Abstract:
Distribution system brings electricity to the end consumers. Conventionally, it
did not have any generating resources embedded in it. Thus, in conventional dis tribution system power flows are expected to be unidirectional. The distribution
systems are commonly modelled as radials of certain length distributing power in
one direction i.e. from source end to the load ends. The issues such as safety,
protection and coordination, network planning and operation, power quality and
design of system components are dictated by this radial architecture of the distri bution systems.
Though, radial distribution architecture is simplest and cheapest, double cir cuit or loop architecture gives redundancy in the system. Modern distribution
systems have embedded/active generations, and bidirectional flows. The possi bility of bidirectional flows has given rise to several challenges for the protection
of distribution systems. This complexity of modern distribution systems calls for
intelligent distribution management systems (DMS) having large amount of In formation and Communication Technology (ICT) components. The deployment
of intelligent DMS has also given way to operation of the distribution systems in
meshed configuration.
A distribution system serves unbalanced three-phase, two-phase and single phase loads. A physical transmission happens over untransposed conductors. This
fact along with variation in the types of loads naturally makes the distribution sys tems characteristically unbalanced. Other problems associated with distribution
systems are under-voltage and over-voltage. The increase in voltage above 110%
of the nominal voltage for more than 1 minute is termed as over-voltage. This
phenomenon usually occurs due to improper compensation and outage of heavi iiily/lightly loaded lines. The under voltage limits are below 90% of the nominal
voltage for more than 1 minute. The under-voltage phenomenon is quite common
in distribution systems due to its radial configuration especially at the far ends of
distribution systems. There are several research papers in the literature addressing
the mitigation of under-voltage / over- voltage problems through reactive power
compensation. Capacitive compensation can also be seen from the perspective of
power factor correction in the system, as the present distribution systems caters
mostly the lagging loads. The low power factor forces high currents in the sys tems calling for larger conductor sizes and larger equipment sizes and increase in
system losses. Higher currents at low power factors also adversely affect voltage
regulation.
The problem of under-voltage/over-voltage and losses may also be addressed
through reconfiguration of distribution network. Reconfiguration is essentially
done using tie-lines provided in the system for redundancy against faults to ensure
service of loads by rerouting the power to the systems loads in the event of line
outages. The rerouting may result in a new configuration giving lower losses hence
option of reconfiguration can also be exercised for loss minimization.
Other problems associated with distribution systems are voltage sag/swell
which are short-time (greater than 10 ms and less than 60 s) phenomenon. Re duction in voltage level in the range of 10 to 90 % of the nominal RMS voltage
is considered as voltage sag and rise in voltage from 110 to 180 % is termed as
voltage swell. Voltage swells are usually caused by system switching operations,
sudden change in ground reference, single line-to- ground fault leading to voltage
rise in unfaulted phases for the duration of fault. Similarly sags are caused by
faults on distribution networks, faults on consumer equipment, and switching in
of heavy loads.
Other short-time phenomena in distribution systems are voltage interrup tion, voltage transients, voltage fluctuations and flicker, dc offset, and harmonic
distortions. Increased presence of small generating resources embedded within
distribution systems has made many of the assumption incorrect with regards to
the radial nature of the distribution systems. Distribution networks are no longer
ivpassive networks; thus basic premise that a radial feeder would become dead upon
opening of breaker, cannot be taken for granted as embedded generation would
still be feeding the line (we can discuss these in future work).
The fact that variation in the types of loads naturally makes the distribution
systems characteristically unbalanced, affects the distribution system in various
ways. The working of several relays in the distribution systems are sensitive to the
sequence currents in the systems and have a tendency to maloperate. Unbalance
of phases in the distribution systems can manifest in the form of current unbalance
and/or voltage unbalance. The voltage unbalance in the system can happen due
to unbalance of parameters of components such as generators, transformers, and
feeders (due to untransposed lines). The voltage unbalance also occurs due to un balanced phase currents leading due to asymmetrical voltage drops in individual
phases. Thus current unbalance also leads to voltage unbalance. Current unbal ance mainly occurs due to single phase loads or two-phase loads and also occurs
in balanced load condition in the presence of voltage unbalance. Thus these two
phenomena are tied together.
The main substation transformers are one of the costliest devices in the dis tribution system. The MVA capacity is divided equally on three phases hence, the
per-phase capacity is one-third of the MVA capacity of the transformer. As the
time progresses the system load grows, however usually the load growth may not
be equal on three phases. When the system loads are severely unbalanced, then if
any of the phases gets loaded up to its capacity, the further loading of the main
transformer gets restricted even when there may be sufficient capacity available
on other two phases. This may lead to load shedding on the loaded phase. Thus,
in any system the load shedding limit would always be less than the total MVA
capacity of the transformer.
The phase unbalance in a system can create large voltage differences among
the phases at a bus. The unbalanced phase voltages lead to unbalanced currents
in three-phase loads at the customer site. This further restricts the transformer
capacity at the customer site. Also, the unbalanced three phase load means higher
negative- and zero-sequence current leading to losses for consumers which is not
vdue to consumer. Also, in several cases the severe unbalance may not allow the
consumer to use the machinery effectively because the voltage unbalance results
in reverse magnetic fields which in turn results in rise of the temperature of ma chine winding and fall in the output torque. The unbalance in the system creates
negative- and zero-sequence currents to flow in the system. The neutral conductor
in the system is designed to carry small current compared to other phases assum ing that the phases are balanced. The increased unbalance may ask for increased
rating of neutral conductors. The neutral has a higher resistance compared to
other phases and therefore, increase in neutral current increases the neutral losses
also.
Distribution systems are getting equipped with Information and Communi cation Technologies (ICT). Various problems of distribution systems which were
considered acceptable in conventional distribution systems have now become solv able. The problem of phase unbalance was not much severe as the amount of load
served were less, also the protection philosophy was quite systems and unbalance
was not seen as a major hindrance to the system operation. As the distribution
systems grew in size, capacity, and in terms of information and communication
infrastructure, the problem of unbalance became significant as well as solvable.
This work considers the availability of communication, control and computa tional infrastructure for solving the problem of unbalance in distribution systems.
The objective of the thesis can be enumerated as follows. 1. Development and
testing of reliable and robust three-phase power flow algorithm for unbalanced
systems. The developed power flow is envisaged to have capacity to handle large
number of PV buses which would be feature of upcoming distribution systems. 2.
Development of phase balancing algorithm for conventional distribution systems
considering switchability of loads from one phase to the other. Further, the test of
efficacy of the developed algorithm is to be carried out in terms of amount of un balance mitigation. 3. Development of phase balancing algorithm for distribution
systems with large number of embedded single-phase generations. Further, the
test of efficacy of the developed algorithm is to be carried out in terms of amount
of unbalance mitigation.
viOutline of the thesis is as follows. Chapter 1 introduces the topic of the thesis
along with survey of relevant literature. In chapter 2, an improved CINR load flow
method has been presented. The developed load flow uses the new equations to
model the PV bus in current injection power flow formulation, which is based
on real and imaginary parts of simple multiplication of voltages and currents of
PV buses. The new CINR load flow technique decreases the required number of
equations and also recovers the convergence property of revised current injection
load flow methods same as conventional NR (CNR) method in the case of PV nodes
particularly. At heavy load and large R/X ratio conditions also the convergence
characteristics improved. The results have also demonstrated that the computation
time of Mod-CINR is less than the fast decoupled NR (FDBX) methods in the
absence of PV buses in systems. All the experiments suggest that the performance
of the improved method in comparison with other techniques is better in terms of
convergence, efficiency, sensitivity, and reliability. In chapter 3, the optimization
techniques used in the chapter 4 and 5 have been described in the context of
the problem addressed in the thesis. In the chapter 4, it has been demonstrated
that re-phasing is sensitive to voltage-dependency of loads. It was found that the
voltage improvement due to re-phasing increases the load demand. The increase in
load demand at individual buses may be insignificant, but for a system as a whole
it is significant enough to reverse the advantage of loss reduction conventionally
expected due to re-phasing. It was found that the system after re-phasing may
suggest more MVA margins at the main substation if appropriate load model is
not considered. Two optimization algorithms, i.e., PSO and Butterfly Optimizer
(BO), have been successfully applied for feeder re-phasing of radial distribution
network. The PSO based method was compared with GA based method on a
system reported in the literature to establish the effectiveness of the approach. To
further validate the effectiveness of the proposed approach a 24-hour load pattern
was taken. It was found that the system could be balanced substantially in terms
of phase currents, phase voltages and losses per phase.
In chapter 5, DG planning for the purpose of phase balancing has been pro posed. The planning is approached in two stages. In stage 1, the optimal DG
viilocations (phase and bus) and sizes are obtained for peak load scenario. The loca tion are then considered to be fixed and the sizes obtained are taken as maximum
available DG real and reactive capacity at the given buses. In the stage 2, the DGs
are optimally scheduled hourly (in term of phase and size) for 24-hour loading sce nario to obtain the phase balancing. It has been established that an effective phase
balancing can be achieved with help of single-phase DGs scheduled in this fashion.
The reduction in losses and the improved voltage profile are added advantages.
Detailed analysis to ascertain the effect of voltage dependency of loads on optimal
scheduling of DGs establishes that for pragmatic optimal solutions, the voltages
dependency of loads must be considered albeit in approximate sense rather than
doing the same using constant power load model.
In this chapter also BO and PSO have been successfully applied for single phase DG re-phasing of radial distribution network.