induction motor
models
13
Where, Va:k = 0.84∠00 pu, Vb:k = 0.92∠ − 1200 pu, Vc:k = 0.92∠1200 pu.
14
This constraints the real power drawn by the motor at the rated value, and allows the reactive
power to be adjusted by the load flow algorithm according to the existing supply voltage condition.
15
Note the deviation in both Pim:a and Qim:a associated with the model given in [11, 12] compared
to the proposed models when kp 6= kp:rated .
16
This constraints both the real and reactive power drawn by the motor at the rated values.
17
Note the deviation in Qa at kp:rated associated with the model given in [4, 5] compared to the
other models.
280
interaction between them are obtained using the component branch currents by:
where,
• for x 6= y, Yxy is equal to the negative value of the phase admittance matrix
Ik = [Ia:k Ib:k Ic:k ]t , any element of the matrix [Ik ], which is equal to the summation
The load flow equations are solved employing the well established Newton-Raphson
1991.
7. P.A.N. Garcia, J.L.R. Pereira, and S.Jr. Carneiro. Voltage Control Devices
Models for Distribution Power Flow Analysis. IEEE Trans. on Power Systems,
8. Paul M. Anderson. Analysis of Faulted Power Systems. John Wiley & Sons,
1995.
282
10. Yuin-Hong Liu, Wei-Jen Lee, and Mo-Shing Chen. Incorporating Induction Mo-
tor Model in a Load Flow Program for Power System Voltage Stability Study.
11. D.R.R. Penido, L.R. Araujo, S. Carneiro Jr., and J.L.R. Pereira. Unbalanced
22 June 2006.
Test Cases - Steady State Solutions. In Proc. IEEE Power Engineering Society
13. R.C. Dugan. Induction Machine Modeling for Distribution System Analysis
14. Daniel Ruiz-Vega, Tomás I. Asiaı́n Olvares, and Daniel Olguı́n Salinas. An
15. Paul C. Krause, Oleg Wasynczuk, and Scott D. Sudhoff. Analysis of Electric