Professional Documents
Culture Documents
Introduction
Electricity is the most preferred used form of energy used in industry, homes, businesses and transportation. It can be easily and efficiently transported from the production centers to the point of
use. It is highly flexible in use as it can be converted to any desired form like mechanical, thermal,
light, chemical etc.
An electrical power system is made up of many components connected together to form a large,
complex system that is capable of generating, transmitting and distributing electrical energy over
large areas.
1.1
The basic structure of modern day power system is shown in Fig. 1.1. A power system is usually
divided into three parts: generation, transmission and distribution system.
1.1.1
Generation
Electricity is produced by converting the mechanical energy into electrical energy. In majority of
cases, the mechanical energy is either obtained from thermal energy or provided by the flowing water.
The main sources of thermal energy sources are coal, natural gas, nuclear fuel and oil. The use of
non-fossil fuels such as wind, solar, tidal, and geothermal and biogas in electricity generation is also
increasing. Hydro-power is the main non-thermal source of mechanical energy used in electricity
generation. The conversion of mechanical to electrical energy is done using synchronous generators
in majority of power plants. Few wind generation systems use induction generators. The power is
usually generated at low voltage, between 11 and 35 KV, and then fed into the transmission system
using a step up transformer.
1.1.2
Transmission system
The electricity is generated in bulk in the generating stations and then transmitted over long distances
to the load points. The transmission system interconnects all the generating stations and major load
centers in the system. It forms the back bone of the power system. Since the power loss in a
transmission line is proportional to the square of line current, the transmission lines operate at the
1
a grid. The grid is the network of multiple generating resources and several layers of transmission
network. The interconnections of power systems offer the following advantages.
(a) Quality: The voltage profile of the transmission network improves as more generators contribute
to the system, resulting in an increased total system capability. This also improves the frequency
behavior of the system following any load perturbation due to increased inertia of the system.
(b) Economy: In interconnected systems, it is possible to reduce the total set of generating plants
required to maintain the desired level of generation reserve. This results in reduction of operational and investment costs. Also, operational (including plant start-ups and shut down) and
generation scheduling of units can be more economically coordinated.
(c) Security: In case of emergency, power can be made available from the neighboring systems and
each system can benefit even when individual spinning reserves may not be sufficient for isolated
operation.
1.1.3
Distribution system
The distribution represents the final stage of power transfer to the individual consumer. The distribution network is generally connected in a radial structure. The primary distribution voltage is
typically between 11 KV and 33 KV. Small industrial customers are supplied by primary feeders at
this voltage level. The secondary distribution feeders supply residential and commercial at 415/240
V. Small generating plants located near the load centers are usually connected to sub-transmission
or distribution system directly.
A power system operates in a normal state, if the following conditions are satisfied:
The bus voltages are within the prescribed limits.
The system frequency is within the specified limits.
The active and reactive power balance exists in the system.
However, the system load varies continuously and hence, in order to ensure satisfactory system
operation, proper controls have to be provided in a power system.
1.1.4
The various elements of power system operation and control are shown in Table 1.1 along with the
time-scale of operation.
1.2
A properly designed and operated power system must meet the following requirements:
3
1
2
3
4
5
6
7
Time period
Multi seconds
Few seconds to few minutes
Few minutes to few hours
Few minutes to few hours
Few hours to few weeks
One month to one year
One year to 10 years
(a) The system must have adequate capability to meet the continuously varying active and reactive
power demand of system load. This requires maintaining and approximately controlling adequate
spinning reserve of active and reactive power at all time instants.
(b) The system should be designed and operated so as to supply electrical energy at minimum cost
and with minimum adverse ecological impact.
(c) The electrical power supplied to the consumers must meet certain minimum quality standards
with respect to the following:
i) The network frequency should be maintained within a range of 3 percent of its nominal
value.
ii) The voltage magnitudes should be maintained within a range of 10 percent of the corresponding nominal value at each network bus bar.
iii) The supply should meet a desired level of reliability to ensure supply continuity as far as
possible.
(d) It should maintain scheduled tie-line flow and contractual power exchange.
To meet the requirements at points i), ii) and iii) above, several levels of controls incorporating
a large number of devices are needed. These controls are as shown in Fig. 1.2.
A brief explanation of various power system controllers is given next.
1.2.1
The controls provided in generating units consist of prime mover control and excitation controls as
shown in Fig. 1.3. The controls are also called as local frequency control (LFC) and automatic
voltage control (AVC).
These controllers are set for a particular operating condition and maintain the frequency and
voltage magnitude within the specified limits following small changes in load demand. If the input to
the prime mover is constant, then an increase in the active power of load at the generator terminals
results in a drop in the prime mover speed. This then, causes a reduction in the frequency. On
4
1.2.2
In LFC, two feedback loops namely, primary and secondary loops are provided. Both the loops
help in maintaining the real power balance by adjusting the turbine input power. The primary LFC
loop senses the generator speed and accordingly controls the turbine input. This is a faster loop
5
1.2.3
In AVC, the bus voltage is measured and compared to a reference. The resulting error voltage is
then amplified and applied to the excitation control system. The output of the exciter controls the
generator field current. An increase in the reactive power load of the generator causes the terminal
voltage to decrease and this results in generation of voltage error signal. The amplified error signal
then increases the exciter field current which in turn increases the exciter terminal voltage. This
increases the generator field current, which results in an increase in the generated emf. The reactive
power generation of the generator is thus increased and the terminal voltage is brought back to its
nominal value.
The generation control maintains the active power balance in the system. It also controls the
division of load active power between the generators in the system to ensure economic operation.
1.2.4
Economic dispatch
Economic operation and planning of electric energy generating system has been accorded due importance by the power system operators. Power systems need to be operated economically to make
6
electrical energy cost-effective to the consumer and profitable for the operator. The operational
economics that deals with power generation and delivery can be divided into two sub-problems.
One dealing with minimum cost of power generation and other dealing with delivery of power with
minimum power loss. The problem of minimum production cost is solved using economic dispatch.
The main aim of economic dispatch problem is to minimize the total cost of generating real
power at different plants in the system while maintaining the real power balance in the system.
For system having hydro-plants, a coordinated dispatch of hydro-thermal units is carried out. The
economic dispatch and minimum loss problems can be solved by means of optimal power flow (OPF)
method. The OPF calculations involve a sequence of load flow solutions in which certain controllable
parameters are automatically adjusted to satisfy the network constraints while minimizing a specified
objective function.
The power system control objectives are dependent on the operating state of the system. Under
normal operating conditions, the controller tries to operate the system as economically as possible
with voltages and frequency maintained close to nominal values.
But abnormal conditions like outage of a larger generator, of a major transmission line or sudden
increase or reduction of system load can cause havoc in the system, if not properly controlled.
Different operating objectives have to be met in order to restore the system to normal operation
after the occurrence of such contingencies.
1.2.5
For the analysis of power system security and development of approximate control systems, the
system operating conditions are classified into five states: normal, alert, emergency, in extremis and
restorative. The state and the transitions between them are shown in Fig. 1.4.
generation eqauls total system load) and inequality(bus voltages and equipment currents within
the limits) constraints being satisfied. In this state, a single contingency cannot disrupt the system
security and cannot cause any variable to violate the limit. The system has adequate spinning
reserve.
Alert state: If the security level of the system falls below some specified threshold, the system
then enters the alert state and is termed as insecure. The system variables are still within limits.
This state may be brought about by a single contingency, large increase in system load or adverse
weather conditions. Preventive control steps taken to restore generation or to eliminate disturbace
can help in restoring the system to the normal state. If these restorative steps do not succeed, the
system remains in the alert state. Occurence of a contingency with the system already in alert state,
may cause overloading of equipments and the system may enter emergency state. If the disturbance
is very severe, the system may enter into extremis state directly from alert state.
Emergency state: If the preventive controls fail or if a severe disturbance occurs, the system
enters emergency state. The transition to this state can occur either from normal state or alert state.
In this state the balance between generation and load is still maintained (equality constraints still
satisfied) and the system remains in synchronism. Some components are however overloaded(some
inequality constrints violated). Failure of these components results in system disintegration. Emergency control actions like disconnection of faulted section, re-routing of power excitation control,
fast valving, and load curtailment have to be taken. It is most urgent that the system be restored
to normal or alert state by means of these actions.
In-extremis state: If the emergency control actions fail when the system is in emergency state,
then the system enters into in-extremis state. The system starts to disintegrate into sections or
islands. Some of these islands may still have sufficient generation to meet the load. The components
are overloaded and the active power balance is also disrupted. Overloaded generators start tripping
leading to cascade outages and possible blackout. Control actions, such as load shedding and
controlled system operation are taken to save as much of the system as possible from a widespread
blackout.
Restorative state: The restorative state represents a condition in which control action is being
taken to restart the tripped generators and restore the interconnections. The system transition can
be either to normal or alert state depending on system conditions. The sequence of events that result
in system transition from normal to in-extremis state may take from few seconds to several minutes.
Bringing the system back to normal through the restorative state is an extremely time consuming
process and may last for hours or may be days. A large generator may require many hours from
restart to synchronization. The switched off loads can be picked up gradually and resynchronization
of operating islands to the grids is also a time consuming process.
The control actions may be initiated from the central energy control centre either through operators or automatically.
1.2.6
Unit Commitment
8
The total load in the power system varies throughout a day and its value also changes with the
day of the week and season. Hence, it is not economical to run all the units available all the time.
Thus, the problem of unit commitment is to determine in advance, the start and the shut down
sequence of the available generators such that the load demand is met and the cost of generation is
minimum.
1.2.7
Maintenance scheduling
Preventive maintenance has to be carried out on power system components to ensure that they
continue to operate efficiently and reliably. Generators are usually put on maintenance once every
year. Their maintenance has to be so scheduled such that the available generation is sufficient to
meet the system load demand. The problem of maintenance scheduling deals with the sequencing of
generator maintenance such that sufficient generation is always available to meet the load demand
and the cost of maintenances and cost of lost generation is minimum.
1.3
System planning
To meet ever increasing load demand, either new power systems have to be built or the existing
power systems are expanded by adding new generators and transmission lines. Many analyses must
be performed to design and study the performance of the system and plan expansion. To study the
system feasibility and performance, the following analyses need to be carried out:
(a) Load flow analysis
(b) Fault analysis/short circuit studies
(c) Stability studies
(d) Contingency analysis
1.3.1
The load flow analysis involves the steady state solution of the power system network to determine power flows and bus voltages of a transmission network for specified generation and loading
conditions. These calculations are required for the study of steady state and dynamic performance
of the system.
The system is assumed to be balanced and hence, single phase representation is used. These studies are important in planning and designing future expansion of power system and also in determining
the best operation of the existing systems.
1.3.2
Fault studies
In these studies the line currents and bus voltages of a system are calculated during various
types of faults. Faults on power system are divided into balanced and unbalanced faults. Three
phase symmetrical faults are balanced faults in which the system retains its balanced nature. The
unbalanced faults are single line to ground fault, line to line fault and double line to ground fault.
The fault currents values are useful in relay setting and co-ordination as well as for selecting the
proper rating of the circuit breakers.
1.3.3
Stability studies
The stability studies ascertain the impact of disturbances on the electrochemical dynamic behavior of the power system. These studies are of two types; small signal stability study and transient
stability study.
The small signal stability studies deal with the behavior of a system following any small disturbances like small change in load, small change in AVR gain etc. As the disturbance is small, the
equations that describe the dynamics of the power system are linearized for the purpose of analysis.
The system is small signal stable for a particular operating point, if following a small disturbance it
returns to essentially the same steady state operating condition.
Transient stability study deals with the response of a power system subjected to a large disturbance such a short circuit, line tripping or loss of large genration. In this study the equations
describing system dynamics are solved using numerical techniques. The power system transient stability problem is then defined as that of assessing whether or not the system will reach an acceptable
steady state operating point following a large disturbance.
We will now start with the study of load flow analysis technique from the next lecture.
10
Module 2
Load Flow Analysis
AC power flow analysis is basically a steady-state analysis of the AC transmission and distribution
grid. Essentially, AC power flow method computes the steady state values of bus voltages and line
power flows from the knowledge of electric loads and generations at different buses of the system
under study. In this module, we will look into the power flow solution of the AC transmission grid
only (the solution methodology of AC distribution grid will not be covered). Further, we will also
study the power flow solution technique when an HVDC link is embedded into an AC transmission
grid. Also, we will be considering only a balanced system in which the transmission lines and loads
are balanced (the impedances are equal in all the three phases) and the generator produces balanced
three phase voltages (magnitudes are equal in all the phases while the angular difference between
any two phases is 120 degree).
2.1
Basically, an AC transmission grid consists of, i) synchronous generator, ii) loads, iii) transformer
and iv) transmission lines. For the purpose of power flow solution, synchronous generators are not
represented explicitly, rather their presence in implicitly modeled. We will look into the implicit
representation of synchronous generators a little later. However, the other three components are
modeled explicitly and their representations are discussed below.
2.1.1
Loads
As we all know, loads can be classified into three categories; i) constant power, ii) constant impedance
and iii) constant current. However, within the normal operating range of the voltage almost all the
loads behave as constant power loads. As the objective of the AC power flow analysis is to compute
the normal steady-state values of the bus voltages, the loads are always represented as constant
power loads. Hence, at any bus k (say), the real and reactive power loads are specified as 100 MW
and 50 MVAR (say) respectively. An important point needs to be mentioned here. As the loads
are always varying with time (the customers are always switching ON and OFF the loads), any
specific value of load (MW and/or MVAR) is valid only at a particular time instant. Hence, AC
11
power flow analysis is always carried out for the load and generator values at a particular instant.
2.1.2
Transmission line
In a transmission grid, the transmission lines are generally of medium length or of long length. A
line of medium length is always represented by the nominal- model as shown in Fig. 2.1, where
z is the total series impedance of the line and Bc is the total shunt charging susceptance of the
line. On the other hand, a long transmission line is most accurately represented by its distributed
parameter model. However, for steady-state analysis, a long line can be accurately represented by
the equivalent- model, which predicts accurate behavior of the line with respect to its terminal
measurements taken at its two ends. The equivalent- model is shown in Fig. 2.2.
Figure 2.2: Equivalent model of a long transmission line connected between buses i and j
In Fig. 2.2,
12
z
is the characteristic impedance of the line
y
Hence, for power system analysis, a transmission line (medium or long) is always represented by
a circuit.
2.1.3
Transformer
For power system steady-state and fault studies, generally the exciting current of the transformer is
neglected as it is quite low compared to the normal load current flowing through the transformer.
Therefore, a two winding transformer connected between buses i and j is represented by its per
unit leakage impedance as shown in Fig. 2.3.
2.2
As the name suggests, the injected power (current) indicates the power (current) which is fed in to
a bus. To understand this concept, let us consider Fig. 2.6. In part (a) of this figure, a generator is
connected at bus k supplying both real and reactive power to the bus and thus, the injected real
and reactive power are taken to be equal to the real (reactive) power supplied by the generator. The
13
Figure 2.4: Equivalent circuit of a regulating transformer with transformation ratio 1:t
Figure 2.5: Equivalent circuit of a regulating transformer with transformation ratio a:1
corresponding injected current is also taken to be equal to the current supplied by the generator.
On the other hand, for a load connected to bus k (as shown in Fig. 2.6(b)), physically the real
(reactive) power consumed by the load flows away from the bus and thus, the injected real (reactive)
power is taken to be the negative of the real (reactive) power consumed by the load. Similarly, the
corresponding injected current Ik is also taken as the negative of the load current. If both a generator
14
and a load are connected at a particular bus (as depicted in Fig. 2.6(c)), then the net injected real
(reactive) power supplied to the bus is equal to the generator real (reactive) power minus the real
(reactive) power consumed by the load. Similarly, the net injected current in this case is taken to
be the difference of the generator current and the load current.
2.3
BUS)
Formation of bus admittance matrix (Y
Let us consider a 5 bus network as shown in Fig. 2.7. In this network, all the transmissions are
represented by models. Therefore, the equivalent circuit of the above network is shown in Fig.
2.8.
In Fig. 2.8, Ik ; k = 1, 2, 3, 4, 5 are the injected currents at bus k. Further, the quantity yij
denotes the series admittance of the line i-j whereas the quantity yijs denotes the half line charging
susceptance of the line i-j. Now applying KCL at each bus k one obtains,
I1 = yT 1 (V1 V2 ) = yT 1 V1 yT 1 V2
15
(2.1)
(2.2)
I3 = (V3 V2 )
y23 + V3 y23s + (V3 V5 )t
yT 2 + t(t 1)
yT 2 V3 + (V3 V4 )
y34 + V3 y34s
= V2 y23 + {
y23 + y23s + t
yT 2 + t(t 1)
yT 2 + y34 + y34s } V3 y34 V4 t
yT 2 V5
(2.3)
I4 = (V4 V2 )
y24 + y24s V4 + (V4 V3 )
y34 + y34s V4
= V2 y24 V3 y34 + (
y24 + y24s + y34 + y34s )V4
(2.4)
I5 = (V5 V3 )t
yT 2 + (1 t)
yT 2 V5
= V3 t
yT 2 + {t
yT 2 + (1 t)
yT 2 } V5
(2.5)
I1 Y11
I2 Y21
I3 = Y31
I Y
4 41
I5 Y51
Y12
Y22
Y32
Y42
Y52
Y13
Y23
Y33
Y43
Y53
Y14
Y24
Y34
Y44
Y54
Y15 V1
Y25 V2
Y35 V3
Y45 V4
Y55 V5
(2.6)
Where, Y11 = yT 1 ;
Y12 =
yT 1 ; Y13 = Y14 = Y15 = 0; Y21 =
yT 1 ;
Y22 = (
yT 1 + y23s + y23 + y24s + y24 ); Y23 =
y23 ; Y24 =
y24 ; Y25 = 0;
Y31 = 0; Y32 =
y23 ; Y33 = {
y23 + y23s + t
yT 2 + t(t 1)
yT 2 + y34 + y34s } ;
Y34 =
y34 ; Y35 = t
yT 2 ; Y41 = 0; Y42 =
y24 ; Y43 =
y34 ;
Y44 = (
y24 + y24s + y34 + y34s ); Y45 = 0; Y51 = Y52 = 0;
Y53 = t
yT 2 ; Y54 = 0; Y55 = {t
yT 2 + (1 t)
yT 2 }
Equation (2.6) can be written as,
IBUS = Y
BUS V
BUS
(2.7)
Where,
17
Similarly, for a n bus power system, the relation given in equation (2.7) holds good, where,
IBUS = [I1 , I2 In ]T (n 1) is the vector of bus injection currents
18
2.4
Zc Zm Ic
Vc
][ ]
[ ] = [
Zm Zd Id
Vd
Or,
1
Vc
Zc Zm
Ic
Vc
Zd Zm
1
]
=
]
[
[ ] = [
]
]
[
[
2
Vd
Zm Zd
Id
Zm
Zc Vd
Zc Zd Zm
Or,
Ic
Yc Ym Vc
][ ]
[ ] = [
Ym Yd Vd
Id
(2.8)
Where,
Yc =
Zd
;
2
Zc Zd Zm
Yd =
Zc
;
2
Zc Zd Zm
and Ym =
Zm
2
Zc Zd Zm
Vu
Vc
Vu Vv
1 1 0 0 Vv
Now from Fig. 2.9, [ ] = [
]=[
]
Vd
Vx Vy
0 0 1 1 Vx
Vy
Or,
Vu
V
Vc
1 1 0 0
v
[ ] = [C] where, C = [
]
Vd
0 0 1 1
Vx
Vy
19
(2.9)
(2.10)
Iu Ic
I I
v c
= =
Ix Id
Iy Id
1 0
c
T I
1 0 Ic
[ ] = [C] [ ]
Id
0 1 Id
0 1
(2.11)
Ic
Yc Ym Vc
Yc
[ ] = [
]
[
]
=
[
Id
Ym Yd Vd
Ym
Or,
Vu
V
Ym
v
[C]
]
Vx
Yd
Vy
Vu
Iu
V
I
c
T I
Yc Ym
T
v
v
[C]
[C] [ ] = = [[C] ] [
]
Vx
Ix
Ym Yd
Id
Vy
Iy
(2.12)
Now,
c
T Y
[C] [
Ym
Ym
[C]
=
]
Yd
1 0
1 0 Yc Ym 1 1 0 0
[
][
]
0 1 Ym Yd
0 0 1 1
0 1
(2.13)
Or,
c
T Y
[C] [
Ym
Ym
[C]
]
=
Yd
1 0
1 0
[
0 1
0 1
Yc Yc
Ym Ym
Ym Ym
]
Yd Yd
(2.14)
Or,
c
T Y
[C] [
Ym
Yc Yc Ym Ym
Y
Ym
Yc Ym Ym
] [C] =
Ym Ym
Yd
Yd Yd
Ym Ym Yd
Yd
20
(2.15)
Iu Yc Yc Ym Ym Vu
I Y
Yc Ym Ym Vv
v
c
=
Ix Ym Ym
Yd Yd Vx
Iy Ym Ym Yd
Yd Vy
(2.16)
Iu = Yc Vu Yc Vv + Ym Vx Ym Vy
= Yc Vu Yc Vv + Ym Vx Ym Vy + Ym Vu Ym Vu
= Yc (Vu Vv ) + (Ym )(Vu Vx ) + Ym (Vu Vy )
(2.17)
Or,
(2.18)
Similarly,
Iv = Yc Vu + Yc Vv Ym Vx + Ym Vy
= Yc Vu + Yc Vv Ym Vx + Ym Vy + Ym Vv Ym Vv
= Yc (Vu Vv ) + (Ym )(Vv Vy ) + Ym (Vv Vx )
(2.19)
Or,
(2.20)
Ix = Ym Vu Ym Vv + Yd Vx Yd Vy
= Ym Vu Ym Vv + Yd Vx Yd Vy + Ym Vx Ym Vx
= Yd (Vx Vy ) + (Ym )(Vx Vu ) + Ym (Vx Vu )
(2.21)
Or,
(2.22)
Equations (2.18), (2.20) and (2.22) can be represented by the partial networks shown in Figs.
2.10, 2.11 and 2.12 respectively. Combining Figs. 2.10, 2.11 and 2.12, Fig. 2.13 is obtained.
Again from the last row of equation (2.16),
Iy = Ym Vu + Ym Vv Yd Vx + Yd Vy
= Ym Vu + Ym Vv Yd Vx + Yd Vy + Ym Vy Ym Vy
= Yd (Vy Vx ) + (Ym )(Yy Vv ) + Ym (Yy Vu )
(2.23)
It can be observed that equation (2.23) is also represented by Fig. 2.13. Therefore, the voltage21
22
23
24
25
2.5
I1 Y11
I Y
2 21
=
In Yn1
Or,
Y12 Y1n V1
Y22 Y2n V2
Yn2 Ynn Vn
(2.24)
Ii = Yij Vj
(2.25)
j=1
Si = Pi + jQi = Vi Ii
Now, Vi = Vi eji ;
Vj = Vj ejj ;
(2.26)
j=1
Or,
n
Pi = Vi Vj Yij cos(i j ij )
(2.27)
j=1
n
Qi = Vi Vj Yij sin(i j ij )
(2.28)
j=1
Equations (2.27) and (2.28) are known as the basic load flow equations. It can be seen that for
any ith bus, there are two equations. Therefore, for a n-bus power system, there are altogether 2n
load-flow equations.
Now, from equations (2.27) and (2.28) it can be seen that there are four variables (Vi , i , Pi and
Qi ) associated with the ith bus. Thus for the n-bus system, there are a total of 4n variables. As
there are only 2n equations available, out of these 4n variables, 2n quantities need to be specified
and remaining 2n quantities are solved from the 2n load-flow equations. As 2n variables are to
be specified in a n bus system, for each bus, two quantities need to be specified. For this purpose,
the buses in a system are classified into three categories and in each category, two different quantities
are specified as described below.
1. PQ Bus: At these buses loads are connected and therefore, these buses are also termed as load
buses. Generally the values of loads (real and reactive) connected at these buses are known
and hence, at these buses Pi and Qi are specified (or known). Consequently, Vi and i need to
be calculated for these buses.
26
2. PV Bus: Physically, these buses are the generator buses. Generally, the real power supplied
by the generator is known (as we say that the generation is supplying 100 MW) and also, the
magnitude of the terminal voltage of the generator is maintained constant at a pre-specified
value by the exciter (provided that the reactive power supplied or absorbed by the generator
is within the limits). Thus, at a PV bus, Pi and Vi are specified and consequently, Qi and i
need to be calculatd.
3. Slack Bus: To calculate the angles i (as discussed above), a reference angle (i = 0) needs to
be specified so that all the other bus voltage angles are calculated with respect to this reference
angle. Moreover, physically, total power supplied by all the generation must be equal to the
sum of total load in the system and system power loss. However, as the system loss cannot be
computed before the load flow problem is solved, the real power output of all the generators in
the system cannot be pre-specified. There should be at least one generator in the system which
would supply the loss (plus its share of the loads) and thus for this generator, the real power
output cant be pre-specified. However, because of the exciter action, Vi for this generator can
still be specified. Hence for this generator, Vi and i (= 0) are specified and the quantities Pi
and Qi are calculated. This generator bus is designated as the slack bus. Usually, the largest
generator in the system is designated as the slack bus.
To summarise, the details of different types of buses in a n bus, m generator power system are
shown in Table 2.1. Now, please note that in a load flow problem, the quantities Pi and Qi (Qi at
Table 2.1: Classification of buses
Type
Specified quantity
Solution quantity
PQ
n-m
P i , Qi
Vi , i
PV
m-1
Pi , Vi
Qi , i
Slack
Vi , i
P i , Qi
PV while Pi and Qi at slack buses) are not directly solved. Only the quantities Vi and i are directly
solved (Vi for all PQ buses while i for all PV and PQ buses). This is because of the fact that once
Vi and i at all PV and PQ buses are solved, then the voltage magnitudes and angles at all the buses
are known (Vi , i at the slack bus are already specified) and subsequently, using equations (2.27)
and (2.28), Pi and Qi at any bus can be calculated.
Therefore, in a n bus, m generator system, the unknown quantities are: Vi (total n-m of
them) and i (total n-1 of them). Therefore, total number of unknown quantities is 2n-m-1. On
the other hand, the specified quantities are: Pi (total n-1 of them) and Qi (total n-m of them).
Hence total number of specified quantities is also 2n-m-1. As the number of unknown quantities is
equal to the number of specified quantities, the load-flow problem is well-posed.
27
Equations (2.27) and (2.28) represent a set of simultaneous, non-linear, algebraic equations. As
the set of equations is non-linear, no closed form, analytical solution for these equations exist. Hence,
these equations can only be solved by using suitable numerical iterative techniques. For solving the
load flow problem, various iterative methods exist. These are:
1. Gauss-seidel method
2. Newton Raphron (polar) technique
3. Newton Raphron (rectangular) technique
4. Fast-decoupled load flow
We will discuss these methods one by one and we start with the Gauss-Seidel method.
2.6
Before discussing the Gauss-Seidel load flow (GSLF) technique, let we first review the basic GaussSeidel procedure for solving a set of non-linear algebraic equations.
Let the following n equations are given for the n unknown quantities x1 , x2 , xn ;
f1 (x1 , x2 xn ) = 0
f2 (x1 , x2 xn ) = 0
fn (x1 , x2 xn ) = 0
(2.29)
It is to be noted that in equation (2.29), the function f1 , f2 , fn are all non-linear in nature
and no particular form of these equations is assumed. Now, with some algebraic manipulation, from
the first equation of equation set (2.29), the variable x1 can be represented in terms of the other
variables. Similarly from the second equation, the variable x2 can be represented in terms of the
other variables. Proceeding in the same way, from the nth equation, the variable xn can be expressed
in terms of the other variables. Therefore, let,
x1
x2
xk
xn
= g1 (x2 , x3 xn )
= g2 (x1 , x3 xn )
= gn (x1 , x2 xn1 )
(2.30)
x(k)
= g1 (x(k1)
, x(k1)
, x(k1)
);
n
1
2
3
(k1)
x(k)
= g2 (x(k)
, x(k1)
);
n
2
1 , x3
(k)
(k)
(k1)
(k1)
xp(k) = gp (x(k)
);
1 , x2 , xp1 , xp+1 , xn
(k)
(k)
x(k)
= gn (x(k)
n
1 , x2 , xn1 );
(k)
(k)
(k1)
for all i = 1, 2, n;
Step 3: Compute ei = xi xi
(k) (k)
(k)
Step 4: Compute er = max(e1 , e2 , en ) ;
Step 5: If er (tolerance limit), stop and print the solution. Else set k = k + 1 and go to step
2.
It is to be noted that in step 2, for updating the variable xp , the most updated values of
x1 , x2 , xp1 (which are before xp in the sequence of the solution variables) are used while for the
variables xp+1 , xp+2 , xn (which are after xp in the sequence of the solution variables), the values
pertaining to previous iteration are used (as these variables have not been updated yet). Subsequently, in steps 3 and 4, the maximum absolute error between the solutions of the current iteration
and previous iteration is calculated. If this maximum absolute error is less then a pre-specified
tolerance value, then the algorithm is considered to be converged. Otherwise, the solution variables
are again updated.
With this background of basic Gauss-Seidel method, we are now in a position of discussing GSLF,
which we will do next.
2.7
n
1
Yii
k=1
k=1
k=1
i
i
29
Pi jQi
. Thus,
Pi + jQi = Vi Ii we get, Ii =
Vi
P jQ
n
1
i
i
ik Vk
Vi =
Yii Vi
k=1
(2.31)
Equation (2.31) is the basic equation for performing GSLF. It is to be noted that without loss
of generality, it is assumed that the m generators are connected to the first m buses (bus 1
being the slack bus) and remaining (n-m) buses are load buses. Now, initially to understand the
basic GSLF procedure, let us assume that m = 1, i.e., there is only one generator (which is also the
slack bus) and the rest (n-1) buses are all load buses. To perform load-flow computation, initial
guesses of the bus voltages are necessary. As any power system is generally expected to operate at
the normal steady-state operating condition (with the bus voltage magnitudes maintained between
(0)
0.95 - 1.05 p.u.), all the unknown bus voltage are initialized to 1.00o p.u (i.e. Vj = 1.00o for
j = 2, 3, n). This process of initializing all bus voltage to 1.00o is called flat start. With these
initial bus voltages, the complete procedure for GSLF (having no PV bus) is as follows.
P jQ
n
1
2
2
(k1)
(k1) Y2j Vj
V
=
Y22 {V2
} j =1
n
1 Pp jQp p1 (k)
(k1)
(k)
Vp
=
(k1) Ypj Vj Ypj Vj
Ypp {Vp
j=p+1
} j=1
Pn jQn n1
1
(k)
(k)
nj Vj
Vn
=
Ynn {Vn(k1) }
j=1
(k)
2
(k)
(k)
(k1)
for all i = 2, n;
Step 3: Compute ei = Vi Vi
(k) (k)
(k)
(k)
Step 4: Compute e = max(e2 , e3 , en ) ;
Step 5: If e(k) (tolerance limit), stop and print the solution. Else set k = k+1 and go to step 2.
With the above understanding of the basic GSLF, we are now in a position to discuss the GSLF
procedure for a system having multiple generators. Before we discuss the GSLF procedure, let
us look into the procedure of initialisation of bus voltages (which is little different than assuming
a flat start for all the bus volatges). For a system having multiple generators, the bus voltage
30
initialisation is carried out in a two step procedure; i) the load buses are initialised with flat start
(0)
(i.e. Vj = 1.00o for j = (m + 1), (m + 2), n) and ii) the magnitudes of the voltages of the
PV buses are initialised with the corresponding specified voltage magnitudes while initialising all
(0)
these voltage angles to 0o (i.e. Vj = Vjsp 0o for j = 2, 3, m, where Vjsp is the specified bus
voltage magnitude of the j th generator). Now, as discussed earlier, the reactive power supplied or
absorbed by a generator (QG ) is calculated by the load flow procedure. However any generator has
a maximum and minimum limit on QG . If the QG from the generator is within these limits, then the
generator excitation system is able to maintain the terminal voltage at the specified value. On the
other hand, if the generator reaches its limit on QG (either maximum or minimum), then because
of the insufficient amount of reactive power (either supplied or absorbed), the generator excitation
system would not be able to maintain the terminal voltage magnitude at the specified value. In that
case the generator bus would behave as a PQ bus (P being already specified for the generator and
Q is set at either maximum or minimum limiting value of QG ). In power system terminology, this
phenomenon (where the generator is behaving like a PQ bus) is termed as PV to PQ switching
which should also be accounted for in any load-flow solution methodology.
This is incorporated in GSLF by the following procedure. At the beginning of each iteration, QG
injection by each generator is calculated. If this calculated QG is found to be within the corresponding limits then this generator continues to behave as a PV bus. Hence Vi of this bus (at which the
generator is connected ) is still maintained at the corresponding specified value and only the angle
of this bus voltage is calculated in the present iteration. On the other hand if QG is found to exceed
any limit (either maximum or minimum), then it is fixed at that limit and the bus is considered
to act like a PQ bus. Thus, both the magnitude and angle of the bus voltage are calculated in
the present iteration. With this background, the complete algorithm of GSLF involving multiple
generator buses is as follows.
(0)
Q(k)
= Vi(k1) Vj(k1) Yij sin (i(k1) j(k1) ij )
i
j=1
(k)
b) If, Qmin
Qi
i
is given by,
(k)
(k)
(k)
(k)
Qmax
; then assign Vi = Visp and i = (Ai ). The quantity Ai
i
(k)
i
n
1 Pi jQi(k) i1 (k)
(k1)
=
Yij Vj
Yij Vj
Yii {Vi(k1) }
j=1
j=i+1
31
(k)
c) If Qi
Qmax
, then calculate
i
(k)
i
V
(k)
d) If Qi
i1
n
1 Pi jQmax
(k)
(k1)
i
ij Vj Yij Vj
=
Yii {Vi(k1) }
j=1
j=i+1
Qmin
, then calculate
i
(k)
i
i1
n
1 Pi jQmin
(k)
(k1)
i
ij Vj Yij Vj
=
Yii {Vi(k1) }
j=1
j=i+1
(k)
i1
n
1 Pi jQ(k)
(k)
(k1)
i
ij Vj Yij Vj
=
Yii {Vi(k1) }
j=1
j=i+1
(k)
(k1)
for all i = 2, n;
Step 4: Compute ei = Vi Vi
(k) (k)
(k)
(k)
Step 5: Compute e = max(e2 , e3 , en ) ;
Step 6: If e(k) , stop and print the solution. Else set k = k + 1 and go to step 2.
We will illustrate the GSLF algorithm with an example in the next lecture.
32
2.7.1
To illustrate the basic procedure of GSLF, let as consider a small 5-bus system as shown in Fig.
2.16. In this system, buses 1-3 are generator buses and buses 4-5 are load buses. Therefore, in this
system, n = 5 and m = 3. Moreover, bus 1 is taken to be the slack bus and thus, buses 2-3 are
considered to be PV buses. The bus data and line data of this system are given in Tables A.1 and
A.2 respectively. From Table A.1, the injected real and reactive powers at different buses can be
obtained as follows: P2 = 0.5 p.u, P3 = 1.0 p.u, P4 = 1.15 p.u, P5 = 0.85 p.u, Q4 = 0.6 p.u, and
BUS of this system (computed from the line data given in Table
Q5 = 0.4 p.u. Moreover, the Y
A.2) is shown in equation (2.32). Note that in this equation, the real part (G) and the imaginary
BUS matrix (Y
BUS = G + jB) are shown separately.
part (B) of the Y
3.2417 1.4006
0
0
1.8412
G= 0
1.8412 4.2294 1.2584 1.1298
0
0
1.2584 2.1921 0.9337
1.8412
0
1.1298 0.9337 3.9047
13.0138 5.6022
0
0
7.4835
0
0
B=
0
7.4835 18.9271 7.1309
4.4768
0
0
7.1309 10.7227 3.7348
7.4835
0
4.4768
3.7348
15.5521
33
(2.32)
For applying GSLF, initially the flat start profile is assumed. Please note that the flat voltage
profile is followed for PQ buses. For PV buses, the initial voltage magnitude is taken to be equal
to their corresponding specified voltage magnitude. However, the initial voltage angles are always
assumed to be zero. Therefore, from the data given in Table A.1, all the 5 bus voltages are initialised
to 1.00o p.u. Now as the system contains both PV and PQ buses, we follow the complete
GSLF algorithm. In step 2(a) of this algorithm, we first calculate the reactive power absorbed
or generated by generators 2 and 3 (corresponding to i = 2 and i = 3). The calculated values of Q2
and Q3 at iteration 1 (in p.u.) are shown in Table 2.2 (denoted as Qcal in this table). Now, the data
in Table A.1 show that the minimum and maximum reactive power limits for both these generators
are -5 p.u. and 5 p.u respectively. Hence, the calculated values of Q2 and Q3 are well within the
corresponding reactive power limits. Therefore, both bus 2 and bus 3 are continued to operate as PV
buses and as a result, their voltage magnitudes are maintained at the corresponding specified values
and only the voltage angles are calculated in step 2(b) (utilising the calculated values of Q2 and Q3 ).
Subsequently in step 3, both the magnitude and angles of buses 4 and 5 are calculated. The results
of iteration 1 are shown in Table 2.2. Finally in steps 4-5, the error is calculated, which is also shown
in Table 2.2. The error is found to be more than the threshold value (taken to be equal to 1.0e12 )
and therefore the algorithm goes back to step 2 again. The iteration wise result for first 6 iterations
are shown in Tables 2.2 and 2.3. Please observed from these two tables that because of high Qmax
G
min
and QG limits, the reactive powers supplied or observed by these two generators are always within
these limits and thus bus 2 and 3 continue to act as PV buses from iteration to iteration. Also note
from Tables 2.2 and 2.3 that the error reduces with iteration. Finally, the algorithm converges after
69 iterations and the final solution is shown in Table 2.4. Please note from this table that the final
values of Q2 and Q3 are -18.51 and 68.87 MVAR respectively. These values are well within their
corresponding reactive power limits and thus, the voltage magnitudes of bus 2 and 3 maintained at
1.0 p.u (pre-specified values). The finally computed values of real and reactive power injection at
all the buses are also shown in Table 2.4, which are also found to be exactly equal to the specified
injected values.
Table 2.2: GSLF results in 5 bus system without any generator Q limit for iterations 1-3
Iteration = 1
Bus no.
1
2
3
4
5
Qcal
Iteration = 2
Qcal
Iteration = 3
Qcal
(p.u)
(p.u)
(deg)
(p.u)
(p.u)
(deg)
(p.u)
(p.u)
(deg)
1.0
0
1.0
0
1.0
0
-0.0720
1.0
2.0533 -0.0663
1.0
4.1038 -0.2022
1.0
3.7091
-0.0932
1.0
3.5968 0.2937
1.0
2.6494 0.5142
1.0
1.6234
0.9394 -3.2379
0.9167 -5.0548
0.9101 -6.0776
0.957 -2.5257
0.9482 -3.2562
0.946 -3.797
error = 0.081708
error = 0.037148
error = 0.017906
Let us now study the behaviour of GSLF when generator reactive power limit is violated. Towards
this goal, let us assume that the maximum reactive power which can be supplied by generator 3 in
34
Table 2.3: GSLF results in 5 bus system without any generator Q limit for iterations 4-6
Iteration = 4
Bus no.
1
2
3
4
5
Qcal
Iteration = 5
Qcal
Iteration = 6
Qcal
(p.u)
(p.u)
(deg)
(p.u)
(p.u)
(deg)
(p.u)
(p.u)
(deg)
1.0
0
1.0
0
1.0
0
-0.2129
1.0
3.1318 -0.2071
1.0
2.6927 -0.2013
1.0
2.3782
0.5926
1.0
0.8641 0.6263
1.0
0.3228 0.6461
1.0
-0.0546
0.9082 -6.7844
0.9074 -7.266
0.9069 -7.5985
0.9452 -4.1735
0.9448 -4.435
0.9446 -4.6163
error = 0.013252
error = 0.0094475
error = 0.0065856
Bus no.
1
2
3
4
5
Pinj
Qinj
(p.u)
(deg)
(p.u)
(p.u)
1.0
0
0.56743 0.26505
1.0
1.65757
0.5
-0.18519
1.0
-0.91206
1.0
0.68875
0.90594 -8.35088
-1.15
-0.6
0.94397 -5.02735
-0.85
-0.4
Total iteration = 69
50 MVAR (it supplies 68.87 MVAR when no limit is imposed on the generators). The iteration
wise solutions for first 6 iterations of the load flow computation with this maximum limit are shown
in Tables 2.5 and 2.6. Now, let us compare Tables 2.2 and 2.5. From these two tables it can be
observed that the load flow solution with generator Q limit proceeds in identical fashion for first two
iterations as in the case with no reactive power limit on the generators. However, from iteration 3
onwards the solution changes. In iteration 3, Q3 calculated is found to be equal to 51.42 MVAR. As
a result, Q3 is limited to 50 MVAR and Bus 3 is converted to a PQ bus, and therefore its voltage
magnitude is calculated using the expression shown in step 2(c). Please observe that this voltage
magnitude is not maintained at 1.0 p.u (in fact, it becomes less than 1.0 p.u. because of insufficient
reactive power). In the subsequent iteration also, calculated Q3 is always found to be more than the
maximum limit and as a result, Q3 is always maintained at 50 MVAR thereby making V3 < 1.0
p.u. The algorithm finally converges with a tolerance of 1.0e12 p.u. after 66 iterations and the final
solution are shown in Table 2.7. In this table, the GSLF results without any reactive power limit (as
shown in Table 2.4) are also reproduced for comparison. Please note from Table 2.7 that because of
cap on Q3 , the overall voltage profile of the system is little lower than that obtained with no limit
on generator reactive power.
As a second example, let us now consider the IEEE-14 bus system. The data of the IEEE 14
bus system are shown in Tables A.3 and A.4. The power flow solution without any limit on the
35
Table 2.5: GSLF results in 5 bus system with generator Q limit on bus 3 for iterations 1-3
Iteration = 1
Bus no.
Qcal
Iteration = 2
Qcal
Iteration = 3
Qcal
(p.u)
(p.u)
(deg)
(p.u)
(p.u)
(deg)
(p.u)
(p.u)
(deg)
1.0
0
1.0
0
1.0
0
-0.0720
1.0
2.0533 -0.0663
1.0
4.1038 -0.2022
1.0
3.7091
-0.0932
1.0
3.5968 0.2937
1.0
2.6494 0.5142 0.9955 1.6318
0.9394 -3.2379
0.9167 -5.0548
0.9071 -6.0918
0.957 -2.5257
0.9482 -3.2562
0.944 -3.8036
error = 0.081708
error = 0.037148
error = 0.019063
1
2
3
4
5
Table 2.6: GSLF results in 5 bus system with generator Q limit on bus 3 for iterations 4-6
Iteration = 1
Bus no.
Qcal
Iteration = 2
Qcal
Iteration = 3
Qcal
(p.u)
(p.u)
(deg)
(p.u)
(p.u)
(deg)
(p.u)
(p.u)
(deg)
1.0
0
1.0
0
1.0
0
-0.1784
1.0
3.1067 -0.1365
1.0
2.6692 -0.1082
1.0
2.3656
0.5337 0.991 0.9288 0.5178 0.9882 0.4399 0.5109 0.9864 0.1025
0.9012 -6.8046
0.8973 -7.2918
0.8947 -7.6271
0.9409 -4.1739
0.9388 -4.4292
0.9374 -4.6044
error = 0.012967
error = 0.0088975
error = 0.0061049
1
2
3
4
5
Table 2.7: Final Results of the 5 bus system with GSLF with generator Q limit
Bus no.
1
2
3
4
5
Pinj
Qinj
Pinj
Qinj
(p.u)
(deg)
(p.u)
(p.u)
(p.u)
(deg)
(p.u)
(p.u)
1.0
0
0.56743 0.26505
1.0
0
0.56979 0.33935
1.0
1.65757
0.5
-0.18519
1.0
1.69679
0.5
-0.04769
1.0
-0.91206
1.0
0.68875 0.9825 -0.63991
1.0
0.5
0.90594 -8.35088
-1.15
-0.6
0.88918 -8.35906
-1.15
-0.6
0.94397 -5.02735
-0.85
-0.4
0.93445 -4.98675
-0.85
-0.4
Total iteration = 69
Total iteration = 66
generator reactive power is shown in Table 2.8. Note that all the terminal voltage of the generators
are maintained at their corresponding specified values. Also observe that the generator connected
at bus 6 supplies a reactive power of 37.27 MVAR. Now assume that generator 6 is constrained to
supply only 30 MVAR. With this Q limit, the power flow solution is also shown in Table 2.8. From
these result followings salient point can be noted:
a. Reactive power supplied by generator 6 is limited at 30 MVAR.
b. As a result, V6 goes down to 1.05497 p.u. (from the specified value of 1.07 p.u.).
36
Bus no.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
(p.u)
1.06
1.045
1.04932
1.03299
1.04015
1.07
1.02076
1.0224
1.0201
1.0211
1.04144
1.0526
1.04494
1.01249
Pinj
(deg)
(p.u)
0
2.37259
-5.17113
0.183
-14.54246
-1.19
-10.39269 -0.4779
-8.76418 -0.07599
-12.52265
0.112
-13.44781
0
-13.47154
0
-13.60908 -0.29499
-13.69541
-0.09
-13.22158 -0.03501
-13.42868 -0.06099
-13.50388 -0.135
-14.60128 -0.14901
Total iteration = 676
Qinj
(p.u)
-0.3308
-0.166
-0.08762
-0.039
-0.01599
0.37278
0
-0.129
-0.16599
-0.05799
-0.018
-0.01599
-0.05799
-0.05001
(p.u)
1.06
1.045
1.04697
1.02902
1.03615
1.05497
1.01266
1.01391
1.0118
1.01154
1.02915
1.03787
1.03063
1.00136
Pinj
(deg)
(p.u)
0
2.37188
-5.17845
0.183
-14.55556
-1.19
-10.35987 -0.4779
-8.71027 -0.07599
-12.45871
0.112
-13.49478
0
-13.5185
0
-13.66101 -0.29499
-13.73679
-0.09
-13.21814 -0.03501
-13.39145 -0.06099
-13.48166 -0.135
-14.64504 -0.14901
Total iteration = 718
Qinj
(p.u)
-0.31249
-0.1066
-0.08762
-0.039
-0.01599
0.3
0
-0.129
-0.16599
-0.05799
-0.018
-0.01599
-0.05799
-0.05001
c. Because of the limit on generator reactive power, the overall voltage profile is on the lower side
as compared to that obtained without any Q limit.
As the last example the 30 bus system is considered. The data of this system are given in Tables
A.5 and A.6. Again initially the load flow solution has been computed without any limit on the
generator reactive power and the result are shown in Table 2.9. Subsequently Q limits have been
imposed on both the generator connected at bus 11 (20 MVAR) and the generator connected at bus
13 (30 MVAR). However, with a tolerance of 10e12 p.u. GSLF algorithm fails to converge even
after 10,000 iteration. When the tolerance is reduced to 10e6 p.u., the algorithm converges in 348
iteration and the result are again shown in Table 2.9. As can be seen from these results, for both
the generation, the reactive power supplied have been fixed at their corresponding limits and as a
result, the overall voltage profile of the system has gone down.
From these results it is observed that the convergence characteristics of the GSLF technique is
quite poor. Usually the number of iteration taken by GSLF is quite large and moreover in many
cases, GSLF even fails to converge. To overcome these difficulties of GSLF, Newton- Raphson(NR)
techniques have been developed, which are our next topics of discussion. These are two versions of
NR techniques, namely, i) NR in polar co-ordinate and ii) NR in rectangular co-ordinate. We will
study both these versions one by one and will start with the NR in polar co-ordinate from the next
lecture.
37
Bus no.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
(p.u)
1.05
1.0338
1.03128
1.02578
1.0058
1.02178
1.00111
1.023
1.04608
1.03606
1.0913
1.04859
1.0883
1.03346
1.02825
1.0359
1.0306
1.01873
1.01626
1.02041
1.02305
1.02343
1.0165
1.00939
1.00048
0.9825
1.00379
1.02049
0.98353
0.97181
Pinj
(deg)
(p.u)
0
2.38673
-4.97945 0.3586
-7.96653
-0.024
-9.58235
-0.076
-13.60103 -0.6964
-11.50296
0
-13.9994
-0.628
-12.56853
-0.45
-13.04088
0
-14.88589 -0.058
-11.16876 0.1793
-13.74947 -0.112
-12.56078 0.1691
-14.71704 -0.062
-14.86737 -0.082
-14.50539 -0.035
-14.98291
-0.09
-15.58107 -0.032
-15.81066 -0.095
-15.63819 -0.022
-15.35955 -0.175
-15.35222
0
-15.41998 -0.032
-15.81043 -0.087
-15.84004
0
-16.27422 -0.035
-15.59587
0
-12.1474
0
-16.87497 -0.024
-17.79427 -0.106
Total iteration = 851
Qinj
(p.u)
-0.29842
-0.05698
-0.012
-0.016
0.05042
0
-0.109
0.12343
0
-0.02
0.24018
-0.075
0.31043
-0.016
-0.025
-0.018
-0.058
-0.009
-0.034
-0.007
-0.112
0
-0.016
-0.067
0
-0.023
0
0
-0.009
-0.019
38
Pinj
Qinj
(p.u)
(deg)
(p.u)
(p.u)
1.05
0
2.3865 -0.29386
1.0338 -4.98084 0.35861 -0.04562
1.03045 -7.95523 -0.02399 -0.012
1.02477 -9.56929 -0.07598 -0.016
1.0058 -13.60836 -0.6964 0.05532
1.02084 -11.49304 0.00003
0
1.00055 -13.99792 -0.628
-0.109
1.023 -12.57567
-0.45
0.15111
1.04006 -13.02865 0.00001
0
1.03117 -14.8866 -0.05797 -0.02001
1.07807 -11.12256 0.1793
0.2
1.04456 -13.75755 -0.11198 -0.075
1.08311 -12.55854 0.1691
0.3
1.02931 -14.73168 -0.062
-0.016
1.02403 -14.88097 -0.08199 -0.025
1.03148 -14.51198 -0.035
-0.018
1.02583 -14.98712
-0.09
-0.058
1.01422 -15.59618 -0.03199 -0.009
1.01159 -15.8251 -0.09499 -0.034
1.01568 -15.64961 -0.022
-0.007
1.01825 -15.36524 -0.17496 -0.11201
1.01867 -15.3581
0
0
1.01226 -15.43637 -0.032
-0.016
1.00517 -15.8277
-0.087
-0.067
0.99748 -15.86849
0
0
0.97944 -16.30534 -0.035
-0.023
1.00158 -15.62827
0
0
1.01959 -12.14235
0
0
0.98126 -16.91314 -0.024
-0.009
0.96951 -17.83675 -0.106
-0.019
Total iteration = 348 (totlerance 10e6 )
2.8
Before discussing the application of NR technique in load flow solution, let us first review the basic
procedure of solving a set of non-linear algebraic equation by means of NR algorithm. Let there be
n equations in n unknown variables x1 , x2 , xn as given below,
f1 (x1 , x2 , xn )
f2 (x1 , x2 , xn )
fn (x1 , x2 , xn )
= b1
= b2
= bn
(2.33)
f1
f1
f1
x1 +
x2 + +
xn = b1
x1
x2
xn
f2
f2
f2
(0)
(0)
(0)
f2 (x1 , x2 , xn ) +
x1 +
x2 + +
xn = b2
x1
x2
xn
fn
fn
fn
(0)
(0)
(0)
fn (x1 , x2 , xn ) +
x1 +
x2 + +
xn = bn
x1
x2
xn
(0)
(0)
f1 (x(0)
1 , x2 , xn ) +
(2.34)
f1
f1 (x(0) ) x1
(0)
f2 (x ) 2
+ x1
fn (x(0) ) f
x1
f1
x2
f2
x2
fn
x2
f1
xn x1 b1
f2
x2 = b2
xn
bn
x
n
fn
xn
(2.35)
In equation (2.35), the matrix containing the partial derivative terms is known as the Jacobin
39
matrix (J). As can be seen, it is a square matrix. Hence, from equation (2.35),
m1
b1 f1 (x(0) )
x1
x
(0) )
1 m2
1 b2 f2 (x
= [J]
= [J]
mn
bn fn (x(0) )
xn
(2.36)
Equation (2.36) is the basic equation for solving the n algebraic equations given in equation
(2.33). The steps of solution are as follow:
Step 1: Assume a vector of initial guess x(0) and set iteration counter k = 0.
Step 2: Compute f1 (x(k) ), f2 (x(k) ), fn (x(k) ).
Step 3: Compute m1 , m2 , mn .
Step 4: Compute error =max [m1 , m2 , mn ]
Step 5: If error (pre - specified tolerance), then the final solution vector is x(k) and print
the results. Otherwise go to step 6.
Step 6: Form the Jacobin matrix analytically and evaluate it at x = x(k) .
T
Step 7: Calculate the correction vector x = [x1 , x2 , xn ] by using equation (2.36).
Step 8: Update the solution vector x(k+1) = x(k) +x and update k = k+1 and go back to step 2.
With this basic understanding of NR technique, we will now discuss the application of NR
technique for load flow solution. We will first discuss the Newton Raphson load- flow (NRLF) in
polar co-ordinates.
2.9
For NRLF techniques, the starting equations are same as those in equations (2.27) and (2.28), which
are reproduced below:
n
Pi = Vi Vj Yij cos(i j ij )
(2.37)
j=1
n
Qi = Vi Vj Yij sin(i j ij )
(2.38)
j=1
Now, as before let us again assume that in a n bus, m machine system, the first m buses
are the generator buses with bus 1 being the slack bus. Therefore, the unknown quantities are;
2 , 3 , n (total n-1 quantities) and Vm+1 , Vm+2 , Vn (total n-m quantities). Thus the total
number of unknown quantities is n 1 + n m = 2n m 1. Against these unknown quantities, the
sp
sp
specified quantities are; P2sp , P3sp , Pnsp (total n-1 quantities) and Qsp
m+1 , Qm+2 , Qn (total
n-m quantities). Hence, the total number of specified quantities is also (2n m 1). Let the
T
T
vectors of unknown quantities be denoted as = [2 , 3 , n ] and V = [Vm+1 , Vm+2 , Vn ] .
T
Similarly let the vector of the specified quantities be denoted as Psp = [P2sp , P3sp , Pnsp ] and
sp
sp
Qsp = [Qsp
m+1 , Qm+2 , Qn ]. Also note from equations (2.37) and (2.38) that the real and reactive
power injections at any bus are functions of and V. Thus, these injection quantities can be
40
P2
Pn
Qm+1
Qn
P2
P2
n
Vm+1
Pn
Pn
n
Vm+1
Qm+1 Qm+1
n
Vm+1
Qn
Qn
n
Vm+1
P2
(0)
sp
2 P2 P2 ( , V(0) )
Vn
Pn
(0)
sp
(0)
(
)
P
P
,
V
n
Vn n n
= sp
(0)
(0)
Qm+1 Vm+1 Q
, V )
m+1 Qm+1 (
Vn
Qn
Qsp Q ( (0) , V(0) )
V
m+1
n
n
Vn
(2.39)
(0)
In equation (2.39), the quantity Pi ( , V(0) ) is nothing but the calculated value of Pi with
(0)
(0)
vectors , V(0) . As a result, commonly, the quantity Pi ( , V(0) ) is denoted as Pical . With
these notations, equation (2.39) can be written as,
Psp Pcal
J1 J2
P
][
] = [ sp
[
]=[
]
cal
J3 J4 V
Q Q
Q
(2.40)
In equation (2.40) the vectors Pcal and Qcal are defined as; Pcal = [P2cal , P3cal , Pncal ] and
cal
cal
sp are of dimension (n 1) 1
Qcal = [Qcal
m+1 , Qm+2 , Qn ]. Also note that the vectors and P
each and the vectors V and Qsp are of dimension (n m) 1 each. Therefore, from equations (2.39)
and (2.40),
T
P2 P2
P2
2 3
n
P3 P3
P
3
= 2 3
J1 =
n
Pn Pn
Pn
2 3
n
(2.41)
P2
P2
P2
Vm+1 Vm+2
Vn
P3
P
P
3
3
= Vm+1 Vm+2
J2 =
Vn
V
Pn
Pn
Pn
Vm+1 Vm+2
Vn
(2.42)
41
Qm+1 Qm+1
Qm+1
2
3
n
Qm+2 Qm+2
Q
m+2
J3 =
= 2
3
n
Qn
Qn
Qn
2
3
n
(2.43)
Qm+1 Qm+1
Qm+1
Vm+1 Vm+2
Vn
Qm+2 Qm+2
Q
m+2
J4 =
= Vm+1 Vm+2
Vn
V
Qn
Qn
Qn
Vm+1 Vm+2
Vn
(2.44)
In equations (2.41) to (2.44) the sizes of the various matrices are as follows: J1 (n1)(n1),
J2 (n 1) (n m), J3 (n m) (n 1) and J4 (n m) (n m). Now, equation (2.40)
can be written in compact form as,
[J] [X] = [M ]
(2.45)
J1 J2
] is known as the Jacobian matrix, the vector X =
J3 J4
Psp Pcal
[
] is known as the correction vector and the vector M = [ sp
] is known as the
V
Q Qcal
mismatch vector. Further, the size of the matrix J is (2n m 1) (2n m 1) while the sizes of
both the vectors X and M is (2n m 1) 1.
In equation (2.45), the matrix J = [
Equation (2.45) forms the basis of the NRLF (polar) algorithm, which is described below. Please
note that in the algorithm described below it is assumed that there is no generator which violates
its reactive power generation or absorption limit. The case of violation of reactive power generation
or absorption limit would be dealt with a little later.
Basic NRLF (polar) algorithm
(0)
(0)
(k1)
j=1
j =1
i
j=1
j =1
i
(2.46)
(2.47)
In the above two equations, the relations Gii = Yii cos(ii ) and Bii = Yii sin(ii ) have been used.
From the expressions of the Pi and Qi in equations (2.46) and (2.47) respectively, the elements of
the Jacobian matrix can be calculated as follows.
Matrix J1 (=
P
) (in this case, i = 2, 3, n,
j = 2, 3, n)
n
Pi
= Vi Vk Yik sin(i k ik );
j
k=1
j=i
(2.48)
Pi
= Vi Vj Yij sin(i j ij );
j
Matrix J2 (=
P
) (in this case, i = 2, 3, n,
V
ji
(2.49)
j = (m + 1), (m + 2), n)
n
Pi
= 2Vi Gii + Vk Yik cos(i k ik );
Vj
k=1
j=i
(2.50)
Pi
= Vi Yij cos(i j ij );
Vj
43
ji
(2.51)
Matrix J3 (=
Q
) (in this case, i = (m + 1), (m + 2), n,
j = 2, 3, n)
n
Qi
= Vi Vk Yik cos(i k ik );
j k = 1
j=i
(2.52)
Qi
= Vi Vj Yij cos(i j ij );
j
Matrix J4 (=
Q
) (in this case, i = (m + 1), (m + 2), n,
V
ji
(2.53)
n
Qi
= 2Vi Bii + Vk Yik sin(i k ik );
Vj
k=1
j=i
(2.54)
Qi
= Vi Yij sin(i j ij );
Vj
ji
(2.55)
With these expressions of Jacobian elements given in equations (2.48)-(2.55), the Jacobian matrix
can be evaluated at each iteration as discussed earlier.
Now, in the basic NRLF (polar) algorithm described earlier, the generator Q-limits have not
been considered. To accommodate the generator Q-limits, at the beginning of each iteration, reactive power absorbed or produced by each generator is calculated. If the calculated reactive power
is within the specified limits, the generator is retained as PV bus, otherwise the generator bus is
converted to a PQ bus, with the voltage at this bus no longer held at the specified value. The
detailed algorithm is as follows.
Complete NRLF (polar) algorithm
(0)
(0)
(k)
(k)
b) If, Qmin
Qi Qmax
; then assign Vi = Vispec and the ith bus is retained as PV bus
i
i
for k th iteration.
(k)
(k)
sp
max
min
c) If Qi > Qmax
, then assign Qsp
or, if Qi < Qmin
. In
i
i = Qi
i , then assign Qi = Qi
44
both the cases, this bus is converted to PQ bus. Hence, its voltage magnitude becomes an unknown
for the present iteration (thereby introducing an extra unknown quantity) and to solve for this extra
unknown quantity, an extra equation is required, which is obtained by the new value of Qsp
i (as
th
shown above). Therefore, when the i bus is converted to a PQ bus, the dimensions of both V
and Q vectors increases by one.
In general, if l generator buses (l (m 1)) violate their corresponding reactive power limits at
step 3, then the dimensions of both V and Q vectors increases from (n m) to (n m + l).
However, the dimensions of both P and vectors remain the same. Therefore, the size of
matrix J2 becomes (n 1) (n m + l), that of matrix J3 becomes (n m + l) (n 1) and the
matrix J4 becomes of size (n m + l) (n m + l). The size of matrix J1 , however, does not change.
Hence, the size of the matrix J becomes (2n m 1 l) (2n m 1 l) while the sizes of both
the vectors X and M (in equation (2.45)) becomes (2n m 1 l) 1. Of course, if there is
no generator reactive power limit violation, then l = 0.
(k1)
Step 4: Compute the vectors Pcal and Qcal with the vectors
and V(k1) thereby forming
T
the vector M . Let this vector be represented as M = [M1 , M2 , M2nm1l ] .
Step 5: Compute error = max (M1 , M2 , M2nm1l ).
(k1)
Step 6: If error (pre - specified tolerance), then the final rotation vectors are
and
(k1)
V
and print the results. Otherwise go to step 7.
(k1)
Step 7: Evaluate the Jacobian matrix with the vectors
and V(k1) .
Step 8: Compute the correction vector X by solving equation (2.45).
(k)
Step 9: Update the solution vectors
= (k1) + and V(k) = V(k1) + V . Update
k = k + 1 and go back to step 3.
In the next lecture, we will look at an example of NRLF (polar) technique.
45
2.9.1
As an example, let us consider again the 5-bus system shown in Fig. 2.16. In this System, n = 5 and
m = 3 (as the number of generator in 3). Initially, let us assume that there is no violation of reactive
power limit at any generator. Therefore, the sizes of The Jacobian sub-matinees are as follows;
J1 (4 4), J2 (4 2), J3 (2 4) and J4 (2 2). Therefore, the size of the combined
Jacobian matrix J is (66). As before, the NRLF algorithm starts with flat voltage profile and with
this assumed voltage profile, the different quantities in the first iteration are calculated as shown in
Table 2.10:
Table 2.10: Initial calculation with NRLF (polar) in the 5 bus system
error = 1.15;
As the error is more than the tolerance value ( = 1012 p.u), the algorithm proceeds and in
the next few steps, the Jacobian matrix, correction vector and the updated values of the error are
calculated as shown in Table 2.11.
As the error is still more than the tolerance value, the algorithm continues and enters 2nd
iteration. The values of the relevant qualities in second iteration are shown in Table 2.12 below.
Since the error is still more than the tolerance value, the algorithm continues and finally converges
after 4 iterations. The final converged solution is shown in Table 2.13. Comparsion of Tables 2.13
and 2.4 (GSLF results) shows that the final results obtained by these two methods are identically
same. However, NRLF achieves this solution in 4 iterations as opposed at 69 iterations required by
GSLF. The convergence behavior of both GSLF and NRLF are shown in Fig. 2.17. From this figure
it can be observed that, NRLF has certainly much better convergence characteristics as compared
to GSLF.
Now let us consider the case where reactive power generation of generator 3 is limited to 50
MVAR. The algorithm again starts with a flat start and the initial calculation are same as shown
in Table 2.10 earlier. As Q3 calculated has not crossed the limit, the program proceeds in the same
way as shown in Table 2.11 till the end of first iteration. With the voltage magnitudes and angles
obtained at the end of first iteration, the reactive power generated by all the machines are again
calculated and the value of Q3 is found to be 53.42 MVAR (0.5342 p.u). Hence, bus 3 (generator
3) is converted to a PQ bus and hence, the PQ buses in the system now are (4, 5, 3). The vectors
of calculated injected real and reactive powers, mismatch vector and final mismatch at the end of
1st iteration are shown in Table 2.14. Note that without any generator reactive power violation, the
size of M vector was (6 1) and with generator limit violation in Q3 , the size has now increased
to (7 1).
At the end of 1st iteration, n = 5 and m = 2 (as the number of generator buses is 2). Hence,
the sizes of the various Jacobian sub-matrices are: J1 (4 4), J2 (4 3), J3 (3 4) and
46
Table 2.11: Calculation at 1st iteration with NRLF (polar) in the 5 bus system without any generator
Q limit violation
13.0858
7.4835
J1 =
0
J3 = [
7.4835
0
0
13.0858
7.4835
J=
0
0
0
1.2584 1.12982
J2 =
2.1921 0.9337
0.9337 3.9047
10.5797 3.7348
J4 = [
];
3.7348 15.4091
7.4835
0
0
0
0
V = [0.0783 0.0475]
error = 0.1087;
J4 (3 3). Thus, the size of the Jacobian matrix increases to (7 7). The Jacobian matrix,
correction vector and the updated value of the mismatch as computed in the 2nd iteration are shown
in Table 2.15. As the mismatch is still more than the tolerance value, the algorithm proceeds further
and finally, the algorithm converges in 5 iterations. The final converged values are shown in Table
2.16. In this table also, the NRLF (polar) results without any reactive power limit (as shown in
Table 2.13) are also reproduced for comparison. Moreover, it is observed that the final converged
values are identically same as those calculated by the GSLF method (Table 2.7).
The results with IEEE - 14 bus system are shown in Table 2.17 with and without limit on
generator reactive power. For this case also, the limit on the generator at bus 6 has been maintained
at 30 MVAR. Comparison of Tables 2.8 and 2.17 shows that the results obtained by GSLF and NRLF
are identical, but due to quadratic convergence characteristics, the number of iteration required by
47
Table 2.12: Calculation at 2nd iteration with NRLF (polar) in the 5 bus system without any generator
Q limit violation
13.1998
7.3995
J1 =
0
J3 = [
7.5543
0
0
13.1998
7.3995
J=
0
0
0
0.4065 0.8098
;
J2 =
0.8909 1.0234
0.7191 2.8658
9.3239 3.3977
J4 = [
];
3.5957 14.4487
7.5543
0
0
0
0
V = [0.0783 0.0475] ;
T
error = 0.0026 ;
NRLF to reach the same solution is much less compared to that taken by GSLF for a tolerance of
10e12 p.u.
The results for 30-bus system are shown in Table 2.18. Without any generator Q-limit, the results
obtained by GSLF and NRLF (polar) are identical, although NRLF (polar) takes only 4 iterations
against 851 iterations taken by GSLF. Also,as mentioned earlier, with a tolerance of 10e12 (p.u),
GSLF does not converge for Q limit on gen 11 and 13 (20 and 30 MVAR respectively). However,
NRLF (polar) does not face any such difficulty in convergence in this case and the algorithm converges
with 7 iterations in this case. The corresponding results are also shown in Table 2.18.
In the next lecture, we will discuss another version of NRLF, namely, the rectangular version,
in which, all the complex quantities are represented in the rectangular co-ordinates instead of polar
co-ordinates as is done in the case of NRLF (polar) technique.
48
Table 2.13: Final Results of the 5 bus system with NRLF (polar) without any generator Q limit
violation
Bus no.
1
2
3
4
5
Pinj
Qinj
(p.u)
(deg)
(p.u)
(p.u)
1.0
0
0.56743 0.26505
1.0
1.65757
0.5
-0.18519
1.0
-0.91206
1.0
0.68875
0.90594 -8.35088
-1.15
-0.6
0.94397 -5.02735
-0.85
-0.4
Total iteration = 4
49
Table 2.15: Calculations at 2nd iteration with NRLF (polar) in the 5 bus system for limit on Q3
0
0
1.5250
7.5543
0
0
;
; J2 =
13.1998 7.5543
0
0
0
0
1.5250
0
6.3896 9.6258 3.2363 0.8909 1.0234 1.9289
1.9289
2.9037
0.9748
9.3239
3.3977
6.3896
1.3756
0.6628
4.3554
3.5957
14.4487
4.1771
13.1998
7.3995
J1 =
0
50
Table 2.16: Final Results of the 5 bus system with NRLF (polar) with generator Q limit
Bus no.
1
2
3
4
5
Pinj
Qinj
Pinj
Qinj
(p.u)
(deg)
(p.u)
(p.u)
(p.u)
(deg)
(p.u)
(p.u)
1.0
0
0.56743 0.26505
1.0
0
0.56979 0.33935
1.0
1.65757
0.5
-0.18519
1.0
1.69679
0.5
-0.04769
1.0
-0.91206
1.0
0.68875 0.9825 -0.63991
1.0
0.5
0.90594 -8.35088
-1.15
-0.6
0.88918 -8.35906
-1.15
-0.6
0.94397 -5.02735
-0.85
-0.4
0.93445 -4.98675
-0.85
-0.4
Total iteration = 4
Total iteration = 5
Table 2.17: Final Results of the 14 bus system with NRLF (Polar)
Without generator Q limit
Bus no.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
(p.u)
1.06
1.045
1.04932
1.03299
1.04015
1.07
1.02076
1.0224
1.0201
1.0211
1.04144
1.0526
1.04494
1.01249
Pinj
(deg)
(p.u)
0
2.37259
-5.17113
0.183
-14.54246
-1.19
-10.39269 -0.4779
-8.76418 -0.07599
-12.52265
0.112
-13.44781
0
-13.47154
0
-13.60908 -0.29499
-13.69541
-0.09
-13.22158 -0.03501
-13.42868 -0.06099
-13.50388 -0.135
-14.60128 -0.14901
Total iteration = 4
Qinj
(p.u)
-0.3308
-0.166
-0.08762
-0.039
-0.01599
0.37278
0
-0.129
-0.16599
-0.05799
-0.018
-0.01599
-0.05799
-0.05001
(p.u)
1.06
1.045
1.04697
1.02902
1.03615
1.05497
1.01266
1.01391
1.0118
1.01154
1.02915
1.03787
1.03063
1.00136
51
Pinj
(deg)
(p.u)
0
2.37188
-5.17845
0.183
-14.55556
-1.19
-10.35987 -0.4779
-8.71027 -0.07599
-12.45871
0.112
-13.49478
0
-13.5185
0
-13.66101 -0.29499
-13.73679
-0.09
-13.21814 -0.03501
-13.39145 -0.06099
-13.48166 -0.135
-14.64504 -0.14901
Total iteration = 6
Qinj
(p.u)
-0.31249
-0.1066
-0.08762
-0.039
-0.01599
0.3
0
-0.129
-0.16599
-0.05799
-0.018
-0.01599
-0.05799
-0.05001
Table 2.18: Final Results of the 30 bus system with NRLF (Polar)
Without generator Q limit
Bus no.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
(p.u)
1.05
1.0338
1.03128
1.02578
1.0058
1.02178
1.00111
1.023
1.04608
1.03606
1.0913
1.04859
1.0883
1.03346
1.02825
1.0359
1.0306
1.01873
1.01626
1.02041
1.02305
1.02343
1.0165
1.00939
1.00048
0.9825
1.00379
1.02049
0.98353
0.97181
Pinj
Qinj
Pinj
Qinj
(deg)
(p.u)
(p.u)
(p.u)
(deg)
(p.u)
(p.u)
0
2.38673 -0.29842
1.05
0
2.3865 -0.29386
-4.97945 0.3586 -0.05698 1.0338 -4.98084 0.35861 -0.04562
-7.96653
-0.024
-0.012 1.03045 -7.95523 -0.02399 -0.012
-9.58235
-0.076
-0.016 1.02477 -9.56929 -0.07598 -0.016
-13.60103 -0.6964 0.05042 1.0058 -13.60836 -0.6964 0.05532
-11.50296
0
0
1.02084 -11.49304 0.00003
0
-13.9994
-0.628
-0.109 1.00055 -13.99792 -0.628
-0.109
-12.56853
-0.45
0.12343
1.023 -12.57567
-0.45
0.15111
-13.04088
0
0
1.04006 -13.02865 0.00001
0
-14.88589 -0.058
-0.02
1.03117 -14.8866 -0.05797 -0.02001
-11.16876 0.1793 0.24018 1.07807 -11.12256 0.1793
0.2
-13.74947 -0.112
-0.075 1.04456 -13.75755 -0.11198 -0.075
-12.56078 0.1691 0.31043 1.08311 -12.55854 0.1691
0.3
-14.71704 -0.062
-0.016 1.02931 -14.73168 -0.062
-0.016
-14.86737 -0.082
-0.025 1.02403 -14.88097 -0.08199 -0.025
-14.50539 -0.035
-0.018 1.03148 -14.51198 -0.035
-0.018
-14.98291
-0.09
-0.058 1.02583 -14.98712
-0.09
-0.058
-15.58107 -0.032
-0.009 1.01422 -15.59618 -0.03199 -0.009
-15.81066 -0.095
-0.034 1.01159 -15.8251 -0.09499 -0.034
-15.63819 -0.022
-0.007 1.01568 -15.64961 -0.022
-0.007
-15.35955 -0.175
-0.112 1.01825 -15.36524 -0.17496 -0.11201
-15.35222
0
0
1.01867 -15.3581
0
0
-15.41998 -0.032
-0.016 1.01226 -15.43637 -0.032
-0.016
-15.81043 -0.087
-0.067 1.00517 -15.8277
-0.087
-0.067
-15.84004
0
0
0.99748 -15.86849
0
0
-16.27422 -0.035
-0.023 0.97944 -16.30534 -0.035
-0.023
-15.59587
0
0
1.00158 -15.62827
0
0
-12.1474
0
0
1.01959 -12.14235
0
0
-16.87497 -0.024
-0.009 0.98126 -16.91314 -0.024
-0.009
-17.79427 -0.106
-0.019 0.96951 -17.83675 -0.106
-0.019
Total iteration = 4
Total iteration = 7
52
2.10
In rectangular co-ordinates, every complex quantity is expressed in terms of its real and imaginary
parts. Hence, let Vk = Vk ejk = ek + jfk ; Ii = ai + jci and Yik = gik + jbik .
n
k=1
k=1
Thus, from equation (2.25), Ii = Yik Vk . Or, ai + jci = (gik + jbik )(ek + jfk ).
Or,
k=1
= (bik ek + gik fk )
k=1
ai = (gik ek bik fk )
ci
(2.56)
Complex power injected at bus i is, Si = Pi + jQi = Vi Ii . Or, Pi + jQi = (ei + jfi )(ai jci ).
Or,
(2.57)
k=1
and
(2.58)
k=1
(2.59)
(2.60)
Now, again let us consider a n bus system having m generators (bus 1 being the slack bus). For
each of the (n-m) PQ buses, both voltage magnitude and angle are unknown. In other words, for
these buses both real and imaginary parts of the voltages are unknown. For each of the (m-1) PV
buses, even though the voltage magnitude is known, real and imaginary parts of the voltage are not
known as there can be many combination of ei and fi to give a specified value of Vi . Hence, for
each of the PV buses also, the real and imaginary parts of the voltage are unknown. Therefore, total
number of unknown quantities is = 2(n m) + 2(m 1) = (2n 2). To solve for these (2n 2)
unknown quantities, (2n 2) independent equations are also needed.
Now, let us look at the specified quantities. As discussed earlier, for each of the (n-1) buses,
the quantity Pi is specified [given in equation (2.59)]. Similarly, for each of the (n-m) buses, the
quantity Q1 is also known [given in equation (2.60)]. Thus, total number of specified quantities is
(n 1 + n m) = (2n m 1). Hence, still 2n 2 (2n m 1) = (m 1) specified quantities are
needed to make the NRLF (rectangular) a well posed problem. These additional specified quantities
would be available from the specified voltage magnitudes at each of the (m-1) buses. At each of
these (m-1) PV buses, following relation holds good between the specified quantity (Vi ) and the
53
(2.61)
Thus, in NRLF rectangular co-ordinates method, the unknown quantities are; (ei , fi ) for i =
2, 3, n (total (2n 2) in number). The specified quantities are; a) Pi ; for i = 2, 3, n; b) Qi ;
for i = (m + 1) n and c) Vi ; for i = 2, 3, m. Hence, the total number of specified quantities
is also (2n 2). The relations connecting the unknown quantities to the specified quantities are
given by equations (2.59) - (2.61), which are solved by the standard Newton-Rabhosn technique to
determine the unknown quantities. As in the case of NRLF (polar) method, in NRLF (rectangular)
technique also, flat voltage profile is assumed at the starting.
The standard, linearized form of Newton-Raphron solution of the equations (2.59) - (2.61) are
given below (following the basic Newton-Raphron solution method discussed previously);
P J J
1 2
Q = J3 J4 [e]
f
V2 J5 J6
(2.62)
e
Q
J4
(n 1) (n 1);
(n m) (n 1);
Q
P
(n 1) (n 1); J3
(n m) (n 1);
f
e
V 2
V 2
J5
(m 1) (n 1); J6
(m 1) (n 1).
e
f
J2
[J] [X] = [M ]
(2.63)
In equation (2.63), as before, the quantities M , X and J are known as the mismatch vector,
correction vector and the Jacobian matrix respectively. Note that the vector M and X each
has a size of (2n 2) 1 whereas the Jacobian matrix has a size of (2n 2) (2n 2).
Equation (2.63) forms the basis of NRLF (rectangular) algorithm as discussed below. Note that
the algorithm discussed below assumes that there is no violation of generator reactive power limits.
P
) (in this case, i = 2, 3, n,
e
j = 2, 3, n)
n
Pi
= 2ei gii + (gik ek bik fk );
ej
k=1
j=i
(2.64)
Pi
= (ei gij + fi bij );
ej
Matrix J2 (=
P
) (in this case, i = 2, 3, n,
f
ji
(2.65)
j = 2, 3, n)
n
Pi
= 2fi gii + (bik ek + gik fk );
fj
k=1
j=i
(2.66)
Pi
= (fi gij ei bij );
fj
Matrix J3 (=
ji
Q
) (in this case, i = (m + 1), (m + 2), n,
e
n
Qi
= 2ei bii + (bik ek gik fk );
ej
k=1
(2.67)
j = 2, 3, n)
j=i
(2.68)
Qi
= (fi gij ei bij );
ej
55
ji
(2.69)
Matrix J4 (=
Q
) (in this case, i = (m + 1), (m + 2), n,
f
n
Qi
= 2fi bii + (gik ek bik fk );
fj
k=1
j = 2, 3, n)
j=i
(2.70)
Qi
= (fi bij + ei gij );
fj
Matrix J5 (=
V 2
) (in this case, i = 2, 3, m,
e
Vi2
= 2ei for j = i;
ej
Matrix J6 (=
Vi2
= 2fi for j = i;
fj
(2.71)
j = 2, 3, n)
and
V 2
) (in this case, i = 2, 3, m,
f
ji
Vi2
= 0 for j i;
ej
(2.72)
j = 2, 3, n)
Vi2
and
= 0 for j i;
fj
(2.73)
With these expressions of Jacobian elements, the matrix J can be evaluated at each iteration as
discussed earlier.
Now, in the above algorithm, the violations of generation reactive power limits have not been
taken into account. As in the case of NRLF (polar), in this case also, if a generator violates its Qlimits at any particular iteration, it is treated as a PQ bus for that iteration. Otherwise, it continues
to be treated as a PV bus. The detailed step-by-step procedure for taking the generation Q limit
violation into account in given below.
Complete NRLF (rectangular) algorithm
(0)
(0)
(m l 1) (n 1) respectively. The sizes of the matrices J1 and J2 remain same. Also, the total
sizes of the mismatch vector, correction vector and the Jacobian matrix remain same.
Step 4: Compute the calculated vectors P, Q and V2 from equations (2.59) - (2.61) with
the vectors e(k1) and f (k1) thereby forming the vector M . Let this vector be represented as
T
M = [M1 , M2 , M2n2 ] .
Step 5: Compute error = max (M1 , M2 , M2n2 ).
Step 6: If error (pre - specified tolerance), then the final rotation vectors are e(k1) and
f (k1) and print the results. Otherwise go to step 7.
Step 7: Evaluate the Jacobian matrix with the vectors e(k1) and f (k1) .
Step 8: Compute the correction vector X by solving equation (2.63).
Step 9: Update the solution vectors e(k) = e(k1) +e and f (k) = f (k1) +f . Update k = k +1
and go back to step 3.
In the next lecture, we will look at an example of NRLF (rectangular) technique.
57
2.10.1
Taking the 5 bus system as an example again, the NRLF algorithm starts with the flat voltage profile
and subsequently different quantitative are calculated as shown in Table 2.19.
Table 2.19: Initial calculation with NRLF (rectangular) in the 5 bus system
As the mismatch (= 1.15) is greater than the tolerance (= 1.0e12 ), the algorithm proceeds further
and calculates the Jacobian matrix. Following the discussion presented earlier, the sizes of various
Jacobian sub matrices should be as follows: J1 (4 4); J2 (4 4); J3 (2 4); J4
(24); J5 (24); J6 (24) and J (88). The computed Jacobian matrices are shown
in Table 2.20. From this set observe that the sizes of the calculated Jacobian matrices are indeed the
same as they are indicated above. After calculating the Jacobian matrix, the algorithm calculates
the correction vector (X) and extracts the vectors e and f from X . With these obtained
vectors e and f , the updated vectors e and f are computed and lastly the mismatch vector
(M ) and the final mismatch are calculated. All these calculations are also shown in Table 2.20.
As the mismatch in still more than the tolerance (although it has reduced from the last value), the
algorithm repeats all these calculation and finally converges with 5 iterations. The final load flow
result is shown in Table 2.21. Observe that the result obtained with NRLF (rectangular) method
are identically same as obtained by NRLF (polar) and GSLF techniques.
The load flow solutions of IEEE-14 bus and 30-bus systems have also been computed with NRLF
(rectangular) method. The results are shown in Tables 2.22 and 2.23 and respectively. Again confirm
yourself that the results shown in these two tables are identically same as the corresponding results
shown earlier with NRLF (polar) and GSLF methods.
Now, let us study the behavior of the algorithm when generator Q-limit is considered. Towards
this end, again let us first consider the 5-bus system and as before, let us consider the limit on Q3 to
be 50 MVAR. TO begin with, the initial calculations for this case are identically same as that shown
in Table 2.19. As no violation of Q3 is detected in this initial calculation, the algorithm proceeds as
usual (without any consideration of generation Q-limit violation) and repeats the same calculations
as shown in Table 2.20. As there is still no violation in Q3 , the algorithm advances to 2nd iteration
and the calculations at the end of 2nd iteration are shown in Table 2.24. At the end of 2nd iteration,
it is found that the calculated value of Q3 is 0.6824 p.u. and as a result, this bus is now converted
from a PV bus to a PQ bus. Hence, the PQ buses are now (4, 5, 3) and only bus 2 is retained as a
PV bus. The corresponding calculated vectors Qcal , Vcal and M are shown in Table 2.24 along
with the final value of the mismatch quantity. As this final value is still more than the tolerance,
the algorithm enters the third iteration.
58
Table 2.20: Calculations at 1st iteration with NRLF (rectangular) in the 5 bus system
3.2417 1.8412
0
0
;
J1 =
0
1.2584
2.1921
0.9337
0
1.1298
0.9337
3.9047
1.8412
0
J =
0
2.000
13.0858
7.4835
J2 =
0
7.4835
0
0
0 0 0 0
];
0 0 0 0
1.8412
0
0
13.0858 7.4835
0
0
0
0
0
0
0
0
0
2.0000
0
0
0
0
0
0
T
In the third iteration the numbers of PQ buses is 3 and the number of PV buses is 1 and
hence, the sizes of matrices J3 and J4 change to (3 4) and those of matrices J5 and J6 become
(1 4). The sizes of the matrices J1 , J2 and J remain same as earlier. With this Jacobian matrix,
the correction vector X is calculated and from this vector X , the vectors e and f are
extracted. Subsequently, the vectors e and f are updated. It is found that Q3 still violates the limit
and the mismatch is also found to be still more than the tolerance at the end of the third iteration.
59
Table 2.21: Final Results of the 5 bus system with NRLF (rectangular) with no generator Q limit
Bus no.
1
2
3
4
5
e
f
V
Pinj
Qinj
(p.u)
(p.u)
(p.u)
(deg)
(p.u)
(p.u)
1
0
1
0
0.56743 0.26505
0.99958 0.02893
1
1.65757
0.5
-0.18519
0.99987 -0.01592
1
-0.91206
1
0.68875
0.89634 -0.13157 0.90594 -8.35088
-1.15
-0.6
0.94034 -0.08272 0.94397 -5.02735
-0.85
-0.4
Total iteration = 5
Table 2.22: Final Results of the 14 bus system with NRLF (rectangular) with no generator Q limit
Bus no.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
e
(p.u)
1.06
1.04075
1.0157
1.01604
1.02801
1.04455
0.99277
0.99427
0.99146
0.99207
1.01383
1.02382
1.01605
0.97979
f
V
(p.u)
(p.u)
(deg)
0
1.06
0
-0.09419 1.045
-5.17113
-0.26348 1.04932 -14.54246
-0.18634 1.03299 -10.39269
-0.15849 1.04015 -8.76418
-0.232
1.07
-12.52265
-0.23739 1.02076 -13.44781
-0.23818 1.0224 -13.47154
-0.24002 1.0201 -13.60908
-0.24176 1.0211 -13.69541
-0.23819 1.04144 -13.22158
-0.24445 1.0526 -13.42868
-0.24401 1.04494 -13.50388
-0.25524 1.01249 -14.60128
Total iteration = 5
Pinj
Qinj
(p.u)
2.37259
0.183
-1.19
-0.4779
-0.07599
0.112
0
0
-0.29499
-0.09
-0.03501
-0.06099
-0.135
-0.14901
(p.u)
-0.3308
-0.166
-0.08762
-0.039
-0.01599
0.37278
0
-0.129
-0.16599
-0.05799
-0.018
-0.01599
-0.05799
-0.05001
All the relevant calculations pertaining to 3rd iteration are shown in Table 2.25. As the mismatch
is still more than the tolerance limit, the algorithm proceeds and finally converges with 5 iterations.
The final results are shown in Table 2.26.
Again the load flow solutions of the IEEE-14 bus and IEEE-30 bus system have been computed
for the same generator reactive power limits as taken for GSLF and NRLF (polar) techniques. The
final solutions are shown in Tables 2.27 and 2.28 respectively. Again cross-check for yourself that the
final solution computed by this method are same as those computed by GSLF and NRLF (polar)
techniques.
We will now discuss the fast-decoupled load flow (FDLF) technique in the next lecture.
60
Table 2.23: Final Results of the 30 bus system with NRLF (rectangular) with no generator Q limit
Bus no.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
e
(p.u)
1.05
1.0299
1.02132
1.01147
0.97759
1.00126
0.97138
0.99849
1.0191
1.00129
1.07063
1.01854
1.06225
0.99956
0.99383
1.00288
0.99556
0.98129
0.97781
0.98263
0.98651
0.98691
0.97991
0.9712
0.96249
0.94313
0.96684
0.99764
0.94118
0.92532
f
V
Pinj
Qinj
(p.u)
(p.u)
(deg)
(p.u)
(p.u)
0
1.05
0
2.38673 -0.29842
-0.08973 1.0338 -4.97945 0.3586 -0.05698
-0.14293 1.03128 -7.96653
-0.024
-0.012
-0.17076 1.02578 -9.58235
-0.076
-0.016
-0.23652 1.0058 -13.60103 -0.6964 0.05042
-0.20376 1.02178 -11.50296
0
0
-0.24218 1.00111 -13.9994
-0.628
-0.109
-0.22261 1.023 -12.56853
-0.45
0.12343
-0.23604 1.04608 -13.04088
0
0
-0.26616 1.03606 -14.88589 -0.058
-0.02
-0.21138 1.0913 -11.16876 0.1793 0.24018
-0.24923 1.04859 -13.74947 -0.112
-0.075
-0.23668 1.0883 -12.56078 0.1691 0.31043
-0.26255 1.03346 -14.71704 -0.062
-0.016
-0.26383 1.02825 -14.86737 -0.082
-0.025
-0.25946 1.0359 -14.50539 -0.035
-0.018
-0.26644 1.0306 -14.98291
-0.09
-0.058
-0.27363 1.01873 -15.58107 -0.032
-0.009
-0.27689 1.01626 -15.81066 -0.095
-0.034
-0.27506 1.02041 -15.63819 -0.022
-0.007
-0.27098 1.02305 -15.35955 -0.175
-0.112
-0.27096 1.02343 -15.35222
0
0
-0.27028 1.0165 -15.41998 -0.032
-0.016
-0.27501 1.00939 -15.81043 -0.087
-0.067
-0.27308 1.00048 -15.84004
0
0
-0.27533 0.9825 -16.27422 -0.035
-0.023
-0.26987 1.00379 -15.59587
0
0
-0.21474 1.02049 -12.1474
0
0
-0.2855 0.98353 -16.87497 -0.024
-0.009
-0.29698 0.97181 -17.79427 -0.106
-0.019
Total iteration = 5
Table 2.24: Calculation at the end of 2nd iteration with NRLF (rectangular) in the 5 bus system for
limit on Q3
Vcal = [1.0];
T
61
error = 0.1824;
Table 2.25: Calculations at 3rd iteration with NRLF (rectangular) in the 5 bus system with limit on
Q3
13.3023
3.3566 1.6230
0
0
7.4539
1.9584 5.5167 1.3701 1.2000
; J2 =
J1 =
0
0
2.0669 2.2170 1.3289
0
1.4328 1.1870 4.0987
0
0
6.2313 8.4978 3.2275
1.9584
7.4539 19.5569 7.1104 4.4586
J5 = [1.9992 0 0 0] ;
3.3566
1.9584
0
J =
0
7.4539
1.992
7.5339
0
0
J6 = [0.0581 0 0 0];
1.6230
0
0
13.3023 7.5339
0
0
0
0
0
0.0581
0
0
0
T
Vcal = [1.0] ;
T
62
Table 2.26: Final Results of the 5 bus system with NRLF (rectangular) with limit on Q3
Bus no.
1
2
3
4
5
e
f
V
Pinj
Qinj
(p.u)
(p.u)
(p.u)
(deg)
(p.u)
(p.u)
1
0
1
0
0.56979 0.33935
0.99956 0.02961
1
1.69679
0.5
-0.04769
0.98244 -0.01097 0.9825 -0.63991
1
0.5
0.87973 -0.12927 0.88918 -8.35906
-1.15
-0.6
0.93092 -0.08123 0.93445 -4.98675
-0.85
-0.4
Total iteration = 5
Table 2.27: Final Results of the 14 bus system with NRLF (rectangular) with generator Q limits
Bus no.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
e
(p.u)
1.06
1.04073
1.01336
1.01224
1.0242
1.03013
0.9847
0.98582
0.98318
0.98261
1.00189
1.00965
1.00223
0.96883
f
V
(p.u)
(p.u)
(deg)
0
1.06
0
-0.09432 1.045
-5.17845
-0.26312 1.04697 -14.55556
-0.18505 1.02902 -10.35987
-0.15691 1.03615 -8.71027
-0.22759 1.05497 -12.45871
-0.23631 1.01266 -13.49478
-0.23701 1.01391 -13.5185
-0.23896 1.0118 -13.66101
-0.2402 1.01154 -13.73679
-0.23533 1.02915 -13.21814
-0.24037 1.03787 -13.39145
-0.24028 1.03063 -13.48166
-0.25317 1.00136 -14.64504
Total iteration = 8
63
Pinj
Qinj
(p.u)
2.37188
0.183
-1.19
-0.4779
-0.07599
0.112
0
0
-0.29499
-0.09
-0.03501
-0.06099
-0.135
-0.14901
(p.u)
-0.31249
-0.1066
-0.08762
-0.039
-0.01599
0.3
0
-0.129
-0.16599
-0.05799
-0.018
-0.01599
-0.05799
-0.05001
Table 2.28: Final Results of the 30 bus system with NRLF (rectangular) with generator Q limits
Bus no.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
e
(p.u)
1.05
1.0299
1.02053
1.01051
0.97756
1.00036
0.97083
0.99845
1.01327
0.99655
1.05781
1.01458
1.05718
0.99547
0.98968
0.99856
0.99092
0.97686
0.97323
0.97802
0.98184
0.98228
0.97574
0.96705
0.95946
0.94003
0.96454
0.99677
0.93881
0.9229
f
V
Pinj
Qinj
(p.u)
(p.u)
(deg)
(p.u)
(p.u)
0
1.05
0
2.38676 -0.29389
-0.08977 1.0338 -4.98141 0.3586 -0.04555
-0.14263 1.03044 -7.95615
-0.024
-0.012
-0.17038 1.02477 -9.57038
-0.076
-0.016
-0.23666 1.0058 -13.6093 -0.6964 0.05533
-0.20342 1.02084 -11.49427
0
0
-0.24204 1.00055 -13.99903 -0.628
-0.109
-0.22276 1.023 -12.57695
-0.45
0.15116
-0.2345 1.04005 -13.0305
0
0
-0.26495 1.03117 -14.8887
-0.058
-0.02
-0.208 1.07807 -11.1244 0.1793
0.2
-0.24844 1.04456 -13.75943 -0.112
-0.075
-0.23554 1.08311 -12.56043 0.1691
0.3
-0.26178 1.02931 -14.73367 -0.062
-0.016
-0.26302 1.02403 -14.88301 -0.082
-0.025
-0.25851 1.03148 -14.51398 -0.035
-0.018
-0.26532 1.02583 -14.98919
-0.09
-0.058
-0.27271 1.01421 -15.59837 -0.032
-0.009
-0.2759 1.01159 -15.82733 -0.095
-0.034
-0.27402 1.01568 -15.65181 -0.022
-0.007
-0.26984 1.01825 -15.36745 -0.175
-0.112
-0.26983 1.01867 -15.36029
0
0
-0.26947 1.01226 -15.43846 -0.032
-0.016
-0.27419 1.00517 -15.82979 -0.087
-0.067
-0.27277 0.99748 -15.87039
0
0
-0.27501 0.97944 -16.30724 -0.035
-0.023
-0.26985 1.00158 -15.62998
0
0
-0.21448 1.01959 -12.14364
0
0
-0.2855 0.98126 -16.91488 -0.024
-0.009
-0.297 0.96951 -17.83848 -0.106
-0.019
Total iteration = 5
64
2.11
An important and useful property of power system is that the change in real power is primarily
governed by the charges in the voltage angles, but not in voltage magnitudes. On the other hand,
the charges in the reactive power are primarily influenced by the charges in voltage magnitudes, but
not in the voltage angles. To see this, let us note the following facts:
(a) Under normal steady state operation, the voltage magnitudes are all nearly equal to 1.0.
(b) As the transmission lines are mostly reactive, the conductances are quite small as compared to the susceptance (Gij << Bij ).
(c) Under normal steady state operation the angular differences among the bus voltages are
quite small (i j 0 (within 5o 10o )).
(d) The injected reactive power at any bus is always much less than the reactive power
consumed by the elements connected to this bus when these elements are shorted to the ground
(Qi << Bii Vi2 ).
With these facts at hand, let us re-visit the equations for Jacobian elements in Newton-Raphson
(polar) method (equation (2.48) to (2.55)). From equations (2.50) and (2.51) we have,
n
Pi
= 2Vi Gii + Vk Yik cos(i k ik )
Vj
k=1
i
n
j=i
(2.74)
Pi
= Vi Yij cos(i j ij )
Vj
= Vi Yij [cos(i j ) cos ij + sin(i j ) sin ij ]
= Vi [Gij cos(i j ) + Bij sin(i j )] ;
ji
(2.75)
Now, Gii and Gij are quite small and negligible and also cos(i j ) 1 and sin(i j ) 0, as
[(i j ) 0]. Hence,
Pi
Pi
0 and
0
Vi
Vj
J2 0
(2.76)
j=i
(2.77)
Qi
= Vi Vj [Gij cos(i j ) + Bij sin(i j )] ;
j
65
ji
(2.78)
Again in light of the natures of the quantities Gii , Gij and (i j ) as discussed above,
Qi
Qi
0 and
0
i
j
J3 0
(2.79)
Substituting equations (2.76) and (2.79) into equation (2.40) one can get,
P
J1 0
[
]=[
][
]
Q
0 J4 V
(2.80)
j=i
j=i
k=1
j=i
Pi
= Vi Vj Yij sin(i j ij ); j i
j
= Vi Vj Yij [sin(i j ) cos ij cos(i j ) sin ij ] ;
= Vi Vj [Gij sin(i j ) Bij cos(i j )] ;
= Vi Vj Bij ;
(2.81)
ji
ji
ji
(2.82)
or,
j=i
Qi
Vi = 2Vi2 Bii + Vi Vk Yik sin(i k ik );
Vj
k=1
j=i
Qi
Vi = Vi2 Bii + Vi Vk Yik sin(i k ik ) = Qi Vi2 Bii ;
Vj
k=1
Qi
Vi = Vi2 Bii ; j = i [as Qi << Bii Vi2 ]
or,
Vj
Qi
or,
= Vi Bii ; j = i
Vj
or,
66
j=i
(2.83)
Qi
= Vi Yij sin(i j ij ); j i
Vj
= Vi Yij [sin(i j ) cos ij cos(i j ) sin ij ] ;
= Vi [Gij sin(i j ) Bij cos(i j )] ;
Vi Bij ;
ji
ji
ji
(2.84)
n
n
Pi
= Vk Bik k
Vi
k=1
(2.85)
Now, as Vi 1.0 under normal steady state operating condition, equation (2.85) reduces to,
n
Pi
= Bik k .
Vi
k=1
Or,
P
= [B] .
V
Or,
= [B ]
V
(2.86)
Matrix B is a constant matrix having a dimension of (n 1) (n 1). Its elements are the
negative of the imaginary part of the element (i, k) of the YBUS matrix where i = 2, 3, n and
k = 2, 3, n.
Again combining equations (2.80), (2.83) and (2.84) we get,
Qi = Vi Bik Vk .
k=1
Or,
n
Qi
= Bik Vk .
Vi
k=1
Or,
= [B ] V
V
(2.87)
Again, [B ] is also a constant matrix having a dimension of (n m) (n m). Its elements are the
negative of the imaginary part of the element (i, k) of the YBUS matrix where i = (m + 1), (m +
2), n and k = (m + 1), (m + 2), n. As the matrixes [B ] and [B ] are constant, it is not
necessary to invert these matrices in each iteration. Rather, the inverse of these matrices can be
stored and used in every iteration, thereby making the algorithm faster. Further simplification in
the FDLF algorithm can be made by,
a. Ignoring the series resistances is calculating the elements of [B ]. Also, by omitting the
elements of [B ] that predominantly affect reactive power flows, i.e., shunt reactances and
transformer off nominal in phase taps.
b. Omitting from [B ] the angle shifting effect of phase shifter, which predominantly affects real
power flow.
2.11.1
As an example, the 5 bus system described earlier is considered again. Starting from the flat start,
the initial calculations are shown in Table 2.29. As the mismatch is more than the tolerance, the
BUS matrix of this system is shown in Table 2.30. In this table, the
algorithm proceeds. The Y
BUS (, m n) represents the elements (of the Y
BUS matrix) corresponding to all rows
notation Y
(denoted by the notation :) and columns spanning from mth column to nth column (denoted by the
BUS matrix, the matrices [B ] and [B ] are constructed as shown in
notation m:n). From this Y
Table 2.30. Note that the size of the matrix [B ] is (4 4) (corresponding to the non slack buses,
i.e. 2, 3, 4 and 5) and the size of the [B ] matrix is (2 2) (corresponding to the PV buses 4
and 5). Also note that the matrices [B ] and [B ] have been formed by taking the negative of the
BUS matrix. With these constant matrices,
imaginary parts of the corresponding elements of the Y
the vectors and V are calculated and subsequently, the vectors and V have been updated.
With these updated values of and V, the final mismatch (error) is again calculated as shown in
Table 2.30. As the error is still more than the tolerance, the algorithm proceeds further and finally
converges in 19 iterations for a tolerance value of 1012 p.u. The final solution is shown in Table
2.31, which happens to be the same as the results obtained by the other methods described earlier.
Now, let us impose the reactive power limit on the generation at bus 3. Starting from the flat
start, the calculation up to 1st iteration are same and hence are not shown here. The calculation
pertaining to 2nd iteration are shown in Table 2.32. From this set, observe that the calculated value
of Q3 has exceeded the limit of 50 MVAR and hence it should be now treated as PQ-bus. Thus,
the dimension of the matrix [B ] increases to (3 3) (corresponding to the buses 4, 5 and 3).
This new, augmented matrix is shown in Table 2.33. With these [B ] and [B ] matrices ([B ]
matrix remains the same), the calculations are further carried out and the algorithm converges in
20 iterations. The final solution is shown in Table 2.31. Comparison of this result with the earlier
results (obtained with other methods) shows that because of the approximations made in FDLF, the
results obtained with FDLF are not identically the same with those obtained by the other methods
but are very close.
The results corresponding to 14-bus and 30-bus systems are shown in Tables 2.34 and 2.35
respectively. From these tables note that the number of iterations taken by FDLF is much higher than
those required by NRLF. This is because of approximations adopted by FDLF (to achieve decoupling)
due to which, the convergence is slower. However, because of the constant Jacobrian matrices, the
68
Table 2.30: Calculations at 1st iteration with FDLF in the 5 bus system
0
1.8412 + 7.4835i 4.2294 18.9271i ;
YBUS (, 1 3) =
0
0
1.2584
+
7.1309i
1.8412 + 7.4835i
0
1.1298
+
4.4768i
0
1.8412 + 7.4835i
0
0
13.0138 7.4835
0
0
; [B ] = [
[B ] =
];
0
7.1309 10.7227 3.7348
3.7348 15.5521
0
4.4768 3.7348 15.5521
V = [0.0528 0.0292] ;
T
execution of each iteration is much faster (as the Jacobrian matrix need not be recomputed and
reversed at each iteration) and hence, the total time taken by FDLF is quiet comparable to that
needed by NRLF. From the two tables it is further noted that, in the absence of any generation
reactive power limit, the results obtained by FDLF are almost identical to those obtained by other
methods. However, in the presence of generation reactive power limits, FDLF results are quiet close
to the results obtained by other methods, though not identical.
We are now at the end of our discussion of AC load flow techniques. In the next lecture, we will
study the method of load flow analysis of an AC system in which a HVDC link is also embedded
(namely, the AC-DC load flow technique).
69
Bus no.
1
2
3
4
5
Pinj
Qinj
Pinj
Qinj
(p.u)
(deg)
(p.u)
(p.u)
(p.u)
(deg)
(p.u)
(p.u)
1.0
0
0.56743 0.26505
1.0
0
0.56985 0.34069
1.0
1.65757
0.5
-0.18519
1.0
1.69742
0.5
-0.04522
1.0
-0.91206
1.0
0.68875 0.98219 -0.63507
1.0
0.49668
0.90594 -8.35088
-1.15
-0.6
0.88888 -8.35938
-1.15
-0.6
0.94397 -5.02735
-0.85
-0.4
0.93428 -4.9861
-0.85
-0.4
Total iteration = 19
Total iteration = 20
Table 2.32: Calculations at 2nd iteration with FDLF in the 5 bus system with limit on Q3
T
V = [0.0399 0.0277] ;
T
70
Table 2.33: Calculations at 3rd iteration with FDLF in the 5 bus system with limit on Q3
13.0138
7.4835
[B ] =
0
7.4835
0
0
Bus no.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
(p.u)
1.06
1.045
1.04932
1.03299
1.04015
1.07
1.02076
1.0224
1.0201
1.0211
1.04144
1.0526
1.04494
1.01249
Pinj
(deg)
(p.u)
0
2.37259
-5.17113
0.183
-14.54246
-1.19
-10.39269 -0.4779
-8.76418 -0.07599
-12.52265
0.112
-13.44781
0
-13.47154
0
-13.60908 -0.29499
-13.69541
-0.09
-13.22158 -0.03501
-13.42868 -0.06099
-13.50388 -0.135
-14.60128 -0.14901
Total iteration = 123
Qinj
(p.u)
-0.3308
-0.166
-0.08762
-0.039
-0.01599
0.37278
0
-0.129
-0.16599
-0.05799
-0.018
-0.01599
-0.05799
-0.05001
(p.u)
1.06
1.045
1.04697
1.02902
1.03615
1.05497
1.01266
1.01391
1.0118
1.01154
1.02915
1.03787
1.03063
1.00136
71
Pinj
(deg)
(p.u)
0
2.37188
-5.17845
0.183
-14.55556
-1.19
-10.35987 -0.4779
-8.71026 -0.07599
-12.4587
0.112
-13.49478
0
-13.5185
0
-13.66102 -0.29499
-13.73679
-0.09
-13.21814 -0.03501
-13.39144 -0.06099
-13.48166 -0.135
-14.64504 -0.14901
Total iteration = 119
Qinj
(p.u)
-0.31249
-0.1066
-0.08762
-0.039
-0.01599
0.29999
0
-0.129
-0.16599
-0.05799
-0.018
-0.01599
-0.05799
-0.05001
Bus no.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
(p.u)
1.05
1.0338
1.03128
1.02578
1.005 8
1.02178
1.00111
1.023
1.04608
1.03606
1.0913
1.04859
1.0883
1.03346
1.02825
1.0359
1.0306
1.01873
1.01626
1.02041
1.02305
1.02343
1.0165
1.00939
1.00048
0.9825
1.00379
1.02049
0.98353
0.97181
Pinj
(deg)
(p.u)
0
2.38673
-4.97945 0.3586
-7.96653
-0.024
-9.58235
-0.076
-13.60103 -0.6964
-11.50296
0
-13.9994
-0.628
-12.56853
-0.45
-13.04088
0
-14.88589 -0.058
-11.16876 0.1793
-13.74947 -0.112
-12.56078 0.1691
-14.71704 -0.062
-14.86737 -0.082
-14.50539 -0.035
-14.98291
-0.09
-15.58107 -0.032
-15.81066 -0.095
-15.63819 -0.022
-15.35955 -0.175
-15.35222
0
-15.41998 -0.032
-15.81043 -0.087
-15.84004
0
-16.27422 -0.035
-15.59587
0
-12.1474
0
-16.87497 -0.024
-17.79427 -0.106
Total iteration = 115
Qinj
(p.u)
-0.29842
-0.05698
-0.012
-0.016
0.05042
0
-0.109
0.12343
0
-0.02
0.24018
-0.075
0.31043
-0.016
-0.025
-0.018
-0.058
-0.009
-0.034
-0.007
-0.112
0
-0.016
-0.067
0
-0.023
0
0
-0.009
-0.019
(p.u)
1.05
1.0338
1.03026
1.02455
1.0058
1.0206
1.00041
1.023
1.03842
1.02995
1.07426
1.04384
1.0824
1.02853
1.0232
1.03054
1.02469
1.01323
1.01052
1.01457
1.01706
1.0175
1.01136
1.00418
0.99677
0.97872
1.00105
1.01937
0.98073
0.96897
72
Pinj
(deg)
(p.u)
0
2.38678
-4.98182 0.3586
-7.95403
-0.024
-9.56795
-0.076
-13.6111 -0.6964
-11.49172
0
-13.9986
-0.628
-12.5786
-0.45
-13.02612
0
-14.88705 -0.058
-11.11026 0.1793
-13.76458 -0.112
-12.56395 0.1691
-14.73984 -0.062
-14.88791 -0.082
-14.51633 -0.035
-14.98908
-0.09
-15.60231 -0.032
-15.83038 -0.095
-15.65374 -0.022
-15.36723 -0.175
-15.36018
0
-15.44311 -0.032
-15.83356 -0.087
-15.87684
0
-16.31432 -0.035
-15.63737
0
-12.14228
0
-16.92364 -0.024
-17.84827 -0.106
Total iteration = 112
Qinj
(p.u)
-0.29289
-0.0429
-0.012
-0.016
0.05654
0
-0.109
0.15797
0
-0.02
0.18809
-0.075
0.29991
-0.016
-0.025
-0.018
-0.058
-0.009
-0.034
-0.007
-0.112
0
-0.016
-0.067
0
-0.023
0
0
-0.009
-0.019
2.12
For solving the load flow problem of an A.C. system in which one or more HVDC links are present,
either of the following two approaches are followed;
a. Simultaneous solution technique
b. Sequential solution technique
In simultaneous solution technique, the equations pertaining to the A.C. system and the equations pertaining to the DC system are solved together. In the sequential method, the AC and DC
systems are solved separately and the coupling between the AC and DC system in accomplished
by injecting an equivalent amount of real and reactive power at the terminal AC buses. In other
words, for an HVDC link existing between buses i and j of an AC system (rectifier at bus i
and inverter at bus j), the effect of the DC link in incorporated into the AC system by injections
(R)
(R)
(I)
(I)
PDCi
and QDCi at the rectifier bus i and PDCj and QDCj at bus j (the super scripts R and
I denote the rectifier and inverter respectively). Therefore the net injected power at bus i and
(R)
(R)
(I)
(I)
j are: Pitotal = PACi + PDCi ; QTi otal = QACi + QDCi ; PjT otal = PACj + PDCj ; QTj otal = QACj + QDCj .
With these net injected powers the AC system is again solved and subsequently, the equivalent in(R)
(R)
(I)
(I)
jected powers (PDCi , QDCi , PDCj , QDCi ) and the total injected powers (PiT otal , QTi otal , PjT otal , QTj otal )
are updated. This process of alternately solving AC and DC system quantities is continued till the
changes in AC system and DC system quantities between two consecutive iterations become less then
a threshold value. Although simultaneous technique gives the solution of the system without any
to and fro switching between the AC and DC systems, the sequential solution technique is actually
quite easy to implement as we will see later. Now let as look at the equations of the DC system.
2.12.1
DC system model
For deriving a suitable model of a HVDC system for steady state operation, few basic assumptions
are adopted as described below;
a. The three A.C. voltages at the terminal bus bar are balanced and sinusoidal.
b. The converter operation is perfectly balanced.
c. The direct current and voltages are smooth.
d. The converter transformer is lossless and the magnetizing admittance is ignored.
With the above assumptions, the equivalent circuit of the converter (either rectifier or inverter)
is shown in Fig. 2.18. In this figure, the notations are as follows;
73
Figure 2.18: Equivalent circuit of the converter under the steady state operation
Vt t
a
Es s
Ip , Is
Vd , Id
It is to be noted that in Fig. 2.18, the angles are referred to the common reference of the entire
AC-DC system. With the above notations, the basic equation governing the HVDC systems are as
follow:
For rectifier
3 2
Vdr =
Nr ar Etr cos r = Vdor cos r
3
Vdr = Vdor cos Xcr Nr Id
For inverter
3 2
Vdi =
Ni ai Eti cos i = Vdoi cos i
3
Vdi = Vdoi cos Xci Ni Id
(2.88)
(2.89)
(2.90)
(2.91)
In the above equations, the subscripts r and i denote the rectifier and inverter side respectively.
The quantity N denotes the number of six-pulse bridges at any partienlar side and the angle
denotes the angular difference between the terminal voltages and primary current of the transformer,
i.e. the power factor of the converter as seen by the AC bus. Xc denotes the commutating reactance
of the converter transformer and the angles and denote the firing angle of the rectifier and
the extinction angle of the inverter respectively.
74
The rectifie and the inverter are interconnected though the following equation:
Vdr Vdi
= Id
Rd
(2.92)
In equation (2.92), the quantity Rd denotes the DC link resistance. Equations (2.88)-(2.92)
describe the operation of a two-terminal HVDC link. Now, as the basic objective of a HVDC link
is to provide complete controllability of power over a transmission corridor, both the rectifies and
the inverter stations are suitably controlled and thus, suitable control equations also need to be
incorporated in the above model. We will discuss these control equations shortly. However, to solve
the above equations, appropriate solution variables must be chosen. Now, for the reason of simplicity,
following set of solution variables is chosen for each converter;
x = [Vd Id a cos ]
(2.93)
Therefore, for a two terminal HVDC link, the complete set of solution vector is;
(2.94)
In equation (2.94), Id has been taken only once as the DC current is same at both the ends. From
equation (2.94) it is observed that there are total 9 unknown variables which need to be solved to
completely determine the HVDC link. However, we have only 5 independent equations as shown in
equations (2.88)-(2.92). Therefore, out of 9 unknown variables, any 4 variables need to be specified
and thereafter, remaining 5 variables can be solved using equations (2.88)-(2.92).
These 4 variables can be specified using the control specification. There can be several combinations of control specification and some of their combination are;
i) , Pdr , , Vdi ;
iii) ar , Pdr , ai , Vdi ;
v)
ar , Pdr , , ai ;
vii)
, Id , , Vdi ;
With any of these four specified control values, the remaining 5 variables can be solved from
equations (2.88)-(2.92) by using standard Newton-Raphoson technique. However, for the sequential
(R)
(R)
(I)
(I)
solution techniques, the quantities PDci , QDci , PDcj and QDcj can be competed in a much easier
way by algebraic manipulation of equations (2.88)-(2.92). we will show this procedure by two of the
eight combinations listed above.
75
Combination 1
In this case, , Pdr , and Vdi are specified. With these known quantities, the calculation procedure
is as follows:
Step 1: We know, Pdr = Vdr Id .
Or, Pdr =
Vdr =
Or,
Vdi
(2.95)
From equation (2.95), two values of Vdr are obtained. Out of these two values, the value of Vdr
which is greater than Vdi is chosen, i.e.
1
Vdr = (Vdi + Vdi2 + 4Rd Pdr )
2
Step 2: Id is calculated as,
Id =
Pdr
Vdr
(2.96)
(2.97)
3
Vdr + Xcr Nr Id
Vdor =
cos
(2.98)
cos r =
Vdr
Vdor
Vdor
ar =
3 2Nr Etr
(2.99)
(2.100)
In equation (2.100) Etr is known as in the sequential solution method, the terminal voltages are
known from the immediate past solution of the AC system equations.
(R)
(R)
Step 5: The quantities PDCi and QDCi are calculated as;
(R)
PDCi
= Pdr
Q(R)
DCi = Pdr tan r
and
(2.101)
3
Vdi + Xci Ni Id
Vdoi =
cos
76
(2.102)
cos i =
Vdi
Vdoi
(2.103)
Vdoi
ai =
3 2Ni Eti
(I)
(2.104)
(I)
(R)
(I)
Q(I)
DCj = PDCj tan i
and
(I)
(2.105)
(I)
With these values of PDCi , QDCi , PDCj and QDCj , the AC system equations are again solved
to obtain the updated values of Etr and Eti and subsequently, steps (1)-(8) are repeated again to
(R)
(R)
(I)
(I)
update the values of PDCi , QDCi , PDCj and QDCj . This alternate process of solving AC and DC
system equations are repeated till convergence in obtained.
Combination 8
In this case, , , Pdi and Vdr are known. With these known quantities, the calculation procedure
is as follows:
Vdr
(2.106)
From the two values of Vdi in equation (2.106), the final value of Vdi is calculated as,
1
Vdi = (Vdr + Vdr2 4Rd Pdi )
2
Step 2: Id is calculated as,
Id =
(2.107)
Pdi
Vdi
(2.108)
With these calculated values of Vdi and Id , steps (3)-(8) of combination-1 are followed to calculate
(R)
the Equivalent power injection values, where PDCi = Vdr Id . With these injected power values, the
AC and DC systems are continued to be solved alternately till convergence in achieved. It is to be
R
noted that at the rectifier end, P(DCi)
= Pdr and Q(R)
DCi = Qdr as the rectifier draws both real and
(I)
I
reactive power from the grid. On the other hand, at the inverter end, P(DCj)
= Pdi and QDCj = Qdi
as the inverter supplies real power to the AC grid and draws reactive power from the AC grid.
In the next lecture, we will look at an example of AC-DC load flow method.
77
2.12.2
To illustrate the application of the above procedure, let us first consider the 5-bus system. In this
system, it is now assumed that one bipolar HVDC link is connected between bus 4 and 5 (rectifier
at bus 4 and inverter at bus 5). Other relevant data for this link are as follows; Rd = 10.0 ;
Nr = Ni = 2;
3
3
Xcr = Xci = 6.0 . Further, let us also assume that the specified values have been
Table 2.36: Final Results of AC-DC load flow of 5 bus system without any generator Q limit violation
Bus no.
1
2
3
4
5
Pinj
Qinj
(p.u)
(deg)
(p.u)
(p.u)
1.0
0
0.68984 0.46301
1.0
-0.63995
0.5
-0.17235
1.0
-4.91128
1.0
1.54134
0.82813 -17.48682
-2.15
-0.76277
0.91332 -3.89028 0.13449 -0.75025
Total iteration = 5
the DC system) are constant. Therefore, by pre-calculating these equivalent power injections and
subsequently incorporating these calculated values into net bus power injections, standard AC load
flow solution can be computed to obtain the solution of the composite AC-DC system.
From the above discussion regarding combination 1 and 8, it may appear that the equivalent real
and reactive power injections (representing the DC system) are always constant for any combination
of the specified control variables. However, this is not true. Depending on the specified control
variables, the equivalent real and reactive power injections may vary from iteration to iteration and
therefore, they need to be calculated in every iteration. As an example, let us consider combination
3. For this combination, the various steps are as follows:
Step 1: Initialise all the bus voltages with flat start. Hence, Etr and Eti are known.
Step 2: From the specified values of Pdr and Vdi , calculate Vdr and Id using equations (2.96)
and (2.97) respectively.
Step 3: Calculate Vdor from equation (2.100).
Step 4: Calculate cos and cos r using equations (2.98) and (2.99) respectively.
(R)
Step 5: From the knowledge of Pdr and cos r , calculate QDCi using equation (2.101).
Step 6: Calculate Vdoi from equation (2.104).
Step 7: Calculate cos and cos i using equations (2.102) and (2.103) respectively.
(I)
(I)
Step 8: Calculate PDCj and QDCj from equation (2.105).
(R)
Please note that in steps 3-5, the quantities Vdor , cos r and QDCi are all dependent on the
(I)
rectifier side AC bus voltage, Etr . Similarly, in steps 6-8, the quantities Vdoi , cos i and QDCj are all
(I)
dependent on the inverter side AC bus voltage, Eti . The quantity PDCj however, depends only on
the DC system quantities and hence remain constant. Thus, the equivalent reactive power injections
at both rectifier and inverter side depend on the AC bus voltage magnitudes (although the equivalent
real power injections at both the sides are independent of AC bus voltage magnitudes). Hence, the
equivalent reactive power injections need to be updated at each iteration and with these updated
power injection values, another iteration of AC load flow is carried out. This process is continued till
convergence is achieved. To illustrate this procedure further, let us assume that the specified values
corresponding to combination 3 are as follows:
79
In
0
1
2
V4
V5
Vdor
Q(R)
DCi
(p.u.) (p.u.)
(kV)
(rad.) (MVAR)
1.0
1.0
356.52 0.778
98.54
0.7843 0.8731 279.63 0.432
46.11
0.7682 0.8703 273.91 0.3842
40.43
Vdoi
Q(I)
DCj
(deg.) (kV)
(rad.) (MVAR)
43.48 356.52 0.7937
100.09
22.33 311.29 0.638
73.04
19.20 310.29 0.634
72.37
MM
(deg.)
44.4
2.15
35.08 0.5578
34.82 0.038
Table 2.38: Final Results of AC-DC load flow of 5 bus system for combination 3 without any
generator Q limit violation
Bus no.
1
2
3
4
5
Pinj
(p.u)
(deg)
(p.u)
1.0
0
0.75382
1.0
-0.99374
0.5
1.0
-5.52481
1.0
0.82813 -18.69153
-2.15
0.91332 -3.65137 0.13449
Total iteration = 70
Qinj
(p.u)
0.78420
-0.16893
2.17338
-1.01522
-1.12440
For further illustration, let us now consider the 30-bus system. In this system, it is now assumed
that one bipolar HVDC link is connected between bus 9 and 28 (rectifier at bus 9 and inverter at bus
28). Other relevant data for this link are as follows; Rd = 10.0 ; Nr = Ni = 2;
3
3
Xcr = Xci = 6.0 .
The load flow has been solved for combination-1, combination-3 and combination-8 (of specified quantities). The specified values which have been considered are as follows;
Combination-1
80
Table 2.39: Results of the 30 bus system with a bipolar HVDC link between bus 9 and 28
Bus no.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
(p.u)
1.05
1.0338
1.02345
1.01637
1.0058
1.01055
0.99437
1.023
1.01796
1.00713
1.0913
1.04132
1.0883
1.02408
1.01562
1.01841
1.00471
0.99979
0.9938
0.99623
0.99497
0.99578
1.00064
0.98942
0.98869
0.97048
0.99711
1.00723
0.97669
0.96489
With combination 1
Pinj
Qinj
With combination 8
Pinj
Qinj
(deg)
(p.u)
(p.u)
(p.u)
(deg)
(p.u)
(p.u)
0
2.42039 -0.26099
1.05
0
2.42229 -0.26023
-5.0387
0.3586 0.06523 1.0338 -5.04273
0.3586
0.06857
-8.08233
-0.024
-0.012 1.02325 -8.08809
-0.024
-0.012
-9.72937
-0.076
-0.016 1.01613 -9.73654
-0.076
-0.016
-13.69708 -0.6964 0.10883 1.0058 -13.70427 -0.6964 0.11036
-11.35395
0
0
1.01026 -11.35703
0
0
-13.96765 -0.628
-0.109 0.99419 -13.97288 -0.628
-0.109
-12.18127
-0.45
0.47454
1.023 -12.18265
-0.45
0.48393
-20.00269
-1
-0.16277 1.01711 -20.13217 -1.01653 -0.16832
-19.52942 -0.058
-0.02
1.00633 -19.6175
-0.058
-0.02
-18.07884 0.1793 0.38778 1.0913 -18.20671 0.1793
0.39224
-16.16702 -0.112
-0.075 1.04109 -16.21744 -0.112
-0.075
-14.97002 0.1691 0.36696 1.0883 -15.02017 0.1691
0.36879
-17.34016 -0.062
-0.016 1.02379 -17.39439 -0.062
-0.016
-17.64022 -0.082
-0.025 1.01524 -17.69657 -0.082
-0.025
-17.83191 -0.035
-0.018 1.01792 -17.89732 -0.035
-0.018
-19.23196
-0.09
-0.058 1.00399 -19.31323
-0.09
-0.058
-19.02914 -0.032
-0.009 0.99925 -19.09708 -0.032
-0.009
-19.66267 -0.095
-0.034 0.99317 -19.73758 -0.095
-0.034
-19.69028 -0.022
-0.007 0.99555 -19.76851 -0.022
-0.007
-19.72658 -0.175
-0.112 0.99419 -19.81012 -0.175
-0.112
-19.62149
0
0
0.99501 -19.7033
0
0
-18.30125 -0.032
-0.016 1.00016 -18.3595
-0.032
-0.016
-18.8339
-0.087
-0.067 0.98881 -18.89458 -0.087
-0.067
-16.70952
0
0
0.98823 -16.73336
0
0
-17.15432 -0.035
-0.023 0.97001 -17.17859 -0.035
-0.023
-15.12179
0
0
0.99676 -15.12284
0
0
-9.39718 0.98449 -0.35025 1.00683 -9.35718
1
-0.35723
-16.41847 -0.024
-0.009 0.97633 -16.42047 -0.024
-0.009
-17.35084 -0.106
-0.019 0.96452 -17.35353 -0.106
-0.019
Total iteration = 4
Total iteration = 4
DC system solutions
Vdr = 253.938 kV; idr = 393.8 Amp.;
ar = 0.7088; Qdr = 16.276 MVAR;
ai = 0.7389; Pdi = 98.45 MW;
Qdi = 35.024 MVAR;
DC system solutions
Vdi = 245.93 kV; idr = 406.613 Amp.;
ar = 0.6988; Pdr = 101.65 MW;
ai = 0.7275; Qdr = 16.831 MVAR;
Qdi = 35.723 MVAR;
We are now at the end of our theoretical study of various load flow techniques. However, in
production grade implementatin of these techniques, the sparsity of the linear equations (connecting
the mismatch and solution vectors) is exploited to reduce the computation time as well as memory requirement. From the next lecture, we will study some methods for solution of sparse linear
82
Table 2.40: Further results of the 30 bus system with a bipolar HVDC link between bus 9 and 28
Bus no.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
(p.u)
1.05000
1.03380
1.02180
1.01437
1.00580
1.00804
0.99286
1.02300
1.00248
0.99624
1.09130
1.03787
1.08830
1.01972
1.01042
1.01181
0.99507
0.99250
0.98530
0.98713
0.98445
0.98541
0.99404
0.98099
0.98192
0.96358
0.99144
1.00345
0.97089
0.95901
With combination 3
Pinj
(deg)
(p.u)
0.00000
2.42241
-5.04581 0.35860
-8.07176 -0.02400
-9.71785 -0.07600
-13.71987 -0.69640
-11.32909 0.00000
-13.96633 -0.62800
-12.20102 -0.45000
-20.07763 -1.00000
-19.56171 -0.05800
-18.12405 0.17930
-16.26936 -0.11200
-15.06838 0.16910
-17.44774 -0.06200
-17.72802 -0.08200
-17.89806 -0.03500
-19.27534 -0.09000
-19.10546 -0.03200
-19.73134 -0.09500
-19.74984 -0.02200
-19.76760 -0.17500
-19.66188 0.00000
-18.37832 -0.03200
-18.89414 -0.08700
-16.75594 0.00000
-17.20702 -0.03500
-15.15766 -0.00000
-9.34449 0.98449
-16.46954 -0.02400
-17.41321 -0.10600
Total iteration = 23
Qinj
(p.u)
-0.25233
0.09175
-0.01200
-0.01600
0.12191
0.00000
-0.10900
0.55554
-0.33808
-0.02000
0.46906
-0.07500
0.39377
-0.01600
-0.02500
-0.01800
-0.05800
-0.00900
-0.03400
-0.00700
-0.11200
-0.00000
-0.01600
-0.06700
0.00000
-0.02300
-0.00000
-0.38369
-0.00900
-0.01900
DC system solutions
Vdr = 253.938 kV; idr = 393.8 Amp.;
= 15.21o ; Qdr = 33.808 MVAR;
= 18.31o ; Pdi = 98.45 MW;
Qdi = 38.37 MVAR;
equations.
83
Module 3
Sparsity Technique
3.1
Sparse matrices
Sparse matrix is a matrix in which most (or, at least, significant number) of the elements are zero.
In the context of power system analysis, the matrices associated with power flow solution are sparse.
For example, let us consider the YBU S matrix. As we have already seen, the off-diagonal elements of
YBU S matrix signifies the connectivity between the nodes. To be more precise, the element (i,j) of
YBU S matrix is non-zero if there is a direct connection between node i and node j, while it is zero if
there is no direct connectivity between these two nodes. Now, in many power systems, generally any
bus is connected to mostly 3-4 buses directly. Therefore, in a 100 buses system (say), there would
be at best 4-5 non-zero terms (including the diagonal) in any row of the YBU S matrix, rest of the
elements being zero. Therefore, out of (100 100) = 10, 000 elements, only about 500 terms would
be non-zero and the other terms (elements) would be zero. Thus, in this case, the YBU S matrix is
almost 95 percent sparse. For any larger system, the percentage of sparsity of the associated YBU S
matrix would be even more.
Because of the sparsity of the YBU S matrix, the Jacobian matrix for load flow solution is also
sparse. To see that, please consider equations (2.48) - (2.55). From these equations it can be seen
that all the elements of the Jacobian matrix depend on the element Yij . Therefore, if this element
Yij is zero, the corresponding elements of the Jacobian matrix would also be zero. As most of the
elements (Yij ) of the YBU S matrix are zero, it immediately follows that most of the elements of the
Jacobian matrix would also be zero, thereby making the Jacobian matrix also quite sparse.
Now, in each iteration of the NRLF technique, (we are considering the polar form here), the
correction vector (X) is computed by inverting the Jacobian matrix and thereafter multiplying
the inverse of the Jacobian matrix with the mismatch vector (M ) (please see equation (2.45)).
However, even though the Jacobian matrix is sparse, its inverse is a full matrix. Hence, computation
of the direct inverse of the sparse matrix involves a lot of computational burden. Therefore, it would
be much less intensive if equation (2.45) can be solved exploiting the sparse nature of the Jacobian
matrix. Apart from this, storing all the elements of a highly sparse matrix also consumes the memory
unnecessarily. Therefore, if only the non-zero elements are stored in appropriate fashion, a lot of
memory can be freed. Of course, with the storage of only the non-zero elements, the complexity of
84
programming will increase. However, for any general purpose load flow program, which is expected
to handle any large size power system, enhancement in the complexity of programming is often a
small cost as compared to the advantage of optimized memory utilization.
Below we will discuss some schemes for solving a set of linear equations (note that equation (2.48)
is a set of linear equations) utilizing the sparse nature of the Jacobian matrix and also some schemes
for storing a sparse matrix. We will start with the Gaussian Elimination method for solving a set of
linear equations.
3.2
Ax = b
(3.1)
Where x and b are both (n1) vectors and A is a (nn) co-efficient matrix. The most obvious
method for solving equation (3.1) is to invert matrix A, that is x = A1 b. However, equation
(3.1) can also be solved indirectly by converting the matrix A into an upper triangular form with
appropriate changes reflected in the vector b and then by back substitution. To illustrate the basic
procedure, let us consider a 4th order system as shown in equations (3.2)-(3.5).
(3.2)
(3.3)
(3.4)
(3.5)
x1 +
a13
a14
b1
a12
x2 +
x3 +
x4 =
a11
a11
a11
a11
(3.6)
b) Multiply equation (3.6) by a21 , a31 , a41 (one by one) and subtract the resulting expression
from equations (3.3), (3.4) and (3.5) respectively to yield:
(a22
a12 a21
a13 a21
a14 a21
b1 a21
) x2 + (a23
) x3 + (a24
) x4 = b2
a11
a11
a11
a11
(3.7)
(a32
a12 a31
a13 a31
a14 a31
b1 a31
) x2 + (a33
) x3 + (a34
) x4 = b3
a11
a11
a11
a11
(3.8)
85
(a42
a13 a41
a14 a41
b1 a41
a12 a41
) x2 + (a43
) x3 + (a44
) x4 = b4
a11
a11
a11
a11
(3.9)
x1 +
a12
a13
a14
b1
x2 +
x3 +
x4 =
a11
a11
a11
a11
(3.10)
(1)
(1)
(1)
a(1)
22 x2 + a23 x3 + a24 x4 = b2
(3.11)
(1)
(1)
(1)
a(1)
32 x2 + a33 x3 + a34 x4 = b3
(3.12)
(1)
(1)
(1)
a(1)
42 x2 + a43 x3 + a44 x4 = b4
(3.13)
a(1)
jk = ajk
aj1 a1k
a11
for j, k = 2, 3, 4
(3.14)
x2 +
(1)
a23
(1)
a22
x3 +
a(1)
24
a(1)
22
(1)
x4 =
b2(1)
(3.15)
a(1)
22
(1)
b) Multiplying equation (3.15) by a32 and a42 (one by one) and subtracting the resulting
expressions from equations (3.12) and (3.13) respectively one can obtain;
(1)
33
[a
(1)
43
[a
(1)
a(1)
23 a32
(1)
22
(1)
a(1)
23 a42
(1)
22
(1)
34
] x3 + [a
(1)
44
] x3 + [a
(1) (1)
a24
a32
(1)
22
(1) (1)
a24
a42
(1)
22
] x4 = [b
(1)
3
] x4 = [b
(1)
4
b2(1)
(1)
22
b2(1)
(1)
22
a(1)
32 ]
(3.16)
a(1)
42 ]
(3.17)
x2 +
(1)
a23
(1)
a22
x3 +
a(1)
24
a(1)
22
x4 =
b2(1)
a(1)
22
(3.18)
(2)
(2)
a33
x3 + a34
x4 = b3(2)
(3.19)
(2)
(2)
a43
x3 + a44
x4 = b4(2)
(3.20)
(1)
jk
=a
(1) (1)
aj2
a2k
(1)
a22
for j, k = 3, 4
Step 3: In this step we will work with equations (3.19) and (3.20).
86
(3.21)
(2)
x3 +
(2)
a34
(2)
a33
x4 =
b3(2)
(3.22)
(2)
a33
(2)
b) Multiplying equation (3.22) by a43 and subtracting it from equation (3.20) one can obtain,
(2)
44
[a
(2)
a(2)
34 a43
(2)
33
] x4 = [b
(2)
4
b3(2)
(2)
33
(2)
]
a43
(3.23)
Equation (3.23) contains only one unknown, x4 . Therefore, the value of x4 can be calculated
from this equation. With the value of x4 thus calculated, x3 can be calculated from equation (3.22).
Going back in this manner, x2 can be calculated from equation (3.18) (with the known values of x3
and x4 ) and lastly, the value of x1 can be calculated from equation (3.10) (with the known values
of x2 , x3 and x4 ).
The steps described in equations (3.6)-(3.23) can easily be expressed in terms of standard matrix
operations. To see this, let us represent equations (3.2)-(3.5) in matrix notation as shown in equation
(3.24). In this equation, it is assumed that a11 0.
a11
a
21
a31
a41
a12
a22
a32
a42
a13
a23
a33
a43
a14 x1 b1
a24 x2 b2
=
a34 x3 b3
a44 x4 b4
(3.24)
Starting with this matrix, the various steps for Gaussian elimination are as follows.
Step M1
On equation (3.24), the operation R1/a11 (where R1 is the first row of the co-efficient matrix
of equation (3.24)) is carried out to obtain equation (3.6) and the resulting matrix equation is shown
in equation (3.25).
a
x b
a
a
a
21
2
22
23
24 2
a31
b3
a
a
a
x
32
33
34
3
a41
a
a
a
x
b
42
43
44
4
4
(3.25)
Step M2
On equation (3.25), the operations (R2 R1 a21 ), (R3 R1 a31 ) and (R4 R1 a41 ) are
carried out (where Ri denotes the ith (i = 1, 2, 3, 4) row of the co-efficient matrix of equation (3.25))
to obtain equations (3.10)-(3.13) and the resulting matrix equation is shown in equation (3.26). In
87
(1)
1 a /a a /a a /a x b /a
12
11
13
11
14
11 1
1 11
(1)
(1)
x2 b(1)
0 a(1)
a
a
2
22
23
24
=
(1)
(1)
(1)
0 a(1)
b
x
a
a
3 3
32
33
34
(1)
(1)
(1)
(1)
0 a42
a43
a44 x4 b4
(3.26)
Step M3
(1)
On equation (3.26), the operation R2/a22 is carried out (corresponding to equation (3.15)) to
obtain the resulting matrix equation shown in equation (3.27).
1 a /a
a13 /a11 a14 /a11 x1 b1 /a11
12
11
(1) (1)
(1)
(1)
(1)
(1)
1
a23 /a22 a24 /a22 x2 b2 /a22
= (1)
(1)
(1)
0 a(1)
x b
a
a
3 3
32
33
34
(1)
(1)
(1)
(1)
0 a42
a43
a44 x4 b4
(3.27)
Step M4
(1)
(1)
On equation (3.27), the operations (R3 R2 a32 ) and (R4 R2 a42 ) are carried out corresponding to the equations (3.18)-(3.21) and the resulting matrix equation is shown in equation
(2)
(3.28). In this equation, it is assumed that a33 0.
1 a /a
a13 /a11 a14 /a11 x1 b1 /a11
12
11
(1) (1)
(1)
(1)
(1)
(1)
0
1
a23 /a22 a24 /a22 x2 b2 /a22
= (2)
(2)
(2)
0
x b
0
a33
a34
3 3
(2)
(2)
(2)
0
0
a
a
x
b
4
43
4
44
(3.28)
Step M5
(2)
On equation (3.28), the operation R3/a33 is carried out to obtain the matrix equation shown
in equation (3.29).
1 a /a
a13 /a11 a14 /a11 x1 b1 /a11
12
11
(1) (1)
(1)
(1)
(1)
(1)
0
1
a23 /a22 a24 /a22 x2 b2 /a22
= (2) (2)
(2)
(2)
0
0
1
a34
/a33
x3 b3 /a33
(2)
(2)
(2)
0
0
a43
a44 x4 b4
(3.29)
Step M6
(2)
Lastly, on equation (3.29), the operation (R4 R3 a43 ) is carried out to obtain the matrix
88
1 a /a
a13 /a11 a14 /a11 x1 b1 /a11
12
11
(1) (1)
(1)
(1)
(1)
(1)
b
/a
x
1
a
/a
a
/a
22
23
22
24
22 2
2
=
(2)
(2)
b(2) /a(2)
0
x
0
1
a
/a
33
34
33 3
(3)
(3)
0
0
0
a44 x4 b4
(3)
(2)
(2)
a(2)
34 a43
(2)
33
(3)
and b4
= b4(2)
b(2)
3
(2)
33
(3.30)
(2)
a43
. From this equation, the
unknowns can be easily solved by back-substitution starting from the last row of the final co-efficient
matrix in equation (3.30). Thus, Gaussian elimination enables us to solve the unknown quantities
in a systematic manner without inverting the co-efficient matrix. Therefore, by adopting the same
procedure, the correction vector (M ) can be computed from equation (2.48) without having to
invert the Jacobian matrix. When a large power system in analyzed, adopting Gaussian elimination
reduces computational burden to a large extent (as compared to inversion of the Jacobian matrix).
(1)
(2)
In the above procedure, the variables a11 , a22 and a33 have been assumed to be non-zero. These
variables, by which the rows of the co-efficient matrix are divided, are called the pivot variables.
However, during the elimination process, it is not necessary that the pivot variables would be always
non-zero. If any pivot variable turns out to be zero at any intermediate step, then the corresponding
row is interchanged with the next row so that the new pivot variable is non-zero and the elimination
process can continue.
We will look into an example of Gaussian elimination procedure in the next lecture.
89
3.2.1
11
23
22
12
17
27
32
15
18
25
34
41
16 x1 10
28 x2 20
=
36 x3 30
36 x4 40
(3.31)
23
27
25
28 x2 20
22
32
34
36 x3 30
12
15
41
36 x4 40
(3.32)
Step M2
On equation (3.32), the operations (R2 R1 A(2, 1)), (R3 R1 A(3, 1)) and (R4 R1 A(4, 1))
are carried out (where A(2, 1) = 23, A(3, 1) = 22 and A(4, 1) = 12) and the resulting matrix equation is given by;
1 1.5455
1.6364
1.4545 x1 0.9091
x
2.0
2.0
4.0
10
3
(3.33)
Step M3
On equation (3.33), the operation R2/A(2, 2) (where, A(2, 2) = 8.5455) is carried out to get;
0
1
1.4787 0.6383 x2 0.1064
0
2.0
2.0
4.0 x3 10
(3.34)
Step M4
On equation (3.34), the operations (R3 R2 A(3, 2)) and (R4 R2 A(4, 2)) are carried
out to obtain (where A(3, 2) = 2.0 and A(4, 2) = 3.5455);
0
1
1.4787 0.6383 x2 0.1064
0
0
0.9574 5.2766 x3 10.2128
0
0
26.6065 20.8085 x4 29.4681
90
(3.35)
Step M5
On equation (3.35), the operation R3/A(3, 3) (where A(3, 3) = 0.9574) is carried out to obtain
the matrix equation shown below:
0
1
1.4787 0.6383 x2 0.1064
0
0
1
5.5114 x3 10.6672
0
0
26.6065 20.8085 x4 29.4681
(3.36)
Step M6
Lastly, on equation (3.36), the operation (R4 R3 A(4, 3)) (where A(4, 3) = 26.6065) is
carried out to get;
x 0.1064
0
1
1.4787
0.6383
0
10.6672
x
0
1
5.5114
3
x4 254.3484
0
0
0
125.83
(3.37)
In equation (3.37), the co-efficient matrix has been converted to an upper-triangular matrix.
From the last row of this equation, x4 can be calculated as x4 = 254.3484/125.83 = 2.0214. Back
substituting this value of x4 in the third row of equation (3.37) one can obtain x3 = 10.6672
5.5114 2.0214 = 0.4735. Similarly, substitution of the values of x3 and x4 in the second row
of equation (3.37) yields the of x2 as x2 = 0.1064 + 1.4787 0.4735 0.6383 2.0214 = 0.4837.
Lastly, substituion of x2 , x3 and x4 in the first row of equation (3.37) gives x1 = 0.9091 + 1.5455
0.4837 + 1.6364 0.4735 1.4545 2.0214 = 0.5086.
3.3
We have seen that the Gaussian Elimination method is quite effective for solving a large set of spare
linear equations without having to invert the co-efficient matrix. Moreover, at every stage, if the
calculations pertaining to Gaussian elimination is carried out only using the non-zero terms, great
saving in the computational burden can be achieved. However, if the elimination process is carried
out in the normal sequence, at any stage of elimination, the original zero-elements may the concerted
into a non-zero element. This is normally termed as fill-in phenomenon. On the other hand, instead
of following the normal sequence, if the elimination process is carried out in an appropriate order,
then the occurrence of fill-in can be avoided to a great extent. A simple example given below
illustrates this point.
In equation (3.38), an initial co-efficient matrix is shown at the left hand side (part a) and the
structure of the co-efficient matrix after step-1 is shown at the right hand side (part b). It is to be
noted that in this equation, only the positions of non-zero terms (denoted by ) and zero terms
(denoted by o) are shown. As can be seen in equation (3.38), after step-1, all the original zero
elements have been converted to non-zero terms (denoted by ), or, in other words, significant
91
1
2
3
4
o o
o o
o o
1
2
3
4
a) Initial A matrix
1
o
o
o
(3.38)
Now, if the original co-efficient matrix shown in part (a) of equation (3.38) is re-arranged as
shown in part (a) of equation (3.39), then after step 1, there would be no fill-in as can be observed
in part (b) of equation (3.39).
4
2
3
1
o o
o o
o o
a) Rearranged A matrix
4
2
3
1
1 o o
o o
o o
o
(3.39)
From the above example, it is apparent that if the rows are eliminated in an optimal order,
then the number of fill-in would be minimum. However, an ideal optimal order is very difficult to
develop and perhaps is impossible. As an alternative, various near optimal ordering schemes have
been developed. Some of them are discussed below:
Scheme 1
In this scheme, before elimination, number the rows of the co-efficient matrix A according to
the number of non-zero, off-diagonal terms. Thus, the rows with only one off-diagonal, non-zero
term are numbered first, those with two non-zero, off- diagonal terms are numbered second and
so on. However, this scheme does not take into account the changes occurring in the co-efficient
matrix during the elimination process. Therefore, this scheme in quite easy and straight forward to
implement.
Scheme 2
In this scheme the rows of the co-efficient matrix A are numbered such that at each step of the
elimination procedure, the row with the fewest number of non-zero off-diagonal terms would be
operated next. If more than one row meets this criterion, then any one row is chosen. Therefore,
this scheme requires the simulation of the elimination procedure to estimate the changes occurring
in the co-efficient matrix in advance. Thus, this method takes longer time as compared to scheme 1
to compute the solution, but is definitely better than scheme 1.
92
Scheme 3
In this scheme, the rows are numbered in such a way so that the row which will introduce fewest nonzero off-diagonal terms would be operated upon next. If more than one row satisfies this criterion,
choose any one row. Again, this scheme also requires the simulation of the elimination process to
study its effects on the co-efficient matrix in advance. Hence, this method also takes longer time
than scheme 1.
Let us now look at another technique for solving a set of linear equations without the need of
inverting the co-efficient matrix, namely, the triangular factorization or LU decomposition.
3.4
Triangular factorization:
a11
a
21
a31
a41
a12
a22
a32
a42
a13
a23
a33
a43
0
0 11 12 13
a14 11 0
0 0 22 23
a24 21 22 0
=
a34 31 32 33 0 0
0 33
a44 41 42 43 44 0
0
0
14
24
34
34
(3.40)
(3.41)
Ly = b
(3.42)
Or,
In equation (3.42), y = Ux. Expanding equation (3.42) we get,
11 0
0
0 y1 b1
0 y2 b2
21 22 0
=
31 32 33 0 y3 b3
41 42 43 44 y4 b4
(3.43)
b1
11
i1
1
=
[bi ij yj ] ;
ii
j=1
y1 =
yi
93
i = 2, 3, N
(3.44)
14 x1 y1
24 x2 y2
=
34 x3 y3
34 x4 y4
11 12 13
22
23
0
0 33
0
0
0
(3.45)
Now, With the knowledge of the intermediate vector y, from equation (3.45), the solution vector
x can be calculated as,
yn
N N
N
1
[yi ij xj ] ;
=
ii
j=1+1
xN =
xi
i = (N 1), (N 2), 1
(3.46)
We will now look into the basic procedure of obtaining the LU decomposition in the next lecturel.
94
3.4.1
Method of LU decomposition
To solve for the matrices L and U for given matrix A, let us write the i, j th element of equation
(3.40). In general, the element aij can be represented as (from equation (3.40)),
(3.47)
The number of terms in the sum of equation (3.47) depends on whether i or j is the smaller
number. We have in fact three distinct cases:
i < j;
i1 1j + i2 2j + + ii ij = aij
(3.48)
i = j;
i1 1j + i2 2j + + ii jj = aij
(3.49)
i > j;
i1 1j + i2 2j + + ij jj = aij
(3.50)
ii = 1;
i = 1, 2, N
(3.51)
With the condition of equation (3.51), an elegant procedure called Crouts algorithm quite
easily solves for the N 2 unknown and by just rearranging the equations in a certain order.
The algorithm is as follows:
step 1: Set ii = 1 for i = 1, 2, 3 N .
step 2: For each j = 1, 2, 3 N , perform the following operation:
a) For i = 1, 2, 3 N solve for ij as
i1
ij = aij ik kj
(3.52)
k=1
(3.53)
These first and second operations both need to be carried out before going to next value of
j.
95
3.4.2
Example of LU decomposition
3 2 7
3 4 1
3 2 7
0 11 12 13
11 0
= 21 22 0 0 22 23
Solution: Let A =
2
3
1
3 4 1 31 32 33 0
0
33
2
;
3
3
= 1;
3
For j = 2;
i = 1 12 = a12 = 2;
5
i = 2 22 = a22 21 12 = ;
3
6
1
(a32 31 12 ) = ;
i = 3 32 =
22
5
For j = 3;
i = 1 13 = a13 = 7;
i = 2 23 = a23 21 13 =
11
;
13
8
i = 3 33 = a33 31 13 32 23 = ;
5
3 2 7 1
0 0 3 2
7
3 4 1 1 6/5 1 0 0 8/5
3.5
There are several methods available in the literature for storing a sparse matrix. Some of these
schemes are described here.
96
Random packing:
In this method, every non-zero element of the matrix is stored in a primary array while its row and
column indices are stored in two secondary arrays. As each element in individually identified, the
elements can be stored in a random manner. For example, the sparse matrixes shown in equation
(3.54) are stored in one primary array and two secondary arrays as shown below.
0
0
0 0 0
A = 0
0 2.67 0 3.12
{0.29
{3
{2
3.12
2
5
(3.54)
-1.25 2.67
3
2
1
3
2.31 0}
3
0}
5
0}
In the above, the zero at the end of each array denotes the termination of the array.
Systematic packing:
In case the elements of a sparse matrix are read or constructed or sorted in a systematic order, then
there is no need to adopt both row and column indices for each element. Instead some alternative,
more efficient schemes can be adopted as described below.
a) The use of row address:
In this scheme, the index of first non-zero element in each row is specified in a separate integer
array. As an example, for the matrix A in equation (3.54) the elements can be represented as,
Real array (elements) = {2.67
Integer array JA
=
{3
Integer array IST
=
{1
3.12
5
1
-1.25
1
3
0.29
2
6
2.31}
5}
}
In this case, the array of row address IST has been constructed such that the number of non-zero
elements in row i is IST(I+1)-IST(I). Thus, for a matrix with m rows, the array IST will have
(m+1) entries. For example, from the array IST, the number of non-zero elements in 1st row of
the matrix A is IST(2) - IST(1) = 1 - 1 = 0, which is indeed true as observed from equation (3.54).
Similarly, the array IST indicates that the number of non-zero elements in 2nd row of the matrix A
is IST(3) - IST(2) = 3 - 1 = 2 and from the first two elements of the array JA, these two elements
are located at columns 3 and 5 of the matrix whereas the elements themselves are given by the first
two elements of the real array (2.67 and 3.12) (which is again true from equation (3.54)). In a similar
way, the elements and the locations of the elements in the third row of the matrix A can easily be
identified from the above three arrays. Moreover, please note that, as the number of rows in the
matrix A is 3, the total number of entries in array IST is 4.
97
= {2 2.67
= {0
3
3.12 3
5
0
-1.25
1
0.29
2
2.31 0}
5
0}
c) Compound identifiers:
In the random packing scheme it in possible to reduce the storage requirement by combining the
two indices for each element so that these can be held in one integer storage. A suitable compound
identifier could be (n i + j) where n is an integer higher than the number of columns in the matrix
and (i, j) denote the position of the non-zero element in the matrix. For example, the matrix A in
equation (3.54) could look like
Real array (elements)
Integer array JA
=
=
{2.67
{2003
3.12 -1.25
2005 3001
0.29
3002
2.31
3005
0}
9999}
In the above, n has been chosen to be equal to 1000. Also, in the real array, the entry zero
indicates the end of the array and the corresponding entry in the integer array is 9999 (to signifies
the end of the integer array).
However, unless compound identifiers yield necessary or highly desirable storage saving, it should
not be used because of the following reasons:
i) Extra programming would be required to interpret the identifiers correctly.
ii) It should not be used for matrices whose order is so large that integer register overflow
results.
d) The use of mixed arrays:
It is possible to use a single array to store both the non-zero elements of the matrix and the
identifiers. For example, the matrix A in equation (3.54) could be stored as,
Real array (B)
{-2
2.67
3.12
-3
-1.25
0.29
2.31
0}
In the above scheme, each non-zero element is preceded by its column number and each non-zero
row is preceded by the negative of the corresponding row number.
With this discussion, we are now at the and of our study of sparse linear systems. From the next
lecture, we will start the discussion of short circuit analysis methods.
98
Module 4
Short Circuit analysis
4.1
For a network with m buses and a reference bus, one can write a relation between bus currents
and bus voltages as
BUS ] [V
BUS ]
[IBUS ] = [Y
(4.1)
Where,
BUS ] [IBUS ]
BUS ] = [Z
[V
(4.2)
Where,
BUS ] = [Y
BUS ]
[Z
From equation (4.2) for the ith bus one can write
(4.3)
ij can be written as
From equation (4.3), Z
Vi
Zij =
Ij Ik = 0; k = 1, 2, m,
99
(4.4)
j
Vi
Zii =
Ii
(4.5)
Ik = 0; k = 1, 2, m, i
BUS matrix
Following points should be noted for the Z
ij is the off-diagonal element of Z
BUS matrix and is called the
Z
impedance between ith and j th bus.
ii is the diagonal element of Z
BUS matrix and is called the
Z
impedance of ith bus.
open-circuit transfer
b=n1
(4.6)
Links : The elements of a graph not included in the tree of the graph are called links. Each
link is associated with a loop. If e is the number of elements in a graph, then the number of
links ` is given by
` = eb = en+1
(4.7)
The above definitions are explained with the help of illustrations as shown below :
Fig. 4.1 is a single line diagram of a power system. It has 4 buses, bus(1) to bus(4) and six
elements element e1 to element e6 . In this figure, bus(0) is taken as the reference bus.
Fig. 4.2 shows the graph of the network depicting the interconnection of the elements and the
reference node.
100
BUS matrix for a partial network with m buses and a reference bus 0, as
us assume that the Z
shown in Fig. 4.4, exists.
The bus voltages and bus currents for the partial network satisfy the relation
m
BUS
m
m
[V
] = [Z
BUS ] [IBUS ]
(4.8)
Where,
m
BUS
V
is m 1 bus voltage vector
Im
BUS is m 1 bus current injection vector
m
Z
BUS is m m bus impedence matrix of the partial network
BUS , one element at a time is added to the partial network, till all the elements are
To build Z
added to the network. The added element may be a branch or a link and hence the four possible
element additions to a partial network are:
a. Addition of a branch between a new node and the reference
b. Addition of a branch between a new node and an existing node
c. Addition of a link between an existing node and the reference
d. Addition of a link between two existing nodes
Let us now discuss these four cases one-by-one in detail.
4.1.1
Fig. 4.5 shows the addition of a branch between a new node q and the reference 0.The addition of
BUS to (m + 1) (m + 1) with the addition
a new node to the partial network increases the size of Z
of a new row and a new column corresponding to the new node q, Let the impedance of this branch
be zq0 . The new network equation can be written as:
V1 Z11
V2 Z21
V Z
p p1
=
Vm Zm1
Vq Zq1
Z1q I1
Z2q I2
Zpq Ip
Zmq Im
Zqq Iq
(4.9)
Figure 4.5: Addition of a branch between a new node and the reference
Vq
Zqq =
Iq Ik = 0; k = 1, 2, m
a current source of Iq = 1 p.u is connected to the q th bus, with all the others buses open, and the
voltage of q th bus (Vq ) is computed, as shown in Fig. 4.6.
From Fig. 4.6 one gets Vq = zq0 Iq , and thus with Iq = 1 p.u.
Vq
Zqq =
= zqo
Iq Ik = 0; k = 1, 2, m,
qi , a current source Ii = 1 p.u. is connected between ith bus and the reference
For finding out Z
bus with all other buses open circuited as shown in Fig. 4.7.
From Fig. 4.7 , Vq = 0, and hence with Ii = 1 p.u.
Vq
Zqi =
Ii Ik = 0; k = 1, 2, m,
=0
qq for Case 1
Figure 4.6: Calculation of Z
Z11
Z21
BUS = Zp1
Z
Zm1
4.1.2
zqo
0
0
(4.10)
Let a branch with impedance zpq be connected between an existing node p and a new node q as
Bus matrix increases by one to (m+1)(m+1) due
shown in Fig. 4.8. In this case also, the size of Z
to the addition of a new node q to the network. The modified network equations can be written
105
qi for case 1
Figure 4.7: Calculation of Z
as:
V1
V2
V
p
=
Vm
Vq
Z11
Z21
Zp1
Zm1
Zq1
I1
I2
Zpq Ip
Zmq Im
Zqq Iq
Z1q
Z2q
(4.11)
m
Even after the addition of branch p-q, the original matrix Z
Bus remains unchanged. Only the
additional elements corresponding to the q th row and column need to be calculated.
qq one can write
For calculating Z
Vq
Zqq =
Iq Ik = 0; k = 1, 2, m
qq current source of Iq = 1 p.u is connected to the q th bus, with all the others buses
To evaluate Z
open circuited, and the voltage of q th bus Vq is computed, as shown in Fig. 4.9. From Fig. 4.9 with
106
Figure 4.8: Addition of a branch between an existing node p and a new node q
V2 = Z2q Iq = Z2q
Vp = Zpq Iq = Zpq
Vm = Zmq Iq = Zmq
Vq = Zqq Iq = Zqq
(4.12)
(4.13)
pq and Vq = Zqq .
Because, from the Fig. 4.10 , ipq = Iq = 1 pu and from equation (4.12) Vp = Z
Thus,
Zqq = Zpq + zpq
(4.14)
qi one can write
For calculating Z
Vq
Zqi =
Ii
Ik = 0; k = 1, 2, m, i
qi a current source of I1 = 1 p.u is connected to the ith bus, with all the others
Hence, to compute Z
107
qq
Figure 4.9: Calculation of Z
V1 = Z1i Ii = Z1i
V2 = Z2i Ii = Z2i
Vp = Zpi Ii = Zpi
Vm = Zmi Ii = Zmi
Vq = Zqi Ii = Zqi
108
qi for case 2
Figure 4.11: Calculation of Z
From Fig. 4.11, Vq = Vp as the current in the branch p q is zero. Hence, from the above equations
one gets
(4.15)
Zqi = Zpi ; i = 1, 2, m
Z11
Z21
BUS = Zp1
Z
Zm1
Zp1
Zpp
Zmp
Zpq + zqp
Z1p
Z2p
(4.16)
So far in this lecture, we have considered the cases of addition of branches only. In the next lecture
we will consider the case of addition of links.
109
4.1.3
When an element is connected between an existing node and the reference, it creates a loop and
thus, the addition of this element is equivalent to the addition of a link. This will not generate any
Bus matrix remains unchanged. However, all the elements are
new node and the size of modified Z
modified and need to be recalculated. Let the added element,with an impedance of zqo , be connected
between an existing node q and the reference node 0 as shown in the Fig. 4.12. I` is the current
Figure 4.12: Addition of a link between an existing node q and the reference
through the link as shown in the Fig. 4.12. This current modifies the current injected into q th bus
from Iq to Iq I` . The modified network equations can be written as,
(4.17)
Vq = zqo I`
(4.18)
Also
Substuting Vq from the equation (4.17) into the equation (4.18) one can write
or
(4.19)
Equations equation (4.17) and equation (4.19) together form the set of (m + 1) simultaneous
network equations which can be expressed in matrix form as:
V1
V2
V
q
=
Vm
Z11 Z12
Z21 Z22
Zq1 Zq2
Zm1 Zm2
Zq1 Zq2
Z1q
Z2q
Z1m
Z2m
Zqq
Zqm
Zmq Zmm
Zqq Zqm
Z1q
Z2q
Zqq
Zmq
Zqq + zqo
I1
I2
I
q
Im
I`
(4.20)
Bus matrix
The link current I` has to be eliminated and hence, the last row and column of modified Z
have to be eliminated. The partitioned matrix relation of equation (4.20) can be written in compact
form as:
m
(m)
[VBus ]
(m)
m
Z
Bus
=
[
T
(1)
[Z]
m
IBus
[Z]
Z``
I`
(1)
(4.21)
where,
T [Im
0 = [Z]
Bus ] + Z`` I`
or,
T [Im ]
[Z]
Bus
I` =
Z``
(4.22)
[V
] = [Z
Bus ][IBus ] + [Z]I`
(4.23)
T
[Z][
Z]
m
Bus
m
] [Im
[V
] = [[Z
]
Bus
Bus ]
Z``
111
(4.24)
Hence,
m
Bus
Bus ][Im
[V
] = [Z
Bus ]
(4.25)
T
[Z][
Z]
Bus ] = [[Z
m
[Z
]
]
Bus
Z``
(4.26)
where
4.1.4
Let an element with impedance zpq be connected between two existing nodes p and q. This is an
addition of a link as it forms a loop encompassing nodes p and q as shown in Fig. 4.13. Let I`
be the current through the link as shown in the Fig. 4.13. This link current changes the injected
current at pth node from Ip to (Ip I` ) , while the injected current at node q th is modified from Iq
to (Iq + I` ). The modified network equations can be written as:
112
V1 = Z11 I1 + Z12 I2 + + Z1p (Ip Il ) + + Z1q (Iq + I` ) + + Z1m Im
(4.27)
Also from Fig. 4.13, the relation between Vp and Vq in terms of I` and zpq can be written as
Vp Vq = zpq I`
(4.28)
0 = Vp + Vq + zpq I`
(4.29)
or
Substituting Vp and Vq from the equation (4.27) into the equation (4.29) in the following relation
is obtained:
0 =(Zq1 Zp1 )I1 + (Zq2 Zp2 )I2 + + (Zqp Zpp )Ip + (Zqq Zpq )Iq + + (Zqm Zpm )Im
+ (Zpp + Zqq 2Zpq + zpq )I`
(4.30)
Equations (4.27) and (4.30) form a set of (m + 1) simultaneous equation which can be written in
matrix form as:
V1
V2
V
q
=
Vm
Z11
Z21
Zq1
Zm1
Z`1
Where,
I1
I2
Zq` Iq
Zm` Im
Z`` I`
Z1`
Z2`
(4.31)
For eliminating the link current I` equation (4.31) can be written in the compact form as:
m
(m)
VBus
(m)
m
Z
[ BusT
(1)
[Z]
m
IBus
[Z]
Z``
I`
(1)
(4.32)
where,
(1)
= (1) [ zp0 ]
Step 3: Pick up another element from the graph. It should either be connected to an existing node
or the reference node. Never select an element connected to two new nodes as it will be isolated from
Bus ] matrix becoming infinite.
the existing partial network and this will result in the elements of [Z
Bus ] matrix building process, if
For instance, with reference to Fig. 4.14, in the next step of [Z
114
element 5 is next added to the partial network as shown in Fig. 4.15, the resultant network is
disjointed. This is an incorrect choice. The proper choice could be any one of the elements 2 or 3 or
4 or 6.
m
If the bus impedance matrix of a partial network with m-nodes, [Z
Bus ], is known, then depending
on whether the added (m + 1)th element is a branch or a link, the following steps are to be followed
Bus ] matrix:
to obtain the new [Z
(a) If the added element is a branch between a new node q and the reference node with an impedance
Bus ] matrix will increase by one and the new matrix is given as :
zqo , then the size of the new [Z
(1)
Bus =
Z
(m)
(q)
(1)
()
(m)
(q)
Z11 Z1m 0
Zm1 Zmm 0
0
0
0 zq0
(b) If the added element is a branch between an existing node p and a new node q with an
impedance zpq , then a new row and column corresponding to the new node q is added to the
m
existing [Z
Bus ] matrix. The new matrix is calculated as follows:
(1)
(p)
Bus =
Z
(m)
(q)
(1)
(p)
(m)
(q)
Zpp + zpq
Z1p
Zpp
Zmp
(c) If the added element is a link between an existing node q and the reference node with an
impedance zqo , then no new node is added to the network. A two-step procedure has to be
followed to find the new bus impedance matrix.
m
In the first step, a column and a row will be temporarily added to existing [Z
Bus ] matrix as:
(1)
(temp)
Z
=
Bus
(q)
(m)
(`)
(1)
Z11
Zq1
Zm1
Zq1
(q)
Z1q
Zqq
Zmq
Zqq
(m)
(`)
Z1q
Zqq
Zmq
Z``
Z1m
Zqm
Zmm
Zqm
where,
Bus ] = [[Z
m
[Z
Bus ]
T
[Z][
Z]
]]
Z``
where,
(1)
(p)
(q)
(m)
(`)
Z11
Zp1
Zq1
Zm1
(Zq1 Zp1 )
(p)
(q)
(m)
Z1p
Z1q
Z1m
Zpp
Zpq
Zpm
Zqp
Zqq
Zqm
Zmp
Zmq
Zmm
(Zqp Zpp ) (Zqq Zpq ) (Zqm Zpm )
where,
(`)
(Z1q Z1p )
(Zpq Zpp )
(Zqq Zqp )
(Zmq Zmp )
Z``
Next eliminate the added row and column ` using the expression:
Bus ] = [[Z
m
[Z
Bus ]
T
[Z][
Z]
]
Z``
where,
117
4.2
The single line diagram of a power system is shown in the Fig. 4.16. The line impedances in pu are
Bus ] matrix formulation is explained as given below:
also given. The step-by-step procedure for [Z
Figure 4.16: Single Line Diagram of the Power System for the example
Preliminary Step: The graph of the network and a tree is shown in Fig. 4.17. Elements 1,2,4
and 5 are the tree branches while 3, 6 and 7 are the links.
Bus ] matrix is
The resulting [Z
(1)
(1)
(2)
(1)
(2)
0
Bus = (1) [ j0.10 0 ] = (1) [ j0.1
Z
]
(2)
0
z20
(2)
0
j0.10
Step 3: Element 3 connected between existing nodes, node 1 (p = 1) and node 2 (q = 2),
having an impedance of z12 = j0.20 p.u. is added to the partial network, as shown in Fig. 4.20.
Since this is an addition of a link to the network a two step procedure is to be followed. In the
119
first step a new row and column is added to the matrix as given below :
(1)
(1)
(temp)
Z
= (2)
Bus
(`)
(`)
(2)
j0.10
0.0
(Z12 Z11 )
0.0
j0.10
(Z22 Z21 )
(Z21 Z11 ) (Z22 Z12 )
Z``
(1)
= (2)
(`)
(1)
(2)
(`)
where,
Z`` = Z11 + Z22 2Z12 + z20 = j0.10 + j0.10 0.0 + j0.20 = j0.40 p.u.
Bus ] matrix as given below:
Next this new row and column is eliminated to restore the size of [Z
j0.10
[
] [j0.10 j0.10]
j0.10
j0.40
Hence, the impedance matrix after the addition of element 3 is found out to be :
(1)
Bus ] =
[Z
(1)
(2)
(2)
j0.075 j0.025
]
j0.025 j0.075
Step 4: The element 4 , which is added next, is connected between an existing node, node 2
(p = 2) and a new node, node 3 (q = 3). The impedance of this element is z23 = j0.30 p.u. and
it is a tree branch hence, a new node, node 3 is added to the partial network. This addition, shown
Bus ] to (3 3).
in Fig. 4.21, thus increases the size of [Z
120
(1)
Bus = (2)
Z
(3)
(2)
(3)
j0.075 j0.025
Z12
j0.025 j0.0.075
Z22
Z21
Z22
Z22 + z23
(1)
= (2)
(3)
(1)
(2)
(3)
Step 5: Element 5 is added next to the existing partial network. This is a tree branch connected
between an existing node, node 3 (p = 3) and a new node, node 4 (q = 4). This is illustrated
in Fig. 4.22.
(1)
Bus = (2)
Z
(3)
(4)
(2)
(3)
(4)
Z13
Z23
Z33 Z33 + z34
(1)
(2)
=
(3)
(4)
(1)
(2)
(3)
(4)
j0.075
j0.025
j0.025
j0.025
j0.025
j0.075
j0.075
j0.075
j0.025
j0.075
j0.375
j0.375
j0.025
j0.075
j0.375
j0.525
Step 6: Next,the element 6 connected between two existing nodes node 1 (p = 1) and node
4 (q = 4) is added to the network, as shown in the Fig. 4.23. The impedance of this element
is z23 = j0.25 p.u. As this is a link addition, the two step procedure is used. The bus impedance
121
(1)
(temp)
Bus
(2)
= (3)
(4)
(`)
(2)
(3)
(4)
j0.075
j0.025
j0.025
j0.025
j0.025
j0.075
j0.075
j0.075
j0.025
j0.075
j0.375
j0.375
j0.025
j0.075
j0.375
j0.525
(Z41 Z11 ) (Z42 Z12 ) (Z43 Z13 ) (Z44 Z14 )
122
(`)
(Z14 Z11 )
(Z24 Z21 )
(Z34 Z31 )
(Z44 Z41 )
Z``
(1)
(temp)
Bus
(2)
= (3)
(4)
(`)
(1)
(2)
(3)
(4)
j0.075
j0.025
j0.25
j0.025
j0.05
j0.025
j0.075
j0.075
j0.075
0.05
j0.025
j0.075
j0.375
j0.375
j0.35
(`)
j0.025 j0.05
j0.075 j0.05
j0.375 j0.35
j0.525 j0.50
j0.50 j0.80
where,
Z`` = Z44 + Z11 2Z14 + z14 = j0.075 + j0.525 2 j0.025 + j0.25 = j0.80 p.u.
The additional row and column ` are to be eliminated to restore the impedance matrix size to
Bus ] matrix after the addition of element 6 is calculated as:
(m m), and the [Z
j0.075
Bus ] = j0.025
[Z
j0.25
j0.025
j0.025
j0.075
j0.075
j0.075
j0.025
j0.075
j0.375
j0.375
j0.05
j0.50
j0.50
j0.075
j0.80
j0.375
j0.525
Hence,
(1)
Bus = (2)
Z
(3)
(4)
(1)
(2)
(3)
(4)
j0.0719
j0.0281
j0.0469
j0.0563
j0.0281
j0.0719
j0.0531
j0.0437
j0.0469
j0.0531
j0.2219
j0.1562
j0.0563
j0.0437
j0.1562
j0.2125
Step 7: Finally the element 7 connected between two existing nodes node 2 (p = 2) and node 4
(q = 4) is added to the partial network of step 6. The impedance of this element is is z23 = j0.40
pu. This is also a link addition, as shown in Fig. 4.24 and hence the two step precedure will be
Bus ] matrix. In the first step the Z
(temp)
is calculated after a row and a
followed to obtain the [Z
Bus
123
Bus as follows:
column are added to the exiting Z
(1)
(1)
(2)
(temp)
Z
= (3)
Bus
(4)
(`)
(2)
(3)
(4)
j0.0719
j0.0281
j0.0469
j0.0563
j0.0281
j0.0719
j0.0531
j0.0437
j0.0469
j0.0531
j0.2219
j0.1562
j0.0563
j0.0437
j0.1562
j0.2125
(Z41 Z21 ) (Z42 Z22 ) (Z43 Z23 ) (Z44 Z24 )
(`)
(Z14 Z12 )
(Z24 Z22 )
(Z34 Z32 )
(Z44 Z42 )
Z``
(1)
(temp)
Bus
(2)
= (3)
(4)
(`)
(1)
(2)
(3)
j0.0719
j0.02810
j0.0469
j0.0563
j0.281
j0.0281
j0.0719
j0.0531
j0.0437
j0.281
j0.0469
j0.0531
j0.2219
j0.1562
j0.1031
(4)
(`)
j0.0563 j0.281
j0.0437 j0.281
j0.1562 j1031
j0.2125 j0.1688
j0.1688 j0.5969
where,
Z`` = Z22 + Z44 2Z24 + z24 = j0.0719 + j0.2125 2 j0.0563 + j0.40 = j0.5969 p.u.
The additional row and column ` are to be eliminated to restore the impedance matrix size to
124
j0.0719
Bus ] = j0.0281
[Z
j0.469
j0.0563
j0.0281
j0.0719
j0.0531
j0.0437
j0.0469
j0.0531
j0.2219
j0.1562
j0.0281
j0.0281
j0.1688
j0.0437
j0.5969
j0.1562
j0.2125
Hence,
(1)
Bus = (2)
Z
(3)
(4)
(1)
(2)
(3)
(4)
j0.0705
j0.0295
j0.0420
j0.0483
j0.0295
j0.0705
j0.0580
j0.0517
j0.0420
j0.0580
j0.2041
j0.1271
j0.0483
j0.0517
j0.1271
j0.1648
4.2.1
Bus ] :
Modifications in the existing [Z
1
result
24
1
1
1
+ add =
=0
z
z24
org
24
or
org
add
z24
=
z24
= j0.40 p.u.
org
add
Hence, by adding an element z24
= j0.4 p.u. in parallel to z24
the removal of line between nodes
2 and 4 can be simulated. The new added fictitious element is a link addition between the two nodes,
p = 2 and q = 4 and is shown in Fig. 4.25 . Hence, this will require a two-step procedure. The
addition of the fictitious element 8 , which is a link, will introduce a temporary row and column.
125
(temp)
The Z
is given as:
Bus
(1)
(1)
(temp)
Bus
(2)
= (3)
(4)
(`)
(3)
(2)
(`)
(4)
j0.0705
j0.0295
j0.0420
j0.0483
j0.0295
j0.0705
j0.0580
j0.0517
j0.0420
j0.0580
j0.2041
j0.1271
j0.0483
j0.0517
j0.1271
j0.1648
(Z14 Z12 )
(Z24 Z22 )
(Z34 Z32 )
(Z44 Z42 )
Z``
(1)
(2)
(temp)
Z
= (3)
Bus
(4)
(`)
(1)
(2)
(3)
j0.0705 j0.0295
j0.0295 j0.0705
j0.0420 j0.0580
j0.0483 j0.0517
j0.0188 j0.0188
j0.0420
j0.0580
j0.2041
j0.1271
j0.0691
(4)
(`)
j0.0483 j0.0188
j0.0517 j0.0188
j0.1271 j0.0691
j0.1648 j0.1131
j0.1131 j0.2681
where,
add
Z`` = Z22 + Z44 2Z24 + z24
= j0.0705 + j0.1648 2 j0.0483 + (j0.40) = j0.2681 p.u.
126
j0.0705
Bus ] = j0.0295
[Z
j0.0420
j0.0483
j0.0295
j0.0705
j0.0580
j0.0517
j0.0420
j0.0580
j0.2041
j0.1271
j0.0188
j0.0188
j0.1131
j0.0517
j0.2681
j0.1271
j0.1648
(1)
Bus = (2)
Z
(3)
(4)
(1)
(2)
(3)
(4)
j0.0719
j0.0281
j0.0469
j0.0563
j0.0281
j0.0719
j0.0531
j0.0437
j0.0469
j0.0531
j0.2219
j0.1562
j0.0563
j0.0437
j0.1562
j0.2125
127
4.3
m
Assume that the bus impedance matrix [Z
BUS ] is known for a partial network of m nodes and
a reference node. The bus voltage and bus current relation for the partial network, shown in Fig.
4.26, can be expressed as:
BUS ] = [Z
m
[V
BUS ] [IBUS ]
(4.33)
In equation (4.33),
4.3.1
V1 Z11
V2 Z21
Vp = Zp1
Vm Zm1
Vq Zq1
Z1q I1
Z2q I2
Zpq Ip
Zmq Im
Zqq Iq
(4.34)
qi = Ziq for i =
The network is assumed to contain bilateral passive elements and hence, Z
1, 2, m, i q . The added branch p-q is assumed to be mutually coupled with one or more
elements of the partial network.
qi , inject a current at ith node and calculate the voltage at q th node with
To determine element Z
respect to reference, as shown in Fig. 4.27.
qi
Calculation of Z
As all other bus currents are zero, bus voltages can be written as,
129
V1 = Z1i Ii
V2 = Z2i Ii
Vp = Zpi Ii
Vm = Zmi Ii
Vq = Zqi Ii
(4.35)
Vq = Vp vpq
(4.36)
Where vpq is the voltage across the added element p q . Also, the currents in the elements of
the network can be related to the voltages across the elements as,
ipq
ypq,pq y
pq, vpq
][ ]
[ ] = [
y
,pq y
, v
i
(4.37)
Where,
j0.60
0
0
0
0
0
j0.50
0
0
0
[
z] = 0
0
j0.50
0
0
0
0
0
j0.25 j1.0
0
0
j1.0
j0.20
130
The inverse of [
z] is [
y], the primitive admittance matrix.
j1.67
0
0
0
0
0
j2.0
0
0
0
[
y] = 0
0
j2.0
0
0
0
0
0
j5.0 j2.5
0
0
j2.5
j6.25
The current ipq in the added branch p q equal to zero as node q is open.
ipq = 0
(4.38)
The voltage vpq , however, is not zero as the added branch is mutually coupled to one or more elements
of the partial network. Thus, the voltage across other elements of the network can be expressed as,
v
= V
(4.39)
and V
are the voltages of the nodes of the partial network. With ipq = 0 from equation
where V
(4.37) one can write,
ypq,pq vpq + y
pq, v
= 0
Hence,
vpq =
y
pq, v
ypq,pq
131
(4.40)
Substituting v
from equation (4.39) and vpq equation (4.36) in equation (4.40) one gets,
V
)
y
pq, (V
Vq = Vp +
ypq,pq
Substitution of Ii = 1 pu in equation (4.35) results in Vp , Vq , V and V being replaced by their
corresponding impedances and hence,
i Z
i )
y
pq, (Z
Zqi = Zpi +
ypq,pq
(4.41)
i = 1, 2, , m, i q
qq , a current Iq = 1 p.u. is injected into q th node with all
For calculating the self impedance Z
other currents equal to zero as shown in Fig. 4.29. Then the voltages of the nodes are calculated
from equation (4.35) , as
V1 = Z1q Iq
V2 = Z2q Iq
Vp = Zpq Iq
Vm = Zmq Iq
Vq = Zqq Iq
(4.42)
Vq = Vp vpq
(4.43)
ipq = Iq = 1
(4.44)
(4.45)
and
vpq =
1+y
pq, v
ypq,pq
(4.46)
V
)
1+y
pq, (V
Vq = Vp +
ypq,pq
132
(4.47)
and V
can be replaced by respective transfer
With Iq = 1 p.u., from equation (4.35) Vp , Vq , V
impedances,
q Z
q )
1+y
pq, (Z
Zqq = Zpq +
ypq,pq
4.3.2
(4.48)
If the added element p q is a link, then a fictitious node ` is created by connecting an ideal voltage
in series with the added element, as shown in Fig. 4.30.
The value of the source voltage is selected such that the current I` through the added link is
zero. If e` is the voltage of node ` with respect to node q and I` is the current injected into node `
from node q . The performance equation of the partial network with the added link p ` and ideal
series voltage source e` is,
V Z
1 11
Vp Zp1
=
Vm Zm1
e` Z`1
. . . Z1p . . . Z1m
. . . Zpp . . . Zpm
. . . Zmp . . . Zmm
. . . Z`p . . . Z`m
133
Z1` I1
Zp` Ip
Zm` Im
Z`` I`
(4.49)
e` = V` Vq ,
`i , 1 = 1, 2, m, i `, of the added row and column, can be calculated by
The elements Z
injecting a current Ii into the ith node and determining the voltage of `th node with repect to q th
node.
Hence,
e`
Z`i = , Ik = 0, k = 1, 2, m, k `
Ii
Also,
e` = Vp Vq vp`
(4.50)
(4.51)
vp` =
y
p`, v
yp`,p`
134
(4.52)
y
p`, = y
pq,
and
yp`,p` = ypq,pq
So,
vp` =
V
)
y
pq, (V
y
pq, v
=
ypq,pq
ypq,pq
(4.53)
(4.54)
With Ii = 1 p.u., substituting Vp , Vq , Vp and V from equation (4.35) and vp` from equation
(4.52) in equation (4.50) one gets
pi Z
i )
y
pq, (Z
Z`i = Zpi Zqi +
ypq,pq
(4.55)
i = 1, 2, m, i `
`` , a current is injected at the `th node with respect to node q , as shown in Fig.
To calculate Z
4.31. As all other node currents are zero, the node voltages can be written as,
Vk = Zk` I` , k = 1, 2, m
135
e` = Z`` I`
(4.56)
(4.57)
vp` =
1+y
pq,p v
p
ypq,pq
(4.58)
1 + ypq,p (Vp V )
e` = Vp Vq +
ypq,pq
, V
and e` from equation (4.56), Z`` is obtained as
With I` = 1 p.u., substituting Vp , Vq , V
p` Z
` )
1+y
pq, (Z
Z`` = Zpi Zqi +
ypq,pq
(4.59)
i = 1, 2, m, i `
In the case of link addition the additional row and column corresponding to fictitious node ` are to
be eliminated.
For this the fictitious series voltage source e` is short circuited. From equation (4.49) the bus
voltages can be written in compact from as
m
m
Bus
m
] = [Z
[V
Bus ][IBus ] + [Z][I` ]
(4.60)
Where,
= [Z1` Z2` Zm` ]T is an (m 1) vector comprising of the entries of the column added
[Z]
m
to the Z
Bus matrix
m
[Im
Bus ], [VBus ] = (m 1) bus current and voltage vectors respectively, of the partial network before
[Z
Bus ] = (m m) [ZBus ] matrix of the partial network before the addition of element p `.
I` = current injected in the link. Also,
T [IBus ] + Z`` I` = 0
[
e` ] = [Z]
136
(4.61)
Z.Z
m
][IBus ]
[VBus ] = [ZBus ] [
Z``
(4.62)
Z.Z
m
Final
[Z
]
=
[
Z
]
[
]
Bus
Bus
Z``
137
(4.63)
4.4
Bus ] example
Figure 4.32: The power system for [Z
A tree for the network is shown in Fig. 4.32. The system data is given in Table 4.1.
138
Mutual
Bus code Impedance
r-s
zpq,rs (p.u.)
0 - 1(2)
j0.2
0 - 1(1)
j0.1
(1)
(1)
(1)
(temp)
Z
=
Bus
(`)
(`)
j0.4 Z1`
[
]
Z`1 Z``
[
z] =
01(1)
01(2)
01(2)
j0.4 j0.2
]
j0.2 j0.5
impedance matrix [
z] and is given by
01(1)
[
y] = [
z]1 =
01(1)
01(2)
01(2)
j3.125 j1.25
]
j1.25 j2.5
matrix is given as
(1)
(`)
(1)
j0.4 j0.2
(temp)
Z
=
[
]
Bus
(`)
j0.2 j0.5
The row and column corresponding to the `th row and column corresponding to a link addition,(shown in red in the above matrix), need to be eliminated as the link addition does not create
Bus ] matrix, after the addition of second element to the partial network, is
a new node. The [Z
calculated using the following expression
Bus ] = [Z
Bus ] Z1` Z`1
[Z
Z``
= j0.4
(j0.2)(j0.2)
j0.5
(1)
140
01(1)
01(1)
[
z] = 01(2)
02
The primitive [
y] matrix is calculated as [
z]
[
y] = 01(2)
02
02
and is equal to
01(1)
01(1)
01(2)
j0.2 j0.5 0
j0.1
0
j0.5
01(2)
02
j3.333 j1.333
j0.667
Bus ] =
[Z
(1)
(2)
j0.32
Z21
(2)
Z12
]
Z22
is [0 1(1) 0 1(2)]
Z21 = Z01 +
Z02 Z11
[
]
y02,01(1) y02,01(2) ] [
Z01 Z11
y02,02
Z01 and Z02 are the transfer impedances associated with the reference node and are equal to zero.
Z21 =
j0.32
[j0.667 j0.2667] [
]
j0.32
j2.133
141
= j0.06
12 = Z21 = j0.06
Hence, Z
Z22 = Z02 +
Z02 Z12
1 + [
]
y02,01(1) y02,01(2) ] [
Z01 Z12
y02,02
1 + [j0.667 j0.2667] [
Z21 =
j0.32
]
j0.32
j2.133
= j0.48
Bus ] matrix is
The modified [Z
(1)
Bus ] =
[Z
(1)
(2)
(2)
j0.32 j0.06
]
j0.06 j0.48
Step 4: On adding element 4 between p = 2 and q = 3, a new node, node 3 is created. Hence, this
Bus ] matrix can be written as
is a branch addition and is shown in Fig. 4.37. The modified [Z
(1)
(1)
Bus = (2)
Z
(3)
(2)
(3)
j0.32 j0.06 Z
13
Z31
Z32 Z33
As this element is not mutually coupled to other elements the elements of vector y
pq, are zero.
Bus ] matrix can be calculated, using the expression given in (4.41),
Hence, the new elements of [Z
as :
Off-diagonal elements
Zqi = Zpi i = 1, 2, 3 i q
142
(2)
(3)
(1)
Bus = (2)
Z
(3)
Step 5: Finally add element 5 between nodes p = 1 and q = 3. This is an addition of a link hence a
temporary row and column are added. Fig. 4.38 showns the final network after the addition of this
(temp)
element. The modified Z
matrix can be written as
Bus
Figure 4.38: The complete network after the addition of link in step 5
(1)
(2)
(temp)
Z
=
Bus
(3)
(`)
(1)
(2)
(3)
(`)
Z`1
Z`2
Z`3 Z``
143
(temp)
Since this element is not mutually coupled to other elements, the new elements of [Z
]
Bus
matrix can be calculated, using the expression of (4.55), as :
Off-diagonal elements
Z`i = Zpi Zqi i = 1, 2, 3
Z1` = Z11 Z13 = j0.32 j0.06 = j0.26 = Z`1
Z2` = Z21 Z23 = j0.06 j0.48 = j0.42 = Z`2
Z3` = Z31 Z33 = j0.06 j0.88 = j0.82 = Z`3
Diagonal element
For calculating the diagonal element, the expression given in (4.59) is used. Hence,
(temp)
(1)
(temp)
[Z
]=
Bus
(2)
(3)
(`)
(1)
(2)
(3)
(`)
Z
Z``
T = [j0.26 j0.42 j0.82]
Z
j0.23
j0.82
144
Bus ] is
Hence, the final matrix [Z
(1)
(1)
Bus ] = (2)
[Z
(3)
(2)
(3)
145
4.5
Fault Analysis:
Under normal conditions, a power system operates under balanced conditions with all equipments
carrying normal load currents and the bus voltages within the prescribed limits. This condition can
be disrupted due to a fault in the system. A fault in a circuit is a failure that interferes with the
normal flow of current. A short circuit fault occurs when the insulation of the system fails resulting
in low impedance path either between phases or phase(s) to ground. This causes excessively high
currents to flow in the circuit, requiring the operation of protective equipments to prevent damage
to equipment. The short circuit faults can be classified as:
Symmetrical faults
Unsymmetrical faults
4.6
Symmetrical faults:
f to the
A three phase symmetrical fault is caused by application of three equal fault impedances Z
f = 0 the fault is called a solid or a bolted fault. These faults
three phases, as shown in Fig. 4.39. If Z
can be of two types: (a) line to line to line to ground fault (LLLG fault) or (b) line to line to line
fault (LLL fault). Since the three phases are equally affected, the system remains balanced. That
is why, this fault is called a symmetrical or a balanced fault and the fault analysis is done on per
phase basis. The behaviour of LLLG fault and LLL fault is identical due to the balanced nature
f = 0, this is usually the
of the fault. This is a very severe fault that can occur in a system and if Z
most severe fault that can occur in a system. Fortunately, such faults occur infrequently and only
about 5% of the system faults are three phase faults.
146
4.7
Unsymmetrical faults:
Faults in which the balanced state of the network is disturbed are called unsymmetrical or unbalanced
faults. The most common type of unbalanced fault in a system is a single line to ground fault (LG
fault). Almost 60 to 75% of faults in a system are LG faults. The other types of unbalanced faults
are line to line faults (LL faults) and double line to ground faults (LLG faults). About 15 to 25%
faults are LLG faults and 5 to 15% are LL faults. These faults are shown in Fig. 4.40.
4.8
In this type of faults all three phases are simultaneously short circuited. Since the network remains
balanced, it is analyzed on per phase basis. The other two phases carry identical currents but with a
phase shift of 120 . A fault in the network is simulated by connecting impedances in the network at
the fault location. The faulted network is then solved using Thevenins equivalent network as seen
from the fault point. The bus impedance matrix is convenient to use for fault studies as its diagonal
elements are the Thevenins impedance of the network as seen from different buses. Prior to the
occurrence of fault, the system is assumed to be in a balanced steady state and hence per phase
network model is used. The generators are represented by a constant voltage source behind a suitable
reactance which may be sub-transient, transient or normal d-axis reactance. The transmission lines
are represented by their -models with all impedances referred to a common base. A typical bus
147
BUS ]
Figure 4.42: Network representation for calculating [V
k are the equivalent load impedances as bus i and k respectively, zik is the impedance of line
and Z
148
BUS (F)] = [V
BUS (0)] + [V
BUS ]
[V
(4.64)
where,
T
BUS (F)] = Vector of bus voltages during fault =[V1 (F ) . . . Vi (F ) . . . Vn (F )]
[V
T
BUS (0)] = Vector of pre-fault bus voltages =[V1 (0) . . . Vi (0) . . . Vn (0)]
[V
T
BUS ] = Vector of change in bus voltages due to fault= [V1 . . . Vk . . . Vn ]
[V
BUS ] [V
BUS ]
[IBUS ] = [Y
(4.65)
n
0 Yn1 Ynk Ynn V
(4.66)
As the fault current Ik (F ) is leaving the bus it is taken as a negative current entering the bus.
Hence,
BUS ] [V
BUS ]
[IBUS (F)] = [Y
(4.67)
BUS ] = [Y
BUS ] [IBUS (F)] = [Z
BUS ] [IBUS (F)]
[V
(4.68)
(4.69)
(4.70)
Vk (F ) = Vk (0) Zkk Ik (F )
(4.71)
Vk (F ) = ZF Ik (F )
(4.72)
f = 0 and hence, Vk (F ) = 0. Thus the fault current Ik (F ) for bolted fault can
For a bolted fault Z
be expressed using equation (4.71) as,
Vk (0)
Ik (F ) =
Zkk
(4.73)
(4.74)
kk in equation (4.73) and equation (4.74) is the Thevenins impedance or openThe quantity Z
circuit impedance of the network as seen from the faulted bus k. From equation (4.70), the bus
voltage after fault for the unfaulted or healthy buses can be written as:
Vi (F ) = Vi (0) Zik Ik (F ) i = 1, 2, n, i k
(4.75)
Vi (F ) = Vi (0)
Zik
Vk (0)
Zkk + Zf
(4.76)
The fault current Iij (F ) flowing in the line connecting ith and j th bus can be calculated as
Vi (F ) Vj (F )
Iij (F ) =
zij
where zij is the impedance of line connecting buses i and j.
150
(4.77)
4.9
For the analysis of unsymmetrical or unbalanced faults, symmetrical component method is used.
The use of symmetrical components simplifies the analysis procedure of unbalanced system and also
helps in improving the understanding of the system behavior during fault conditions.
A review of symmetrical components is presented next.
4.9.1
Symmetrical components:
Any unbalanced set of three phase voltage or current phasors can be replaced by three balanced sets
of three phase voltage or current phasors. These three balanced set of voltage or current phasors
are called symmetrical components of voltages or currents. Let Ia , Ib , and Ic be an arbitrary set of
three current phasors representing phase currents. Then using symmetrical components they can be
expressed as:
I I I I
a a0 a1 a2
Ib = Ib0 + Ib1 + Ib2
Ic Ic0 Ic1 Ic2
(4.78)
Or,
Ib2 120
a = 1120
a2 = 1240
a3 = 1360
also 1 + a + a2 = 0
Ia1 = Ia1 1
Ib1 = a2 Ia1
Ic1 = aIa1
The phase sequence of the positive component set is abc.
Similarly the negative sequence set can be written as:
Ia2 = Ia2 2
where, 2 is the angle of phase a negative sequence current.
Ib2 = aIa2
Ic2 = a2 Ia2
152
I 1 1 1 I
a0
a
Ib = 1 a2 a Ia1
Ic 1 a a2 Ia2
(4.79)
[I]
[I]abc = [A]
012
where, [
I]
abc
(4.80)
T
[I]012 = set of sequence quantities = [Ia0 Ia1 Ia2 ]
1 1 1
1 a a2
1 1 1
1
1
= 1 a a2 thus,
where, A
3
1 a2 a
[I]
[I]012 = [A]
abc
(4.81)
1
Ia0 = [Ia + Ib + Ic ]
3
1
Ia1 = [Ia + aIb + a2 Ic ]
3
1
Ia2 = [Ia + a2 Ib + aIc ]
3
(4.82)
[V]
[V]
= [A]
abc
012
(4.83)
To summarize:
For voltage:
153
[V]
[V]
= [A]
012
abc
(4.84)
[I]
[I]abc = [A]
012
1
[I]
[I]012 = [A]
abc
(4.85)
(4.86)
where, [
I]abc is the set of phase voltages, and [I]012 is the set of sequence voltages.
Before starting unbalanced fault analysis, it is necessary to learn about the sequence networks of
different power system components, which we will discuss in the next lecture.
154
4.9.2
[E]abc = a2 [Ea]
a
(4.87)
As the generator is supplying a three-phase balanced load, the following KVL equations can be
written for each phase :
Va = Ea Zs Ia Zn In
Vb = Eb Zs Ib Zn In
Vc = Ec Zs Ic Zn In
155
(4.88)
Substituting the neutral current In = Ia + Ib + Ic in equation (4.88), and writing the resulting equation in matrix form, we get:
V E Z + Z
Zn
Zn Ia
a a s
n
Vb = Eb Zn
Zs + Zn
Zn Ib
Vc Ec Zn
Zn
Zs + Zn Ic
(4.89)
[V]
[I]abc
= [E]
[Z]
abc
abc
abc
(4.90)
where,
T
abc = [Va Vb Vc ] is the vector of terminal phase voltages.
[V]
T
[I]abc = [Ia Ib Ic ] is the vector of terminal phase currents.
abc is the impedance matrix which can be easily identified from equation (4.89).
[Z]
Replacing the phase quantities of equation (4.90) by corresponding sequence quantities, using
the transformation equation (4.83) and equation (4.85) one can write:
[I]
[E]
[V]
[A]
[A]
[Z]
= [A]
012
abc
012
012
[V]
[I]012
= [E]
[Z]
012
012
012
where,[Z]
012
(4.91)
(4.92)
Z + 3Z 0 0
s
0
0 Zs
[E]
012
are always balanced and contain only the positive sequence component.
Substituting [E]
012
and [Z]
012
V 0 Z 0 0 I
a0 0
a0
Va1 = Ea 0 Z1 0 Ia1
Va2 0 0 0 Z2 Ia2
(4.93)
(4.94)
Va2 = Z2 Ia2
From equation (4.94), it is evident that the three sequence components are independent of each
other.The current of a particular sequence produces a voltage drop of that sequence only, hence
the three sequences are decoupled from each other. The three sequence networks of a synchronous
generator are shown in Fig. 4.45.
4.9.3
For a static device such as a transmission line, the phase sequence of voltages and currents have no
effect on the impedance offered by the line as both positive and negative phase sequences encounter
identical line geometry. Hence, the positive and negative sequence impedances offered by a line are
1 = Z2 .
identical i.e. Z
The zero sequence currents, however, are in phase and flow through the conductors and return
through grounded neutral and/or ground wires. As a result, the ground or ground wire are to be
0 is, therefore,
included in the path of the zero sequence currents. The zero sequence impedance Z
1 and Z2 due to the inclusion of the ground return path. Z0 is usually more than
different from Z
1 or Z2 . The three sequence networks of the transmission lines are shown in Fig.
three times of Z
4.46.
4.9.4
For short circuit studies, the shunt magnetizing branch of transformer is neglected as the current
through it is negiligible as compared to short circuit current. The transformer is, therefore, modelled with an equivalent series leakage impedance. Since the transformer is also a static device like a
157
Figure 4.47: The zero-sequence equivalent circuit of a Star-Star transformer with both neutrals
grounded
(a) Star-Star connections with both neutrals grounded: Since both the neutrals are grounded,the
phasor sum of three unbalanced phase currents is equal to three times the zero sequence current
Ia0 (equation (4.82)). Hence, the zero sequence currents can flow in the primary and secondary
158
windings and the transhomer, therefore, can be represented by the equivalent zero-sequence
leakage impedance. The equivalent circuit is shown in Fig. 4.47.
(b) Star-Star connections with only one neutral grounded: When the neutral of only one winding is
grounded, the phase currents of the ungrounded winding must add up to zero. This implies that
the zero sequence currents can not exist in the ungrounded winding and hence the zero sequence
currents can not exist even in the transformer side with neutral grounded. The transformer in
this case, is represented as an open circuit between primary and secondary windings and the
equivalent circuit is shown in Fig. 4.48.
Figure 4.48: The zero-sequence equivalent circuit of a Star-Star transformer with only one neutral
grounded
(c) Star-Star connections with only no neutral grounded: In this case also the phasor sum of the
phase curents of both the windings is zero and hence the zero sequence currents can not exist on
any winding in this case also. The zero sequence equivalent network is represented as an open
circuit between the two windings and is shown in Fig. 4.49.
Figure 4.49: The zero-sequence equivalent circuit of a Star-Star transformer with both neutrals
ungrounded
159
(d) Star-delta connections with neutral grounded: A zero sequence current on the grounded starwinding will cause a circulating zero-sequence current in the closed delta-winding. However,
the zero-sequence current on the delta-winding can not exist on line side of the winding and
is confined only to the closed delta-winding. As a result, an open circuit exists between the
star and the delta sides. But, as the zero-sequence currents can exist on the line-side of the
grounded star winding, the zero-sequence leakage impedance of the transformer is connected to
ground on the star side of the transformer and an open circuit exists between the two windings.
The equivalent circuit for this connection is shown in Fig. 4.50.
Figure 4.50: The zero-sequence equivalent circuit of a Star-Delta transformer with Star-side neutral
grounded
(e) Star-delta connections with ungrounded neutral grounded: Since the neutral is isolated, no zerosequence current can exist in the star side of the transformer and as a consequence zero-sequence
currents can not exist in the delta side. The transformer is, therefore, represented as an opencircuit and the equivalent circuit is shown in Fig. 4.51.
Figure 4.51: The zero-sequence equivalent circuit of a Star-Delta trnasformer with Star-side neutral
ungrounded
(f) Delta-delta connections with ungrounded neutral grounded: In this case, the zero-sequence currents can only circulate within the closed delta windings and can not exit on line sides of both
160
the windings. Hence, an open circuit exists between the two windings as far as zero-sequence currents are concerned. To permit the circulating zero-sequence current to exist, the zero-sequence
leakage impedance is represented as a closed path with the ground. The equivalent circuit is
shown in Fig. 4.52.
Figure 4.53: The zero-sequence equivalent circuit of a Star-Delta transformer with neutral grounded
through impedance
Point to remember: If the neutral of a transformer is grounded through a grounding impedance
Zn , as shown in Fig. 4.53, then, the total zero-sequence equivalent impdance to be used in the
equivalent circuit is
Z0total = Z0 + 3Zn
(4.95)
This is due to the fact that the neutral current is 3 times the zero-sequence current per phase.
Next, the concepts of unsymmetrical fault analysis are developed with help of Thevenins equivalent circuit of sequence networks and symmetrical components in the next lecture.
161
4.9.5
An unloaded balanced three-phase synchronous generator with neutral grounded through an impedance
Zn is shown in Fig. 4.54. Suppose a single line to ground fault (LG) occurs on phase a though
f .
an impedance Z
Since the generator is unloaded, the following terminal conditions exist at the fault point:
Va = Zf Ia
Ib = 0
(4.96)
Ic = 0
Substituting Ib = Ic = 0 in equation (4.86), the symmetrical components of currents can be
calculated as:
1 1 1 I
I
a
a0
1
Ia1 = 1 a a2 0
3
Ia2
1 a 2 a 0
162
(4.97)
Solving the above equation, the values of the symmetrical components of fault current Ia are:
1
Ia0 = Ia1 = Ia2 = Ia
3
(4.98)
The voltage of phase a can be expressed in terms of symmetrical components from equation (4.83),
as
Va = Va0 + Va1 + Va2
(4.99)
Substituing in the equation the values of Va0 , Va1 and Va2 from equation (4.94) into equation (4.99),
Va can be written as (with Ia0 = Ia1 = Ia2 from equation (4.98)):
Va = Ea (Z0 + Z1 + Z2 )Ia0
(4.100)
Ia0 =
Ea
Z0 + Z1 + Z2 + 3Zf
(4.101)
If = Ia = 3Ia0 =
3Ea
Z0 + Z1 + Z2 + 3Zf
(4.102)
From equations (4.98) and (4.101), it be easily interpreted that the three sequence networks are
connected in series as shown in Fig. 4.55.
alent circuit (as seen from the fault point) is obtained, individually for the three sequence networks.
1th ,
For the positive sequence network Vth , the open circuit pre-fault voltage at the fault point, and Z
the positive sequence Thevenins equivalent impedance as seen from the fault point are determined.
2th and Z0th ,
For negative and zero sequence networks, only the Thevenins equivalent impedances Z
respectively are calculated. The three Thevenins equivalent networks are then connected in series.
4.9.6
f , on
Fig. 4.56 shows a line to line fault (LL) between phases b and c through an impedance Z
an unloaded three phase generator. The terminal conditions at the fault point are:
Vb Vc = Zf Ib
Ib + Ic = 0
164
(4.103)
Ia = 0
Substituting Ia = 0 and Ib = Ic in equation (4.86), the symmetrical components of cuurents can be
calculated as:
I
1 1 1 0
a0
1
Ia1 = 1 a a2 Ib
3
2
Ia2
1 a a Ib
(4.104)
Solving the above equation, the values of the symmetrical components of the current Ia are:
Ia0 = 0
1
Ia1 = (a a2 )Ib
3
1
Ia2 = (a2 a)Ib = Ia1
3
(4.105)
(4.106)
Substituting Va1 and Va2 from equation (4.94) and noting that Ia1 = Ia2 , one can write:
(4.107)
Ib =
3Ia1
(a a2 )
(4.108)
3Zf Ia1
(a a2 )(a2 a)
Since, (a a2 )(a2 a) = 3, the above expression can be simplified and written as:
Ia1 =
Ea
(Z1 + Z2 + Zf )
(4.109)
I 1 1 1 0
a
Ib = 1 a2 a Ia1
Ic 1 a a2 Ia1
Solving for the phase currents, the expressions for Ib and Ic can be written as:
165
(4.110)
Ib = Ic = (a2 a)Ia1
(4.111)
Figure 4.57: Connection of sequence networks for an LL fault between phases b and c of an
unloaded generator
Extending the above concept to LL fault calculations in a power system, it can be concluded that,
the Thevenins equivalent positive and negative sequence networks, as seen from the fault point, can
be connected in phase opposition through the fault impdedance for calculating fault current.
4.9.7
Fig. 4.58 shows a double line to ground (LLG) fault on phases b and c through an impedance
Zf on an unloaded three phase generator. The terminal conditions at the fault point are:
Vb = Vc = Zf If = Zf (Ib + Ic )
Ia = Ia1 + Ia2 + Ia0 = 0
(4.112)
b and V
c can be written as:
From equation (4.83), V
Vb = Va0 + a2 Va1 + aVa2
Vc = Va0 + aVa1 + a2 Va2
166
(4.113)
Va1 = Va2
(4.114)
Substituting Ib and Ic in terms of their sequence components from equation (4.85), voltage of phase
b can be expressed as:
Vb = 3Zf Ia0
(4.115)
Further substituting Vb from equation (4.115) and the condition of equation (4.114) in equation
(4.113), we get:
(4.116)
Substituting Va0 and Va1 from equation (4.94) in equation (4.116), the zero sequence component of
Figure 4.59: Connection of sequence networks for an LLG fault between phases b and c of an
unloaded generator
(Ea Z1 Ia1 )
Ia0 =
(Z0 + 3Zf )
(4.117)
For calculating the negative sequence component of current, Ia2 , substitute Va1 and Va2 from equation
(4.94) in equation (4.114). The expression for Ia2 is:
(Ea Z1 Ia1 )
Ia2 =
Z2
(4.118)
Finally, by substituting Ia0 and Ia2 from equations (4.117) and (4.118) in equation (4.112), the value
168
Ia1 =
Ea
Z2 (Z0 + 3Zf )
Z1 +
(Z0 + Z2 + 3Zf )
(4.119)
(4.120)
The equivalent circuit for the fault in terms of the sequence networks is shown in Fig. 4.59. The
circuit shown in Fig. 4.59 is based on equations (4.114) and (4.116). For LLG fault calculations in
a power system, the Thevenins equivalent of the three sequence networks, as seen from the fault
point, are found out. The positive and negative sequence equivalents are connected in parallel and
f . In the next lecture,
the combination is then connected to the zero sequence network through 3Z
BUS matrix.
we will study the procedure of unbalanced fault analysis using Z
169
4.10
BUS matrix:
Unbalanced fault analysis using Z
In the previous section, it is observed that, for fault calculations the Thevenins equivalent networks,
at the fault point, are needed for the three sequence networks. Since the three sequence networks are
BUS matrices of these sequence networks can be found seperately. The diagonal
independent, the Z
BUS matrices infact, are the Thevenins equivalent impedances
elements of the three sequence Z
of the sequence networks as seen from the various buses. Let, the three sequence bus impedance
matrices for zero sequence, positive sequence and negative sequence networks be represented as
th bus, then Z(0) , Z(1) and Z(2)
(0)
(1)
(2)
[Z
BUS ], [ZBUS ] and [ZBUS ] respectively. If the fault is at the k
kk
kk
kk
of the sequence bus impedance matrices are the zero, positive and negative Thevenins equivalent
impedances, respectively, as seen from the faulted bus. Hence, the first step in the fault analysis
BUS matrix is the determination of the three sequence networks and subsequently, finding
using Z
the bus impedance matrix for each sequence network.
To illustrate this step, consider the single line diagram of the power system shown in Fig. 4.60.
f as shown in Fig.
(a) LG fault: Let the fault be on phase a of bus k with a fault impedance Z
4.64.
From equations (4.98) and (4.101), it can be seen that the three equivalent sequence networks
are in series for calculating the sequence components of the fault currents. Hence, generalizing
equation (4.101) for fault at kth bus, the expression for sequence component of fault current can
be written as:
Vk (0)
(1)
(2)
(0)
Z(kk)
+ Z(kk)
+ Z(kk)
+ 3Zf
(4.121)
[I(012) (F)]
[I(abc)
(F)] = [A]
k
k
(4.122)
(b) LL fault: Let the fault be between phases phase b and phase c of bus k through an
f as shown in Fig. 4.65. From equation (4.109) and Fig. 4.57 it is observed that the
impedance Z
positive sequence and negative sequence equivalent networks are connected in phase opposition.
Thus, the expression of equation (4.109) for the sequence components of fault current at bus k
can be generalized as:
Ik(0) (F ) = 0
172
Ik(1) (F ) =
Vk (0)
= Ik(2) (F )
(0)
(1)
(2)
Z + Z + Z
kk
kk
(4.123)
kk
The phase components of fault current is the calculated from equation (4.122)
Ik (F ) = Ik(b) (F ) = Ik(c) (F )
(4.124)
(c) LLG fault: Fig. 4.66 shows an LLG fault involving phases phase b and phase c of bus k
f . Referring to equation (4.119) and Fig.4.66, the generalized expression
through an impedance Z
Ik(1) (F ) =
Vk (0)
(2)
(0)
Zkk
(Zkk
+ 3Zf )
(1)
Zkk + (2)
Z + Z (0) + 3Zf
kk
173
kk
(1) (1)
Vk (0) Zkk
Ik (F )
(2)
Ik (F ) =
(2)
Zkk
(4.125)
(1) (1)
Vk (0) Zkk
Ik (F )
(0)
Ik (F ) =
(0)
Zkk + 3Zf
The phase currents can be obtained from equation (4.122), the fault current is then calculated
as
Ik (F ) = Ik(b) (F ) + Ik(c) (F )
4.10.1
(4.126)
To calculate the voltages of buses during fault equation (4.94) can be generalized as:
(4.127)
[Vi(012) (F )]
Vi(abc) (F ) = [A]
(4.128)
I (F ) =
Iij(1) (F ) =
(2)
ij
I (F ) =
(0)
(1)
Vi(0) (F ) Vj(0) (F )
zij(0)
Vi(1) (F ) Vj(1) (F )
zij(1)
(4.129)
Vi(2) (F ) Vj(2) (F )
zij(2)
(2)
where zij , zij and zij are the zero, positive and negative sequence impedance respectively of
the line between bus i and bus j. The phase currents for the line can be calculated from the
symmetrical components using the relation:
174
Iij012 (F )]
[Iijabc (F )] = [A][
(4.130)
The process of fault analysis of a power system network is illustrated in the next lecture with
the help of an example.
175
4.11
1 = X
2 = 0.2 pu, X
0 = 0.05 pu
Generators G1 and G2 : X
1 = X
2 = X
0 = X
` = 0.05 pu
Transformers T1 and T2 : X
1 = X
2 = 0.1 pu, X
0 = 0.3 pu
Transmission Lines L1 , L2 and L3 : X
Figure 4.67: Single line diagram of the power System of the example
Prefault voltage for all buses is taken as Vi (0) = 1.000 pu i = 1, 2, 3.
We wish to carry out the complete short-circuit analysis of the system for:
(a) three phase bolted fault at bus 5
f = 0.1 pu at bus 5
(b) LG fault with Z
f = 0.1 pu at bus 5
(c) LL fault with Z
f = 0.0 pu at bus 5
(d) LLG fault with Z
Solution:
(a) Three phase fault at bus 5
For the three phase bolted fault, only positive sequence network and the positive sequence bus
(1)
impedance matrix [Z
Bus ] is required. The positive sequence network for the power system of
Fig. 4.67 is shown in Fig. 4.68. In this diagram all the elements have been replaced by their per
unit positive sequence impedances.
(1)
The [Z
Bus ] matrix for the network of the Fig. 4.68 is given below:
176
1
(1)
Bus
[Z
]=3
4
5
j0.1294
j0.0706
j0.1118
j0.0882
j0.10
j0.0706
j0.1294
j0.0882
j0.1118
j0.10
j0.1118
j0.0882
j0.1397
j0.1103
j0.1250
j0.0882 j0.10
j0.1118 j0.10
j0.1103 j0.1250 pu
j0.1397 j0.1250
j0.1250 j0.1750
The sequence component of three phase fault current at bus 5 are given as, from equation (4.73):
0
0
1
(012)
[I5 (F)] = 1 =
= j5.7143 pu
Z55 j0.1750
0 0
The phase components of the fault current are calculated using equation (4.122):
I5(012) (F)]
[I5(abc) (F)] = [A][
1 1 1
5.7143 900
0
(abc)
(abc)
[Ifault ] = [I5 (F)] = 1 a2 a j5.7143 = 5.71431500 pu
1 a a2
5.7143300
0
(1) (1)
V1(1) (F ) = V1(1) (0) Z15
I5 (F )
0.4285700
(abc)
0
0.428571200
Bus 2:
(1) (1)
V2(1) (F ) = V2(1) (0) Z25
I5 (F )
0.4285700
(abc)
[V2 (F)] = 0.42857 1200 pu
0.428571200
Bus 3:
(1) (1)
V3(1) (F ) = V3(1) (0) Z35
I5 (F )
(1)
0.2857100
(abc)
0
0.285711200
Bus 4:
178
(1) (1)
V4(1) (F ) = V4(1) (0) Z45
I5 (F )
0.2857100
(abc)
[V4 (F)] = 0.28571 1200 pu
0.285711200
0
(abc)
5 (F) = 0 pu because the fault
The bus voltage of bus 5 under faulted condition is V
0
impedance is zero.
For line L1 from bus 3 to bus 4 the positive sequence component for line current (I34 (F )) is
calculated as:
0
(abc)
I34 (F) = 0 pu
0
(1)
For line L2 from bus 3 to bus 5 the positive sequence component for line current (I35 (F )) is
calculated as:
I35 (F ) =
=
= 2.8571 900 pu
(1)
j0.1
Z35
Hence, the phase components of line current are
2.8571 900
(abc)
0
2.8571300
179
(1)
For line L3 from bus 4 to bus 5 the positive sequence component for line current (I45 (F )) is
calculated as:
=
I45 (F ) =
= 2.8571 900 pu
(1)
j0.1
Z45
Hence, the phase components of line current are
2.8571 900
(abc)
0
2.8571300
I13 (F ) =
=
= 2.8571 900 pu
(1)
j0.05
zT1
The phase components of the transformer T1 current are:
2.8571 900
(abc)
0
2.8571300
For transformer T2 between bus 2 and bus 4 the positive sequence component fault current
(1)
(I24
(F )) is calculated as:
2.8571 900
(abc)
0
2.8571300
180
2.8571 900
(abc)
0
2.8571300
(1)
For Generator G2 connected at bus 2 the positive sequence component fault current (IG2 (F ))
is calculated as:
2.8571 900
(abc)
0
2.8571300
The flow of fault current in the system is shown in the single line diagram of Fig. 4.69.
(b) Single line to ground fault at bus 5
181
In this case all the sequence networks are required. The positive sequence network is same as
(1)
the one shown in the Fig. 4.68 and [Z
Bus ] is identical to the matrix used in three phase fault
analysis.
The negative sequence equivalent network for this network is as shown in Fig. 4.70. The network
(1)
Bus ] = [Z
Bus ].
is identical to the network of Fig. 4.68 except for the voltage sources. Hence, [Z
The zero sequence equivalent network is drawn next considering the transformer connections
and grounding as well as generator grounding. The equivalent zero sequence networks is shown
in Fig. 4.71.
Transformer T2 has both the winding connected in delta, hence, no connection exists between
the primary and secondary sides for zero sequence currents to flow. To represent circulating
zero sequence currents in the delta connected transformer winding, it is represented as a short
circuited winding.
(0)
[Z
Bus ], the zero sequence bus impedance matrix is then calculated using the step-by-step ZBus
building algorithm. The zero sequence bus impedance matrix is given as:
1
1
(0)
Bus
[Z
]=3
4
5
Fault current:
The sequence component of the fault current at bus 5 are given as, from equation (4.121):
Vk (0)
1.0
= j1.538 pu
=
(1)
(2)
(0)
j0.175 + j0.175 + j0.30
Z55 + Z55 + Z55
1 1 1 j1.538
(abc)
(abc)
[Ifault ] = [I5 (F)] = 1 a2 a j1.538
1 a a2 j1.538
4.6154 900
(abc)
pu
[Ifault ] =
0
Bus voltages:
The bus voltage in sequence components, during fault, are calculated using equation (4.127)
written in compact form as:
V (0) (F ) 0 Z (0) 0
0 Ik(0) (F )
i
ik
(1)
V (F ) = Vi 0 Z (1) 0 I(1) (F )
i
ik
k
(2)
(2) (2)
Vi (F ) 0 0
0 Zik Ik (F )
(4.131)
V (0) (F ) 0 j0.05
0
0 j1.538
1
(1)
V (F ) = 1.0 0
j0.10
0 j1.538
1
(2)
V1 (F ) 0 0
0
j0.10 j1.538
Or,
0.0769
(012)
0.1538
The bus voltage in the phase form is calculated using equation (4.128).
0.615400
(abc)
1 (F)] = 0.9638 116.040 pu
[V
0.9638 116.040
V (0) (F ) 0 0.0
0
0 j1.538
2
(1)
V (F ) = 1.0 0 j0.10
0 j1.538
2
(2)
V2 (F ) 0 0
0
j0.10 j1.538
Or,
0.0
(012)
2 (F)] = 0.8462 pu
[V
0.1538
The bus voltage in the phase form is calculated using equation (4.128).
0.692300
(abc)
2 (F)] = 0.9326 111.790 pu
[V
0.9326 111.790
V (0) (F ) 0 j0.10
0
0 j1.538
3
(1)
V (F ) = 1.0 0
j0.125
0 j1.538
3
(2)
V3 (F ) 0 0
0
j0.125 j1.538
Or,
0.1538
(012)
3 (F)] = 0.8077 pu
[V
0.1923
184
The bus voltage in the phase form is calculated using equation (4.128).
0.461500
(abc)
3 (F)] = 0.9813 118.050 pu
[V
0.9813 118.050
V (0) (F ) 0 j0.20
0
0 j1.538
4
(1)
j1.538
V (F ) = 1.0 0
j0.125
0
(2)
V4 (F ) 0 0
j1.538
0
j0.125
Or,
0.3076
(012)
0.1923
The bus voltage in the phase form is calculated using equation (4.128).
0.307700
(abc)
0
1.0624 125.400
V (0) (F ) 0 j0.30
0
0 j1.538
5
(1)
j1.538
V (F ) = 1.0 0
j0.175
0
5
(2)
V5 (F ) 0 0
j1.538
0
j0.175
Or,
0.4615
(012)
0.2692
The bus voltage in the phase form is calculated using equation (4.128)
0.000
(abc)
0
1.087 128.640
Observe that the phase voltage of the faulted phase a is zero due to a zero impedance fault.
Line Currents
185
The sequence components of line currents during fault are calculated using equation (4.129),
written here in compact form as
(0) 0
0
V (0) (F ) V (0) (F )
I(0) (F ) zij
i
ij
j
1
(1)
(1)
(1)
I (F ) = 0
0 Vi (F ) Vj (F )
(1)
ij
zij
(2)
(2)
(2)
Iij (F )
1 Vi (F ) Vj (F )
0
0
zij(2)
(4.132)
zij(0) , zij(1) , zij(2) represent the respective sequence impedances of the line i j
Vi(0) (F ), Vi(1) (F ), Vi(2) (F ), Vj(0) (F ), Vj(1) (F ), Vj(2) (F ) are the sequence components of voltages of ith and jth buses respectively during fault.
Line 1: The sequence components of line current are
1.0
0
0
I(0) (F ) j0.3
34
0.1538 (0.3076)
(1)
1.0
I (F ) = 0
0 0.8077 0.8077
34
j0.10
(2)
I34 (F )
1.0 0.1923 (0.1923)
0
0
j0.10
Or,
j0.5128
(012)
pu
[I34 (F)] =
0
0.5128 900
(abc)
[I34 (F)] = 0.5128 900 pu
0.5128 900
1.0
0
0
I(0) (F ) j0.3
35
0.1538 (0.4615)
(1)
1.0
I (F ) = 0
0 0.8077 0.7308
35
j0.10
(2)
I35 (F )
1.0 0.1923 (0.2692)
0
0
j0.10
186
Or,
j1.0256
(012)
j0.7692
2.5641 900
(abc)
[I35 (F)] = 0.2564 900 pu
0.2564 900
1.0
0
0
I(0) (F ) j0.3
45
0.3077 (0.4615)
(1)
1.0
I (F ) = 0
0 0.8077 0.7308
45
j0.10
(2)
I45 (F )
1.0 0.1923 (0.2692)
0
0
j0.10
Or,
j0.5128
(012)
[I45 (F)] = j0.7692 pu
j0.7692
2.0513 900
(abc)
0
0.2564 900
Transformer Currents
1.0
0
0
I(0) (F ) j0.05
13
0.0769 (0.1538)
(1)
1.0
I (F ) = 0
0.8462 0.8077
0
13
j0.05
(2)
I13 (F )
0.1538
(0.1923)
1.0
j0.05
Or,
j1.538
(012)
j0.7692
187
3.0769 900
(abc)
[I13 (F)] = 0.7692 900 pu
0.7692 900
0
0
I(0) (F )
24
0 (0.3076)
1.0
(1)
I (F ) = 0
0 0.8462 0.8077
24
j0.05
(2)
I24 (F )
0.1538 (0.1923)
1.0
j0.05
(012)
j0.7692
1.538 900
(abc)
[I24 (F)] = 0.7692 900 pu
0.7692 900
Generator Currents
The sequence components of generator currents during fault are calculated using the expression
(0) 0
0
I(0) (F ) zgi
E (0) (F ) V (0) (F )
Gi
Gi
ti
1
(1)
(1)
(1)
I (F ) = 0
0
EGi (F ) Vti (F )
(1)
Gi
zgi
(2)
(2)
(2)
IGi (F )
1 EGi (F ) Vti (F )
0
0
(2)
zgi
(4.133)
where,
(0)
(1)
(2)
EGi
(F ), EGi
(F ), EGi
(F ) the zero, positive and negative sequence generated voltages respecth
tively of i generator.
Vti(0) (F ), Vti(1) (F ), Vti(2) (F ) are the zero, positive and negative sequence terminal voltages respectively of ith generator after fault.
(0)
(1)
(2)
zgi
(F ), zgi
(F ) and zgi
(F ) are the sequence impedances of the ith generator.
1.0
0
0
0 (0.0769)
I(0) (F ) j0.05
G1
(1)
1.0
I (F ) = 0
0 1 0.8462
G1
j0.20
(2)
IG1 (F )
1.0 0 (0.1538)
0
j0.20
j1.538
(012)
[IG1 (F)] = j0.7692 pu
j0.7692
3.0769 900
(abc)
0
0.7692 900
1.0
0
0
I(0) (F ) j0.05
00
G2
(1)
1.0
I (F ) = 0
0 1 0.8462
G2
j0.20
(2)
IG2 (F )
1.0 0 (0.1538)
0
0
j0.20
(012)
[IG2 (F)] = j0.7692 pu
j0.7692
1.538 900
(abc)
0
0.7692 900
The flow of sequence currents in the sequence networks is shown next in the Fig. 4.72. From
Fig. 4.72 the following points are worth observing:
Both generators contribute equal amount of positive and negative sequence currents as the
network is symmetrical as seen from the fault point.
Since the positive and negative sequence fault voltages are equal for buses 3 and 4, the
positive and negative sequence currents through line L1 between buses 3 and 4 are zero.
189
Moreover, the zero sequence network is not symmetrical, hence, zero sequence voltages of
buses 3 and 4 are not equal and as a result a zero sequence current flows through line L1 .
In the next lecture, we will look into the examples of short circuit fault calculation for LL and
LLG faults.
191
4.12
(2)
Bus ] and [Z
Bus ]
In this case only positive and negative sequence networks are required. Hence [Z
f = j0.1 pu.
as calculated previously will be used. Let the fault impedance be Z
Fault current calculations: The sequence components of fault current are calculated using equation (4.123).
I5(1) (F ) = 0
As zero sequence current can not flow without a ground path.
I5(1) (F ) =
1.0
V5(0)
=
= j2.2222 pu
(1)
(2)
Z55 + Z55 + Zf j0.175 + j0.175 + j0.1
I5(2) (F ) = I5(1) (F ) = j2.222 pu
Hence
(012)
j2.222
(abc)
3.849
Zero Sequence
V1(0) (F ) = 0.0 pu
Positive Sequence
(1) (1)
V1(1) (F ) = V1(1) (0) Z15
I5 (F ) = 1.0 j0.10 (j2.222)
V1(1) (F ) = 0.7778 pu
Negative Sequence
(2) (2)
V1(2) (F ) = Z15
I5 (F ) = j0.10 (j2.222)
V1(2) (F ) = 0.2222 pu
Hence, the sequence components of bus 1 voltage are:
012
1 (F)] = 0.7778 pu
[V
0.2222
1.0
(abc)
0
0.69391360
V2(0) (F ) = 0.0 pu
Positive Sequence
(1) (1)
V2(1) (F ) = V2(1) (0) Z25
I5 (F ) = 1.0 j0.10 (j2.222)
V2(1) (F ) = 0.7778 pu
Negative Sequence
(2) (2)
V2(2) (F ) = Z25
I5 (F ) = j0.10 (j2.222)
193
V2(2) (F ) = 0.2222 pu
Hence, the sequence components of bus 2 voltage are:
012
0.2222
1.0
(abc)
0
[V
(2) (F)] = 0.6939 136 pu
0.69391360
V3(0) (F ) = 0.0 pu
Positive Sequence
(1) (1)
V3(1) (F ) = V3(1) (0) Z35
I5 (F ) = 1.0 j0.125 (j2.222)
V3(1) (F ) = 0.7222 pu
Negative Sequence
(2) (2)
V3(2) (F ) = Z35
I5 (F ) = j0.125 (j2.222)
V3(2) (F ) = 0.2778 pu
Hence, the sequence components of bus 3 voltage are:
012
3 (F)] = 0.7222 pu
[V
0.2778
1.0
(abc)
0.6310 142.410 pu
[V
=
(F)]
(3)
0
0.6310142.41
V4(0) (F ) = 0.0 pu
Positive Sequence
(1) (1)
V4(1) (F ) = V4(1) (0) Z45
I5 (F ) = 1.0 j0.125 (j2.222)
V4(1) (F ) = 0.7222 pu
Negative Sequence
(2) (2)
V4(2) (F ) = Z45
I5 (F ) = j0.125 (j2.222)
V4(2) (F ) = 0.2778 pu
Hence, the sequence components of bus 4 voltage are:
012
0.2778
1.0
(abc)
0
0.6310142.410
V5(0) (F ) = 0.0 pu
Positive Sequence
(1) (1)
V5(1) (F ) = V5(1) (0) Z55
I5 (F ) = 1.0 j0.175 (j2.222)
195
V5(1) (F ) = 0.6111 pu
Negative Sequence
(2) (2)
V5(2) (F ) = Z55
I5 (F ) = j0.175 (j2.222)
V5(2) (F ) = 0.3889 pu
Hence, the sequence components of bus 5 voltage are:
012
5 (F)] = 0.6111 pu
[V
0.3889
1.0
(abc)
0
0.5358158.950
1.0
0
0
I(0) (F ) j0.3
0
34
(1)
1.0
I (F ) = 0
0 0.7222 0.7222
34
j0.10
(2)
I34 (F )
1.0 0.2778 0.2778
0
0
j0.10
0
(012)
[I34 (F)] = 0 pu
0
The line current in phase form is calculated as:
0
(abc)
[I34 (F)] = 0 pu
0
Line 2: The sequence components of line current are
196
1.0
0
0
I(0) (F ) j0.3
0
35
(1)
1.0
I (F ) = 0
0 0.7222 0.6111
35
j0.10
(2)
I35 (F )
1.0 0.2778 0.3889
0
0
j0.10
(012)
j1.1111
(abc)
[I35 (F)] = 1.9245 pu
1.9245
1.0
0
0
I(0) (F ) j0.3
45
(1)
1.0
I (F ) = 0
0 0.7222 0.6111
45
j0.10
(2)
I45 (F )
1.0 0.2778 0.3889
0
0
j0.10
(012)
[I45 (F)] = j1.1111 pu
j1.1111
(abc)
[I45 (F)] = 1.9245 pu
1.9245
1.0
0
0
I(0) (F ) j0.05
0
13
(1)
1.0
I (F ) = 0
0 (0.7778 0.7222)
13
j0.05
(2)
I13 (F )
1.0 (0.2222 0.2778)
0
0
j0.05
197
(012)
[I13 (F)] = j1.1111 pu
j1.1111
(abc)
1.9245
1.0
0
0
I (F ) j0.05
1.0
I (F ) = 0
0 (0.7778 0.7222)
j0.05
I (F )
1.0 (0.2222 0.2778)
0
0
j0.05
(0)
24
(1)
24
(2)
24
(012)
j1.1111
(abc)
1.9245
1.0
0
0
I(0) (F ) j0.05
0
G1
(1)
1.0
I (F ) = 0
0 (1 0.7778)
G1
j0.20
(2)
IG1 (F )
(0 0.2222)
1.0
j0.20
(012)
j1.1111
198
(abc)
1.9245
1.0
0
0
I(0) (F ) j0.05
0
G2
(1)
1.0
I (F ) = 0
0 (1 0.7778)
G2
j0.20
(2)
IG2 (F )
1.0 (0 0.2222)
0
j0.20
(012)
j1.1111
(abc)
1.9245
(2)
(0)
(1)
In this case all the three sequence networks are required. Hence, [Z
Bus ],[ZBus ],and [ZBus ] as
f = 0.
calculated earlier will be used. It is assumed that the fault impedance Z
Fault current calculations:
The sequence components of fault current are calculated using equation (4.125) as follows:
Positive Sequence Current
199
I5(1) (F ) =
I5(1) (F ) =
(1)
Z55
+
1.0
(2) (0)
Z55
(Z55 + 3Zf )
(2)
(0)
Z55
+ Z55
+ 3Zf
1.0
j0.175(j0.175 + 3 0)
j0.175 +
j0.175 + j0.3 + 3 0
I5(1) (F ) = j3.5023 pu
The negative and zero sequence currents are calculated using current division as:
Negative Sequence Current
(0)
Z55
I5(1) (F )
(0)
(2)
Z55 + Z55
j0.3
=
(j3.5023)
(j0.3 + j0.175)
I5(2) (F ) =
I5(2) (F ) = j2.212 pu
Zero Sequence Current
200
(2)
Z55
I5(1) (F )
(0)
(2)
Z55 + Z55
j0.175
(j3.5023)
=
(j0.3 + j0.175)
I5(0) (F ) =
I5(0) (F ) = j1.2903 pu
Hence, the fault current in sequence components is :
j1.290
(012)
[I5 (F)] = j3.503 pu
j2.212
(abc)
0
5.313721.360
) = I5(b) (F ) + I5(c) (F )
I(F
= 5.3137158.640 + 5.313721.360
) = 3.871900 pu
I(F
Bus Voltage Calculations:
Bus 1:
The sequence components of bus 1 voltage are calculated as
V (0) (F ) 0 j0.05
0
0 j1.290
1
(1)
V (F ) = 1.0 0
j0.10
0 j3.503
1
(2)
V1 (F ) 0 0
0
j0.10 j2.212
201
0.0645
(012)
1 (F)] = 0.6497 pu
[V
0.2212
0.935500
(abc)
0
0.5248134.990
Bus 2:
The sequence components of bus 2 voltage are calculated as
V (0) (F ) 0 0
0
0 j1.290
2
(1)
j3.503
V (F ) = 1.0 0 j0.10
0
(2)
V2 (F ) 0 0
j2.212
0
j0.10
(012)
2 (F)] = 0.6497 pu
[V
0.2212
0.871000
(abc)
0
0.5722139.560
Bus 3:
The sequence components of bus 3 voltage are calculated as
V (0) (F ) 0 j0.10
0
0 j1.290
3
(1)
V (F ) = 1.0 0
j0.125
0 j3.503
3
(2)
V3 (F ) 0 0
0
j0.125 j2.212
0.1290
(012)
0.2765
0.976600
(abc)
3 (F)] = 0.3815 139.560 pu
[V
0.3815139.560
Bus 4:
The sequence components of bus 5 voltage are calculated as
V (0) (F ) 0 j0.20
0
0 j1.290
4
(1)
V (F ) = 1.0 0
j0.125
0 j3.503
4
(2)
V4 (F ) 0 0
0
j0.125 j2.212
0.2581
(012)
0.2765
1.096800
(abc)
4 (F)] = 0.2954 123.100 pu
[V
0.2954123.100
Bus 5:
The sequence components of bus 5 voltage are calculated as
V (0) (F ) 0 j0.30
0
0 j1.290
5
(1)
V (F ) = 1.0 0
j0.175
0 j3.503
5
(2)
V5 (F ) 0 0
0
j0.175 j2.212
0.3871
(012)
5 (F)] = 0.3871 pu
[V
0.3871
1.161300
(abc)
pu
[V5 (F)] =
0
Line L1
203
0
0
I(0) (F ) j0.30
34
(0.1290 0.2581)
(1)
1
I (F ) = 0
(0.5622 0.5622)
0
34
j0.10
(2)
I34 (F )
(0.2765
0.2765)
1
j0.10
Hence,
j0.4301
(012)
[I34 (F)] = 0 pu
j0.4301
(abc)
j0.4301
Line L2
The sequence components of Line 2 current are calculated as
0
0
I(0) (F ) j0.30
(0.1290 0.3871)
35
(1)
1
I (F ) = 0
0 (0.5622 0.3871)
35
j0.10
(2)
I35 (F )
1 (0.2765 0.3871)
0
0
j0.10
Hence,
j0.8602
(012)
[I35 (F)] = j1.7511 pu
j1.1060
0.2151900
(abc)
0
0.295425.550
Line L3
204
0
0
I(0) (F ) j0.30
45
(0.2581 0.3871)
(1)
1
I (F ) = 0
(0.5622 0.3871)
0
45
j0.10
(2)
I45 (F )
(0.2765
0.3871)
1
j0.10
Hence,
j0.4301
(012)
[I45 (F)] = j1.7511 pu
j1.1060
0.2151 900
(abc)
[I45 (F)] = 2.5863163.080 pu
2.586316.920
Transformer T1
The sequence components of Transformer 1 current are calculated as:
0
0
I(0) (F ) j0.05
(0.0645 0.1290)
13
(1)
1
I (F ) = 0
0 (0.6497 0.5622)
13
j0.05
(2)
I13 (F )
1 (0.2212 0.2765)
0
0
j0.05
Hence,
j1.290
(012)
j1.106
0.6452900
(abc)
0
2.953633.100
205
Transformer T2
The sequence components of Transformer 2 current are calculated as
0
0
I(0) (F )
(0.0 0.2581)
24
1
(1)
I (F ) = 0
0 (0.6497 0.5622)
24
j0.05
(2)
I24 (F )
1 (0.2212 0.2765)
0
0
j0.05
Hence,
(012)
[I24 (F)] = j1.750 pu
j1.106
0.6452 900
(abc)
0
2.49537.430
0
0
I(0) (F ) j0.05
G1
(0.0 0.0645)
(1)
1
I (F ) = 0
0 (1.0 0.6497)
G1
j0.20
(2)
IG1 (F )
1 (0.0 0.2212)
0
0
j0.20
206
j1.290
(012)
[IG1 (F)] = j1.750 pu
j1.106
0.645200
(abc)
[IG1 (F)] = 2.9536146.900 pu
2.953633.100
0
0
I(0) (F ) j0.05
G2
(0.0 0.0)
(1)
1
I (F ) = 0
0 (1.0 0.6497)
G2
j0.20
(2)
IG2 (F )
1 (0.0 0.2212)
0
j0.20
0.0
(012)
j1.106
0.645200
(abc)
0
2.49537.430
The flow of sequence currents in the sequence networks is shown in the Fig.4.74.
We will be discussing open-conductor faults in the next lecture.
207
4.13
When one or two phases of a balanced three-phase line opens it creates an unbalance in the
system and results in the flow of unbalanced currents. Such conditions occur in the system when
one or two conductors of a trnansmission line are broken due to storm or if fuses, isolators or
circuit breakers operate only on one or two phases leaving others connected. Such open conductor
Bus ] matrices of sequence networks.
faults can also be analysed with the help of [Z
0 ] be the vector of open-circuit bus voltages corresponding to the initial (pre-fault) value
Let [V
Bus ]. We can
of bus current vector [
I0 ] injected in a network with bus impedance matrix [Z
then write
208
0 ] = [Z
Bus ][I0 ]
[V
(4.134)
can be
If the bus currents are changed to a new value, [
I0 + I], the new bus voltage [V]
expressed as:
= [Z
Bus ] [I0 + I]
[V]
Bus ][I0 ] + [Z
Bus ][I]
= [Z
(4.135)
0 ] + [V]
= [V
represents the change in the values of the original bus voltage [V
0 ].
where,[V]
Fig. 4.76 represents a power system with buses i and j taken out along with the reference node.
0 ] and [I0 ] are zero. Currents [Ii ] and [Ij ] are
The circuit is not energised so that [V
injected into the ith and jth buses respectively, through current sources connected between the
node and the reference node.
Vi i
V = j
VN N
Z11 Z1i
Zi1 Zii
Zj1 Zji
ZN 1 ZN i
Z1j Z1N
Zij ZiN
Zjj ZjN
ZN j ZN N
I
1
Ii
I
j
IN
V Z I + Z I
1j
j
1
1i i
Vi Zii Ii + Zij Ij
V = Z I + Z I
jj
j
j
ji i
VN ZN i Ii + ZN j Ij
(4.136)
(4.137)
(4.138)
(4.139)
(4.140)
Zji = Zij
(4.141)
Thus the equations (4.138) and (4.140) can be represented by an equivalent circuit shown in
Fig. 4.77, which is also the Thevenins Equivalent circuit of the network as seen from the ith
and jth buses.
From the figure it can be observed that the Thevenins open circuit voltage between ith and jth
buses is (Vi0 Vj0 ).
210
Vi = (Zii Zij )I
(4.142)
(4.143)
Next, calculate the voltage difference Vij between ith and jth buses as:
(4.144)
Hence,
ZT hevenin,ij =
Vij
= (Zii + Zjj 2Zij )
I
(4.145)
Once the Thevenins equivalent is established, the analysis of open-conductor faults can proceed
further. The opening of all the three phases is equivalent to the removal of the line i j totally
(0)
(1)
(2)
from the network. If zij , zij and zij are the the three sequence impedance of the line i
j, then the removal of this line from the network can be simulated by adding
zij(0) ,
zij(1) and
zij(2) to the corresponding Thevenins equivalent network of the three sequence networks of the
original network as seen from ith and jth buses.
Let x represents the fractional length of the broken line i j from ith bus to the break point
k, where 0 x 1.
211
The positive sequence impedance of the conductor segment between the ith bus and the point of
break k is x
zij(1) , and the positive sequence impedance of the remaining conductor from point k to
j th bus is (1 x)
zij(1) . These two impedances are then added to represent the broken conductor.
This is illustrated in Fig. 4.78
Figure 4.78: Positive sequence equivalent network with line open between buses k and k
(a)
(b)
(c)
If Vkk , Vkk and Vkk represent the phase component of voltage drops between points k and
(0)
(1)
(2)
k, then Vkk , Vkk and Vkk represent the sequence components of the voltage drops between
points k and k. These sequence voltages have different values depending on the type of open
conductor fault.
(1)
To further simplify the circuit, the voltage Vkk and the total series impedance [x
zij(1) + (1
x)
zij(1) ] = zij(1) is replaced by a current source
Vkk(1)
(1)
ij
(1)
Fig. 4.79.
(1)
(1)
Vkk(1) Vkk(2)
,
(1)
zij
zij(2)
and
Vkk(0)
zij(0)
k. If no conductor is open then the sequence voltages are all zero and the current sources are
not present in the equivalent circuit. Further, the current sources can be regarded as current
injections into the buses i and j of the original sequence networks. All through the calculation
212
Figure 4.80: Final positive sequence Thevenins Equivalent circuit repesenting the opening of line i
j between buses k and k
(0)
(1)
(2)
Bus ], [Z
Bus ] and [Z
Bus ] of the original network are
process, the bus impedance matrices [Z
used.
Figure 4.81: Final (a) negative sequence (b) zero sequence Thevenins Equivalent circuit repesenting
the opening of line i j between k and k
Positive Sequence Negative Sequence Zero sequence
at ith bus
at j th bus
Vkk(1)
Vkk(2)
zij(1)
Vkk(1)
(1)
zij
Vkk(0)
zij(2)
Vkk(2)
(2)
zij
(0)
(1)
zij(0)
Vkk(0)
(0)
zij
(2)
The sequence voltage drops Vn , Vn and Vn at any bus n due to the current injections
at the buses i and j can be calculates from equation (4.136) as:
(0)
n
Vn(1) =
Vn(2) =
(0)
(0) (0)
(Zni
Znj
)Vkk
zij(0)
(1)
(1) (1)
(Zni
Znj
)Vkk
zij(1)
(2)
(2) (2)
(Zni
Znj
)V
(2)
ij
(4.146)
kk
Next, the Thevenins equivalent impedances for each sequence network, as seen from the busesk
and k, are calculated as follows:
(1)
Zth,ij
(
zij(1) )
+ (1 x)
zij(1)
(1)
(1)
Z
+ (
zij )
th,ij
214
(1)
kk
(
zij(1) )2
= (1)
Z
zij(1)
(4.147)
th,ij
Similarly from Fig. 4.81 (a) and (b), the negative sequence and zero sequence Thevenins equivalent impedances can be expressed as:
(2)
Zkk
=
(
zij(2) )2
Z (2) zij(2)
th,ij
(4.148)
(0)
Zkk
=
(
zij(0) )2
Z (0) zij(0)
th,ij
(
zij(1) )2
(Vi(1) Vj(1) )
= (1)
(1)
Z
zij
(4.149)
th,ij
substituting
(1)
Zkk
zij(1)
(1)
Vth,kk
is obtained as:
zij(1)
(1)
Zth,ij
zij(1)
(1)
Zkk
(1)
Vth,kk
=
(Vi(1) Vj(1) )
(1)
zij
(4.150)
(1)
Also prior to the occurance of open-conductor fault on any conductor, the current Iij flowing
in phase a is the positive sequence component and is given by the relation:
Iij(1) =
(Vi(1) Vj(1) )
zij(1)
(4.151)
(4.152)
The Thevenins equivalent network as seen from points k and k for the three sequence networks
are shown in Fig. 4.82
We are now ready to discuss the two possible cases of open-circuit fault i.e. (a) open phase open
(b) two phases open. We will be discussing them in the next lecture.
215
4.13.1
Consider that phase a conductor is open as shown in Fig. 4.75(a), hence phase a current Ia = 0.
As a result:
(2)
(4.153)
(0)
where, Ia , Ia and Ia are the symmetrical components of phase a current. Since phases b and
c are closed , the voltage drops .
Vkk ,b = 0 Vkk ,c = 0
216
(4.154)
The symmetrical components of voltage drops across the fault point can be calculated as :
V (0)
1 1 1 V
V
kk ,a
a
kk ,a
(1) 1
1
Va = 1 a a2 0 = Vkk ,a
3
(2)
2
Va
1 a a 0
Vkk ,a
(4.155)
1
Va(0) = Va(1) = Va(2) = Vkk ,a
3
(4.156)
Hence,
It implies that open conductor in phase a causes equal voltages to appear across points k and
k of each sequence network. Hence, the three equivalent sequence networks can be connected in
parallel across points k and k as shown in Fig. 4.83.
Figure 4.83: Connection of Equivalent sequence networks to represent open phase a between k and
k
(1)
The current Ia is given as :
Ia(1) = Iij
(1)
Zkk
+
(1)
Zkk
(2)
Z Z (0)
kk
kk
(2)
(0)
Zkk
+ Zkk
simplifying
Ia(1) = Iij
(1) (2)
(0)
Zkk
[Zkk + Zkk ]
Z (0) Z (1) + Z (1) Z (2) + Z (2) Z (0)
kk
kk
kk
kk
kk
(4.157)
kk
(1)
(2)
(0)
The sequence voltage drops Vkk , Vkk and Vkk can be calculated with reference to Fig.4.83 as:
(2) (0)
(1)
(1) Zkk Zkk
Vkk = Ia
(2)
(0)
Zkk
+ Zkk
(1)
Substituting Ia and simplifying we get:
217
(1)
(2)
kk
kk
kk
kk
(4.158)
kk
(0)
, Z , and Z are obtained from the impedance parameters of the sequence networks
Z
kk
kk
kk
[equation (4.147) and equation (4.148)].
Iij is the pre-fault current or load current in phase a of the line i j
Next, the equivalent injected currents
(0)
(1)
(2)
Vkk(1) Vkk(2)
,
(1)
zij
zij(2)
and
Vkk(0)
zij(0)
are calculated.
4.13.2
(4.159)
(4.160)
I(0)
1 1 1 I
a
a
(1) 1
Ia = 1 a a2 0
3
(2)
2
1 a a 0
Ia
Simplifying one gets the condition:
1
Ia(0) = Ia(1) = Ia(2) = Ia
3
(4.161)
equation (4.161) indicates that the three equivalent sequence networks are in series and to ensure
V + Va(1) + Va(2) = 0 the circuit should be closed.The interconnection of the sequence networks is
shown in Fig.4.84.
From the equivalent circuit of Fig.4.84, the sequence currents can be calculated as :
(0)
a
218
Figure 4.84: Connection of Equivalent sequence networks to represent open phases b and c between
k and k
(1)
Zkk
(0)
(1)
Z + Z + Z (2)
kk
kk
(4.162)
kk
ij (0)
(1)
(2)
Zkk + Zkk
+ Zkk
(1) (0)
Zkk
Zkk
(0)
ij (0)
(1)
(2)
Zkk + Zkk
+ Zkk
(4.163)
Remaining calculations are similar to those performed for single conductor open case. In the
next lecture we will be looking at an example of open conductor fault analysis.
219
4.14
(2)
(1)
Zth,45
= Zth,45
= j0.0647 pu
(0)
(0)
(0)
(0)
Zth,45
= Z44
+ Z55
2Z45
= j0.30 + j0.30 2 0.20 = j0.20 pu
(1)
(2)
(0)
Zkk = (1)
=
= j0.2833 pu
(1)
j0.0647 j0.10
Zth,45 z45
(2)
(1)
Zkk
= Zkk = j0.2833 pu
(0) 2
(
z45
)
(j0.30)2
(0)
Zkk = (0)
=
= j0.90 pu
(0)
j0.20 j0.30
Zth,45 z45
(1)
The current Ia is calculated using equation (4.157)
Ia(1) = Iij
(1) (2)
(0)
Zkk
[Zkk + Zkk ]
Z (0) Z (1) + Z (1) Z (2) + Z (2) Z (0)
kk
kk
kk
kk
kk
kk
j0.2833(j0.2833 + j0.3)
= (0.2170 j0.2146) pu
j0.9 j0.2833 + j0.2833 j0.2833 + j0.2833 j0.9
220
(1)
(2)
(0)
The sequence voltage drops Vkk , Vkk and Vkk are then calculated using the equation:
(2) (0)
(1)
(1) Zkk Zkk
Vkk = Ia
(2)
(0)
Zkk
+ Zkk
j0.2833 j0.9
Vkk(1) = (0.2170 j0.2146)
= (0.0463 + j0.0468) pu
j0.2833 + j0.9
Vkk(2) = Vkk(0) = Vkk(1) = (0.0463 + j0.0468) pu
As a check calculate
Ia(2)
Vkk(2)
Z (2)
0.0463 + j0.0468
= (0.1651 + j0.1633) pu
j0.2833
Ia(0)
V (0)
= kk
Z (0)
0.0463 + j0.0468
= (0.052 + j0.0514) pu
j0.90
kk
kk
Q.E.D.
V3 =
=
(0.0463 + j0.0468) = 0 pu
(0)
j0.30
z45
(1)
(1)
(Z34
Z35
)Vkk(1) j0.1103 j0.1250
(1)
(0)
(0)
(Z44
Z45
)Vkk(0) j0.30 j0.20
(0)
V4 =
=
(0.0463 + j0.0468) = (0.0154 + j0.0156) pu
(0)
j0.3
z45
221
(1)
(1)
(Z44
Z45
)Vkk(1) j0.01397 j0.125
(1)
V4 =
=
(0.0463 + j0.0468) = (0.0068 + j0.0069) pu
(1)
j0.10
z45
(0)
(0)
(Z54
Z55
)Vkk(0) j0.20 j0.30
(0)
V5 =
=
(0.0463 + j0.0468) = (0.0154 j0.0156) pu
(0)
j0.3
z45
(1)
(1)
(Z54
Z45
)Vkk(1) j0.1250 j0.1750
(1)
=
(0.0463 + j0.0468) = (0.0231 j0.0234) pu
V5 =
(1)
j0.3
z45
V3 (F ) = 0.9053 9.750 pu
V4 (F ) = 0.9392 8.020 pu
V5 (F ) = 0.8430 18.040 pu
The phase components of current in line 4-5 are:
45
(b)
I = 1 a2 a (0.2170 j0.2146) = 0.4782 147.840 pu
45
(c)
0
2
I45 1 a a (0.1651 + j0.1633) 0.478258.56
TWO CONDUCTORS OPEN
The sequence currents can be calculated with the help of equation (4.162) as:
Ia(1) = Iij
(1)
Zkk
(0.3821 j0.3779) j0.2833
=
(0)
(1)
(2)
j0.9 + j0.2833 + j0.2833
Zkk + Zkk + Zkk
(0)
(2)
Vkk(1) = Ia(1) (Zkk
+ Zkk ) = (0.0738 j0.073) (j0.9 + j0.2833) = (0.0864 + j0.0873) pu
(2)
Vkk(2) = Ia(2) Zkk
= (0.0738 j0.073)(j0.2833) = (0.0207 + j0.0209) pu
(0)
Vkk(0) = Ia(0) Zkk
= (0.0738 j0.073)(j0.9) = (0.0657 + j0.0664) pu
V3 =
=
(0.0657 j0.0664) = 0 pu
(0)
j0.3
z45
(1)
(1)
(Z34
Z35
)Vkk(1) j0.1103 j0.1250
(1)
V3 =
=
(0.0864 + j0.0873) = (0.0127 j0.0128) pu
(1)
j0.1
z45
(2)
(2)
(Z34
Z35
)Vkk(2) j0.1103 j0.1250
(2)
V3 =
=
(0.0207 j0.0208) = (0.0030 + j0.0030) pu
(2)
j0.1
z45
V3 (F ) = 0.9085 9.460 pu
Bus 4
(0)
(0)
(Z44
Z45
)Vkk(0) j0.30 j0.20
(0)
=
(0.0657 j0.0664) = (0.0219 j0.0221) pu
V4 =
(0)
j0.30
z45
(1)
(1)
(Z44
Z45
)Vkk(1) j0.1397 j0.1250
(1)
=
(0.0864 + j0.0873) = (0.0127 + j0.0128) pu
V4 =
(1)
j0.1
z45
(2)
(2)
(Z44
Z45
)Vkk(1) j0.1397 j0.1250
(2)
V4 =
=
(0.0207 j0.0209) = (0.0030 j0.0031) pu
(2)
j0.1
z45
V4 (F ) = 0.9054 11.00 pu
Bus 5
(0)
(0)
(Z54
Z55
)Vkk(0) j0.20 j0.30
(0)
=
(0.0657 j0.0664) = (0.0219 + j0.0221) pu
V5 =
(0)
j0.30
z45
(1)
(1)
(Z54
Z55
)Vkk(1) j0.1250 j0.1750
(1)
V5 =
=
(0.0864 + j0.0873) = (0.0432 j0.0437) pu
(1)
j0.10
z45
224
(2)
(2)
(Z54
Z55
)Vkk(2) j0.1250 j0.1750
(2)
V5 =
=
(0.0207 j0.0209) = (0.0103 + j0.0105) pu
(2)
j0.10
z45
(b)
I = 1 a2 a (0.0738 j0.073) =
pu
0
45
(c)
2
I45 1 a a (0.0738 j0.073)
225
Module 5
Contingency Analysis
5.1
In Chapter 2, we have discussed about the methods for analyzing the power system performance at a
particular operating point. However, for practical system operation, apart from ensuring the satisfactory operation of the system at a particular operating condition, it is also equally important to make
sure that the system operates with adequate level of security. Broadly, the term security implies
the ability of the system to operate within system constraints (on bus voltage magnitudes, current
and power flow over the lines) in the event of outage (contingency) of any component (generator or
transmission line). Now, if the system is operating at high loading (light loading) conditions, then
the post-contingency system condition would be highly stressed (lightly stressed). Therefore, the
post contingency values of different quantities (voltages, current/power flow) depend on the present
operating condition. In case the post-outage (post contingency) does not involve any violation of
any operating constraints, the system is said to be operating securely. Otherwise, the system is said
to enter an emergency operating condition. Therefore, for detecting the possibility of appearance
of emergency operating conditions, analysis of the post-contingency scenario (henceforth termed as
contingency analysis) of the system needs to be carried out.
Now, in any practical sized power system, there is a very large number of elements. Hence, for
carrying out contingency analysis, outages of all these elements (preferably) need to be carried out
one-by-one corresponding to any particular operating condition. However, in any power system, the
operating point of the system changes quite frequently with change is loading/generating conditions.
With the change in system operating conditions, the contingency analysis exercise needs to be
carried out again at the new operating point. Thus, for proper monitoring of system security, a
large number of outage cases need to be simulated repeatedly over a short span of time. Ideally,
these outage cases should be studied with the help of full AC load flow solutions. However, analysis
of thousands of outage cases with full AC power flow technique will involve a significant amount of
computation time and as a result, it might not be possible to complete this entire exercise before the
new operating condition emerges. Therefore, instead of using full non-linear AC power flow analysis,
approximate, but much faster techniques based on linear sensitivity factors are used to estimate the
post contingency values of different quantities of interest. The basic concept of sensitivity factors
226
is described below. Essestially, the linear sensitivity factors approximately estimate the changes
in different line flows for any particular outage condition without the need of full AC power flow
solution.
Basically, there are two types of sensitivity factors and these are:
a. Generation outage sensitivity factor (GOSF)
b. Line outage sensitivity factor (LOSF)
GOSF relates the approximate change in power flow in line i-j (i.e. between bus i and j) due
to the outage of generator at bus k, whereas LOSF helps to calculate the approximate change in
power flow in line i-j due to outage of line m-n.
The generation outage sensitivity factor is defined by,
ijk =
fij
Pk
(5.1)
k
where, ij
GOSF of line i-j for generation change at bus k
k
The factor ij
denotes the sensitivity of the line flow on line i-j due to change in generation
at bus k. In equation (5.1), it is assumed that the generation lost at bus k would be exactly
compensated by the reference or slack bus. Now, if the generation at bus k was generating an
amount of power equal to Pk0 , then to represent the outage condition, Pk = Pk0 .
Hence, the new power flow over the line i-j would be given as,
(5.2)
k
The factor ij
would be pre-calculated and stored in the memory. As we will see later, the values
k
of ij depend only on the network parameters and therefore, are constant. However, it should be
k
m
noted that for any particular line i-j, the factors ij
and ij
(for generation outage at bus m) are
different and therefore need to be pre-calculated separately. Once these factors are pre-calculated
and stored, the new values of line flow over any line can easily be estimated very quickly from
equation (5.2). If the new power flow over any line is found to be more than the corresponding limit,
then the operator can be alerted for taking an appropriate pre-emptive action.
In equation (5.2), it is assumed that the lost generation at bus k would be taken up by the
slack bus. However, it is also quite possible that the lost generation would be compensated by
all the remaining on-line generators combinedly, in which, each of the on-line generators would
take up some fraction of the lost generation in some particular ratio. One of the most frequently
used methods assumes that the on-line generators share the lost generation in proportion to their
maximum MW rating. Thus, the proportion of generation picked up by generation g is given by
227
g k,
gk =
Pgmax
(5.3)
Pamax
a=1
k
(5.4)
a=1
In equation (5.4) it is assumed that no remaining on-line generation hits the generation limit.
The line outage distribution factors are also defined similarly. The LOSF is defined by,
ij, mn =
fij
(5.5)
(0)
fmn
Where, ij, mn Line outage distribution factor for line i-j under outage of line m-n.
(0)
fmn
Power flow over line m-n in the pre-outage condition.
Therefore, for the outage of line m-n, the new flow over line i-j is given by,
(0)
fij(n) = fij(0) + ij, mn fmn
(5.6)
Again, as we will show later, the factors ij,mn are constant as they are dependent only on the line
parameters. Therefore, they would be pre-calculated and stored in the memory. As a result, for the
outage of any line m-n, the new power flows over all the other lines can be estimated very quickly.
5.2
We have already discussed the concepts of linear sensitivity factors. These factors are calculated
based on the concept of DC power flow and hence, let us first have a look at DC power flow
technique.
5.2.1
DC power flow
From FDLF method, we know, P = [B ] , where each elements of the matrix [B ] is negative
BUS bus matrix. Now, in DC power
of the imaginary parts of the corresponding elements of the Y
228
flow analysis, apart from using the above decoupled relation between P and , several other
simplifying assumptions are also taken as follows:
a. The system is lossless (i.e. line resistance is neglected) and therefore, each line is represented
by its reactance only.
b. The voltage of each bus is maintained at the rated voltage, i.e. 1.0 p.u.
c. For any line m-n, the angular difference between its terminal buses is quite small, and hence,
cos m cos n (as m n ) and sin(m n ) (m n ) rad (as m n 0) .
With these assumptions, the power flow over a line becomes,
Pij =
Vi Vj
1
sin(i j )
(i j ) (p.u)
xij
xij
(5.7)
In equation (5.7), the quantity xij denotes the reactance of the line i-j. From this equation it
is observed that the line power flow is basically a linear combination of the terminal bus voltage
angles. Moreover, the current flow over the line i-j is given by,
Vi Vj
Iij =
jxij
(Vi cos i Vj cos j ) + j(Vi sin i Vj sin j )
=
jxij
1
1
=
(sin i sin j )
(i j )
xij
xij
(5.8)
Equation (5.8) has been written under an added assumption that both the angles i and j are
individually quite small in magnitude. From equations (5.7) and (5.8) it is observed that in DC
power flow model, the expressions of line power flow and line current are same in per unit. We will
utilize this fact for computing LOSF in future.
In the next lecture, we will discuss a method for calculating GOSF.
229
5.2.2
Calculation of GOSF
1
X
3 32 X33 X3n P3
(5.9)
In equation (5.9), the matrix [X] is of size (n 1) (n 1) and any element Xij is actually
located in the location (i 1), (j 1). As the matrix [B ] is a constant real matrix dependent only
on the line parameters, matrix [X] is also a constant real matrix dependent on the line parameters.
Furthermore, in equation (5.9), the quantity 1 is not included in the vector as the reference
angle does not change with any outage in the system.
Now, to calculate the GOSF for the outage of generator at bus k, the perturbation at bus k
is set to Pk and the perturbation at the slack bus is set to Pk (assuming that the entire lost
generation is taken up by the slack bus) while keeping the perturbations at the other buses equal to
zero. Therefore, the perturbed values of the bus angles can be given as,
i = Xik Pk for i = 2, n
(5.10)
Now, from equation (5.7), the change in power flow over line i-j is given by,
Pij =
Therefore,
1
1
(i j ) = (Xik Xjk )Pk , where, xl = xij is the reactance of the line i-j.
xl
xl
ijk =
(5.11)
k
As can be seen from equation (5.11), the factor ij
is dependent only on the line parameters.
Let us now look at the application and utility of these sensitivity factors. For this purpose, let
as consider the IEEE-14 bus system (the data of which are given in Tables A.3 and A.4). In this
system, apart from the slack bus,there are two other generations at bus 2 and bus 6 (refer Table A.3).
The real power flows in all the lines have been calculated with all the three generatiors operating in
the system and are shown in column 2 of Table 5.1 (under the heading Pline(ori)). The GSOFs for
these two generators have been calculated using equation (5.11) and are shown is columns 3 and 7 of
Table 5.1 respectively (under the heading GSOF(2) and GSOF(6) respectively). From Table A.3 it
can be seen that the specified real power generated by these two generations are 18.3 MW and 11.2
MW respectively. Hence, following the argument given for equation (5.2), the quantities P2 and
P6 are equal to -0.183(p.u) and 0.112 (p.u) respectively (on a 100 MVA base) as shown in columns
4 and 8 of Table 5.1. With these information, the estimated line flows after the outage of generation
2 and 6 are calculated using equation (5.2) and are shown in columns 5 and 9 respectively. Lastly,
full AC power flow studies have been carried out by removing generator 2 and 6 one by one (by
modeling them as PQ bus after reducing their real power generation to zero) and the results of line
230
flows (obtained with full AC power flow) are shown in columns 6 and 10 respectively. From columns
5 and 6 as well as from columns 9 and 10 it is observed that the post-outage line-flows estimated by
the sensitivity analysis technique match quite closely with those obtained by the full AC power flow
method.
Table 5.1: Results of generator outage analysis in IEEE 14 bus system (all powers are given in p.u.)
For outage of generator at bus 2
Line Pline
GSOF
Pline
Pline
P2
no.
(ori)
(2)
(cal)
(ACLF)
1
1.6262 -0.9709 -0.183 1.803875 1.7891
2
0.7464 -0.1935 -0.183 0.781811 0.7776
3
0.8655 0.0089 -0.183 0.863871 0.8606
4
0.5314 0.0556 -0.183 0.521225 0.5207
5
0.3662 0.0821 -0.183 0.351176 0.3511
6
-0.358
0.047 -0.183 -0.3666
-0.3622
7
-0.7068 0.1064 -0.183 -0.72627 -0.7217
8
0.2689 -0.0013 -0.183 0.269138 0.2691
9
0.1063 -0.0004 -0.183 0.106373 0.1064
10
0.2893 -0.0078 -0.183 0.290727 0.2887
11
0.1156 -0.0059 -0.183 0.11668
0.1154
12
0.0852 -0.0007 -0.183 0.085328 0.0851
13
0.2005 -0.0031 -0.183 0.201067 0.2003
14
0.0025 -0.0065 -0.183 0.00369
0.0025
15
0.2664 0.0052 -0.183 0.265448 0.2667
16
0.0123 0.0055 -0.183 0.011294 0.0124
17
0.0655 0.0037 -0.183 0.064823 0.0657
18 -0.0777 0.0057 -0.183 -0.07874 -0.0776
19
0.0233 -0.0012 -0.183 0.02352
0.0232
20
0.0857 -0.0038 -0.183 0.086395 0.0854
For outage of
GSOF
P6
(6)
-0.7628 -0.112
-0.4344 -0.112
-0.1568 -0.112
-0.2928 -0.112
-0.2977 -0.112
-0.1203 -0.112
-0.0033 -0.112
-0.2562 -0.112
-0.1009 -0.112
-0.6729 -0.112
0.2459 -0.112
0.0277 -0.112
0.1312 -0.112
-0.0204 -0.112
-0.2358 -0.112
-0.2287 -0.112
-0.1548 -0.112
-0.2368 -0.112
0.05
-0.112
0.1574 -0.112
generator at bus 6
Pline
Pline
(cal)
(ACLF)
1.711634 1.7203
0.795053 0.7753
0.883062 0.8888
0.564194 0.5678
0.399542
0.395
-0.34453 -0.3357
-0.70643 -0.7189
0.297594 0.3173
0.117601 0.1255
0.364665 0.3309
0.088059 0.0748
0.082098 0.0778
0.185806 0.1782
0.004785 0.0022
0.29281
0.3151
0.037914 0.0512
0.082838 0.0944
-0.05118
-0.039
0.0177
0.016
0.068071 0.0567
To calculate the LOSF, we first need to understand the concept of Thevenin equivalent impedance
of a power system. Let as first have a look at that.
5.2.3
V
where I, V
and Y
are the vector of bus injection currents, vector of bus voltages
We know
I=Y
=Z
I where Z
is the bus impedance matrix and
and the admittance matrix respectively. Hence,V
=Y
1 . As the matrix Z
is a constant matrix, the relation V
=Z
I denotes a linear
is given by Z
and I and hence, for incremental charges in V
and I the same relation also
equation connecting V
231
= Z
I. Expanding this relation we get,
holds good, i.e. V
V1 Z11 Z12 Z1n I1
I
V Z
Z
2 21
2
22
2n
Vn Zn1 Zn2 Znn In
(5.12)
Now, suppose that there is an incremental charge is the current of bus k only while the incremental changes at the other buses are zero. Hence, Ik 0 and Ii = 0 for i = 1, n; k . Hence,
ik Ik for i = 1, n. Hence,
from equation (5.12), the changes in bus voltages are given by, Vi = Z
(0)
if the initial bus voltage is Vk , then the final voltage after perturbation is given by,
Vk = Vk(0) + Zkk Ik
(5.13)
Equation (5.13) can be represented as an equivalent circuit as shown in Fig. 5.1, which shows
the Thevenins equivalent circuit at bus k. From this circuit, the Thevenins equivalent impedance
kk .
of the system (looking from bus k) in equal to Z
232
5.3
Calculation of LOSF
In the last lecture, we have looked at the Thevenins impedance of the system looking from a
particular bus k. Let us now look at the Thevenins equivalent impedance of any system between
bus m and bus n. To achieve that let us assume that the incremental currents Im and In are
injected at bus m and n respectively (the incremental current at all the other buses are zero).
Consequently, from equation (5.12), the changes in the bus voltages (Vm and Vn ) can be given
as,
Vm = Zmm Im + Zmm In
(5.14)
Vn = Znm Im + Znn In
(5.15)
(5.16)
(5.17)
From equations (5.16) and (5.17), the equivalent circuit of the power system looking from bus
m and bus n can be drawn as shown in Fig. 5.2.
(0)
From Fig. 5.2 it is observed that the open-circuit voltage between bus m and bus n is Vm Vn
233
and the Thevenins equivalent impedance between bus m and bus n is given by,
(5.18)
From Fig. 5.2 it is also observed that the Thevenins equivalent impedance between bus m and
mm Znm + Znm ) = Zmm and that between bus n and the reference is (Znn Znm +
reference is (Z
Zmn ) = Znn which is in complete agreement with the results already obtained in equation (5.13).
b is connected between
Also, from Fig. 5.2 it can be readily seen that if a line having an impedance Z
bus m and n, then the current through this line is given by,
Vm(0) Vn(0)
Ib =
ZT h, mn + Zb
(5.19)
b is removed
Now, let as consider the case when a line or transformer having an impedance Z
from the system. Let us assume that this branch is removed between bus m and n. The removal
b between bus m and n. The
of this branch can be represented by addition of an impedance Z
situation is shown in Fig. 5.3.
b and this
Now, in Fig. 5.3, it can be seen that the current Ia flows in the added impedance Z
current is given by,
Vm(0) Vn(0)
Ia =
ZT h, mn Zb
(5.20)
This current Ia flows out of bus m and flows into bus n. Therefore, because of the addition
234
(5.21)
(5.22)
and
Ipq =
(5.23)
[
]
Ipq =
Zc
ZT h, mn Zb
(5.24)
Imn =
Zb
(5.25)
Or,
Ipq =
[
]
Zc
ZT h, mn Zb
Zb
(5.26)
(5.27)
Now, in equations (5.7) and (5.8) we have seen that under the assumptions of DC power flow
method, the expressions of power flow and current flow in a line are same in per unit. Therefore,
if we impose the same assumptions in equation (5.27), then all the impedances would be replaced
by the corresponding reactances and Ipq (Imn ) would be replaced by Ppq (Pmn ). Therefore, the
LOSF (pq, mn ) is given by,
pq, mn =
(5.28)
Again, as an example, IEEE 14 bus system is chosen and the estimated and calculated line-power
flows for two line outage cases are shown in Table 5.2. In this table, the pre-outage real power flows
235
in the line are shown in column 2. The values of LOSF are shown in columns 3 and 7. Using these
LOSF, the estimated values of power flow in the lines are tabulated in columns 5 and 9. Finally,
the actual power flow in the lines obtained with full AC power flow (after removing the line under
consideration) are shown in columns 6 and 10 respectively. From this table, it is observed that
the estimated and calculated values are reasonably close to each other, thereby establishing the
usefulness of the line outage sensitivity factors.
Table 5.2: Results of line outage analysis in IEEE 14 bus system (all powers are given in p.u.)
For outage of line no. 18
Line Pline
LSOF
Pline
Pline
P18
no.
(ori)
(2)
(cal)
(ACLF)
1
1.6262 -0.0363 -0.0777 1.6290205 1.6309
2
0.7464 0.0351 -0.0777 0.7436727
0.742
3
0.8655 -0.0294 -0.0777 0.8677844
0.869
4
0.5314 -0.0682 -0.0777 0.5366991 0.5364
5
0.3662 0.0574 -0.0777 0.36174
0.362
6
-0.358 -0.0363 -0.0777 -0.355179 -0.3545
7
-0.7068 0.5224 -0.0777 -0.747390 -0.7442
8
0.2689 -0.4868 -0.0777 0.3067244 0.3014
9
0.1063 -0.1923 -0.0777 0.1212417 0.1192
10
0.2893 0.6604 -0.0777 0.2379869 0.2433
11
0.1156 0.9961 -0.0777 0.038203
0.0351
12
0.0852 -0.0857 -0.0777 0.0918589 0.0925
13
0.2005 -0.325 -0.0777 0.2257525 0.2277
14
0.0025 -0.0125 -0.0777 0.0034713 0.0024
15
0.2664 -0.4743 -0.0777 0.3032531 0.2991
16
0.0123 -1.0039 -0.0777 0.090303
0.0904
17
0.0655 0.4085 -0.0777 0.0337596 0.0329
18 -0.0777
-0.0777
19
0.0233 -0.1021 -0.0777 0.0312332 0.0305
20
0.0857 -0.4102 -0.0777 0.1175725 0.1192
0.7464
2
Pline
(ACLF)
2.4451
0.9987
0.7986
0.7256
-0.2352
-0.3481
0.2815
0.1113
0.272
0.1046
0.0842
0.1952
0.0024
0.2791
0.0234
0.0719
-0.0666
0.0224
0.0794
So far, we have discussed the case of single contingency only. In the next lecture, we will start
looking into the procedure of analysis of multiple contingencies.
236
5.4
Let us now examine the case of outage of more than one elements. We will discuss here the case
of outage of two transmission lines. Although theoretically the outage of any number of lines can
be analyzed, but practically it is not feasible as outage of more than two lines would almost surely
overload the other existing lines and hence trip the whole system. We will also not discuss the case
of outage of two generations as outage of two generators would also most probably create a huge
deficit of power supply in the system, which would necessitate shedding of loads to maintain the
frequency of the grid within acceptable limits. Hence, we will confine ourselves to the case of outage
of two lines only. Towards this goal, let us first analyze the situation where two transmission lines are
simultaneously added to the power system, as the expressions derived would be directly applicable
to the case of our interest, i.e. outage of two lines.
Let us consider an N-bus power system in which I1 , I2 , IN are the bus injection currents and
V1 , V2 , VN are the bus voltages. The bus voltages and the bus injection currents are related by
and the relation is given as,
the bus impedance matrix Z
V1 Z11 Z12 Z1N I1
V Z
2 21 Z22 Z2N I2
=
VN ZN 1 ZN 2 ZN N IN
(5.29)
a 0 Ia
e V
Z
V
a Ia = Ve V and Zb Ib = Vg V . Hence, [
From Fig. 5.4, Z
] [ ] = [ f ].
f
h
0 Zb Ib
Vg Vh
237
Or,
Za
[
0
V
1
V2
V
e
e
f
g
h
0 Ia
0 0 1 1 0 0 0 0 0 Vf
][ ] = [
]
Zb Ib
0 0 0 0 0 1 0 1 0
V
g
Vh
VN
= AV
238
(5.30)
In equation (5.30), the matrix A can easily be identified. Now, from Fig. 5.4 it can be seen
a and Zb , extra currents Ia and Ib are being injected into buses f
that due to the addition of lines Z
and h respectively (which were not present before the addition of these two lines). Similarly, extra
currents Ia and Ib are being extracted from the buses e and g respectively (which were also not
present before the addition of these two lines). Therefore, it can be said that, due to the addition
of these two lines, injection currents at these four buses have changed by Ie , If , Ig and Ih
respectively, where Ie = Ia ; If = +Ia ; Ig = Ib and Ih = +Ib and due to these extra injection
to V
. Therefore, the addition of these two lines can
currents, the bus voltages have changed from V
be indirectly represented by these extra four injection currents which will produce the same effect
on the existing system. This situation is shown in Fig. 5.5.
Now, the vector of change in injection currents can be expressed as;
T
I = [00 Ie If Ig Ih 00] . Or,
0 0 0
0 0 0
I 1 0
Ia 1 0 Ia
[ ] = AT [Ia ]
I = =
Ib
Ib
Ib 0 1
Ib 0 1
0 0 0
(5.31)
= Z
I
Due to the above change in current
I, the bus voltages would undergo a change V
= V
+ Z
I = V
ZA
T [Ia ]
V
Ib
Ia
Ib
= AV
AZA
T [ ];
Or, AV
Or,
(5.32)
Za 0 Ia
AZA
T [Ia ] (from equation (5.30))
]
[
]
=
A
V
0 Zb Ib
Ib
Or, [
1
Ia
Za 0
T } AV
=Z
1
[ ] = {[
] + AZA
1 AV
Ib
0 Zb
(5.33)
Where,
1 = [Za 0 ] + AZA
T
Z
0 Zb
239
(5.34)
(5.35)
ZT h,ef + Za
(Zeg Zeh ) (Zf g Zf h )
1 = [
Z
]
(Zge Zgf ) (Zhe Zhf )
ZT h,gh + Zb
(5.36)
Or,
T h,gh and ZT h,gf denote the thevenin equivalent impedance of the system as observed
Where, Z
from terminals e-f and g-h respectively (please see equation (5.18)). Reproducing equation (5.33)
we have,
Ia
1 Ve Vf
1
[] = Z
A
V
=
Z
[
]
1
1
Ib
Vg Vh
(5.37)
From equations (5.36) and (5.37), the vector [Ia Ib ]T is calculated and subsequently, from
can be calculated.
equation (5.32), the new bus voltage vector V
We are now ready to analyze the effect of outages of two lines, which we will do in the next
lecture.
240
5.5
(5.38)
a by Za and Zb by Zb respectively
Note that equation (5.38) has been obtained by replacing Z
in equation (5.36). Hence, from equation (5.37) we have,
ZT h, ef Za
(Zeg Zef ) (Zf g Zf h ) Ia
Ve Vf
]
]
=
[
[
]
[
Vg Vh
Ib
(Zge Zgf ) (Zhe Zhf )
ZT h, gh Zb
(5.39)
V V
e
f
(
Z
Z
)
(
Z
Z
)
eg
ef
fg
fh
1
Ia ZT h, ef Za
ZT h, ef Za
(Z Z ) (Z Z )
ge
gf
he
hf
Ib Vg Vh
ZT h, gh Zb
ZT h, gh Zb
(5.40)
(5.41)
(5.42)
241
(say)
Za
Lef, gh
Zb
V V
e
f
Zb
Lgh, ef Ia ZT h, ef Za
Za
I Vg Vh
1
ZT h, gh Zb
Or,
a
1
1L
gh, ef Lef, gh Za
Ib
Lef, gh
Zb
V V
e
f
Zb
Lgh, ef ZT h, ef Za
Za
Vg Vh
1
ZT h, gh Zb
(5.43)
(5.44)
and Igh =
Vg Vh
Zb
(5.45)
Ia =
Ib =
1
1 Lgh, ef Lef, gh
1
1 Lgh, ef Lef, gh
Z 2 Lgh, ef
Za
Ief + b
Igh ]
ZT h, ef Za
Za ZT h, gh Zb
(5.46)
Zb
Za2 Lef, gh
Ief +
Igh ]
Zb ZT h, ef Za
ZT h, gh Zb
(5.47)
c ) is given by,
Now, the change in current in line i-j (having an impedance Z
Iij =
Vi Vj
Zc
(5.48)
Now,
Thus,
Iij =
(5.49)
(5.50)
(5.51)
Iij =
1
1 Lgh, ef Lef, gh
1 Ief + X
2 Igh ]
[X
(5.52)
Where,
2
Lef, gh
1 = Za (Zif Zie ) (Zjf Zje ) + Za
[(Zih Zig ) (Zjh Zjg )]
X
Zc
ZT h, ef Za
Zb Zc ZT h, ef Za
(5.53)
Now, we recognize,
Lij, ef =
(5.54)
Hence,
Lef, gh
Za2
(5.55)
Now, expanding the second term of equation (5.55), we have [using equation (5.41)],
Lef, gh
Za2
[(Zih Zig ) (Zjh Zjg )]
Zb Zc ZT h, ef Za
Z 2 Zb (Zeh Zeg ) (Zf h Zf g ) (Zih Zig ) (Zjh Zjg )
= a
Zb Zc Za
ZT h,gh Zb
ZT h,ef Za
=[
Or,
Lef, gh
Za2
[(Zih Zig ) (Zjh Zjg )] = Lij, gh Lgh, ef
Zb Zc ZT h, ef Za
(5.56)
Equation (5.56) has been written using equation (5.42) and by noting that,
Lij, gh =
(5.57)
Hence,
(5.58)
2
Lgh, ef
Zb (Zih Zig ) (Zjh Zjg )
2 = Zb
[(Zif Zie ) (Zjf Zje )] +
X
Za Zc ZT h, gh Zb
Zc
ZT h, gh Zb
Again,
Lgh, ef
Zb2
[(Zif Zie ) (Zjf Zje )]
Za Zc ZT h, gh Zb
243
(5.59)
Or,
Lgh, ef
Zb2
[(Zif Zie ) (Zjf Zje )] = Lij, ef Lef, gh
Za Zc ZT h, gh Zb
(5.60)
Please note that equation (5.60) has been obtained by using equations (5.41) and (5.54). Hence,
(5.61)
Iij =
1
1 Lef, gh Lgh, ef
(5.62)
Equation (5.62) gives the change in the current of line i-j in terms of the original currents of
the lines which have gone out of service. Now, in equation (5.62) all the factors are in terms of
complex impedances. However, imposition of the conditions of the DC power flow makes all the
factors being represented by the corresponding reactances (i.e. all complex impedances are replaced
by their corresponding reactances). Now, from equation (5.28) we see that when these functions are
represented by reactances only, they actually represent the different line outage sensitivity factors
(LOSF). As already noted earlier, there LOSFs are already pre-calculated and stored. Also, as
discussed earlier, under the assumption of DC power flow, for contingency calculation, the current
flow quantities are replaced by corresponding power flow quantities. Therefore, the change in power
flow in line i-j can very easily be calculated from equation (5.62) using the information of pre-outage
power flow in lines e-f and g-h.
In the next lecture, we will look into an example of contingency analysis for outages of two lines.
244
5.5.1
As an example, let us consider outage of two lines in the IEEE-14 bus system. Towards this goal,
let us assume that line no. 3 (let it be considered as line e-f) and 17 (let it be considered as line
g-h) have been taken out of the system. Subsequently, for any line i-j (in the set of lines which
are still remaining in the system), the factors LOSFM1 and LOSFM2 are calculated as follows.
LOSF M 1 =
(5.63)
LOSF M 2 =
(5.64)
Now, under the assumption of DC power flow, for contingency calculation, the impedances are
replaced by corresponding reactances. Therefore, for calculating the quantities Lef, gh , Lgh, ef , Lij, ef
and Lij, gh , the expressions given in equations (5.41), (5.42), (5.54) and (5.57) have been used respectively, with the impedances replaced by the corresponding reactances. After obtaining the values of
these four quantities, the factors LOSFM1 and LOSFM2 have been calculated from equations (5.63)
and (5.64). The calculated values of these two factors for all the lines remaining in the system are
shown in columns 5 and 6 of Table 5.3. Please note that for lines 3 and 17, thes two factors have
no relevance and therefore, these values have been indicated as zero in this table for these two lines.
After obtaining these two factors, the change in power flow in line i-j are calculated from equation
(5.62) using the information of pre-outage power flow in line no. 3 and 17. For ready reference, the
pre-outage power flows of line 3 and 17 are given in columns 3 and 4 of Table 5.3 respectively. After
calculating the change in power flow of any line i-j, its post-outage power flow is obtained by adding
its pre-outage power flow to the change in power flow. Again, for ready reference, the pre-outage line
power flows for all the lines are given in column 2 of Table 5.3. The calculated post-outage power
flows in all the lines are given in column 7 of this table. Please note that, as line no. 3 and 17 are now
out of the circuit, the post-outage power flows in these two lines are zero. Finally, for the purpose
of comparison, complete AC power flow solution of the IEEE-14 bus system has been obtained after
removing line no. 3 and 17 from the system. The line power flows obtained through AC power flow
study are shown in the last column of Table 5.3. From the results given in last two columns it is
observed that there is some difference between the power flows calculated by complete AC power
flow analysis and contingency analysis. Moreover, comparison of Tables 5.2 and 5.3 shows that this
difference is more for double line contingencies as compared to that for single line contingency.
5.6
From the results of the Tables 5.1 and 5.2 it can be observed that the sensitivity factors give
reasonably close estimates of real power flows in the lines in the event of outage of a generator
or a line. However, the sensitivity factors give the estimate of only the real power flows over the
lines. On the other hand, in several situations, it is also equally important to consider the bus
245
Table 5.3: Contingency calculation for outage of line no. 3 and 17 (all powers are given in p.u.)
Line Pline
P3
P17
LOSFM1
no.
(ori)
1
1.6262 0.8655 0.0655 -0.2053
2
0.7464 0.8655 0.0655
0.2177
3
0.8655 0.8655 0.0655
0
4
0.5314 0.8655 0.0655
0.4758
5
0.3662 0.8655 0.0655
0.3491
6
-0.358 0.8655 0.0655
-1.072
7
-0.7068 0.8655 0.0655 -0.5508
8
0.2689 0.8655 0.0655 -0.0274
9
0.1063 0.8655 0.0655 -0.0107
10
0.2893 0.8655 0.0655
0.0238
11
0.1156 0.8655 0.0655
0.0237
12
0.0852 0.8655 0.0655
0
13
0.2005 0.8655 0.0655
0
14
0.0025 0.8655 0.0655 -0.0097
15
0.2664 0.8655 0.0655 -0.0177
16
0.0123 0.8655 0.0655 -0.0237
17
0.0655 0.8655 0.0655
0
18 -0.0777 0.8655 0.0655 -0.0237
19
0.0233 0.8655 0.0655
0
20
0.0857 0.8655 0.0655
0
LOSFM2
-0.025
0.0237
0
-0.0678
0.0389
-0.004
0.4444
-0.3998
-0.1579
0.5427
-0.5152
0.2174
0.7805
-0.01
-0.3898
0.507
0
0.5111
0.2304
0.9991
Pline
Pline
(cal)
(ACLF)
1.44687535
1.5449
0.9363717
0.9864
0
0
0.938764
0.9803
0.670894
0.7059
-1.286078
-1.1902
-1.1544092 -1.1585
0.2189984
0.218
0.0866967
0.0862
0.34544575
0.3668
0.10236675
0.1236
0.0994397
0.0994
0.25162275
0.2558
-0.00655035 0.0023
0.22554875
0.2157
0.02499615
0.0069
0
0
-0.0647353 -0.0833
0.0383912
0.0373
0.15114105
0.1533
voltage variations as well as the reactive power flows over the lines in the event of any outage. In
these situations, the full AC power flow analysis needs to be carried out as the sensitivity factors are
not able to estimate the changes in bus voltage and the reactive power flows in the lines. However,
the full AC power flow analysis is considerably slower than the sensitivity based methods and these
are not suitable for analyzing thousands of potential outage cases within the time frame required by
on-line contingency analysis.
Thus, we have a contradicting situation here. On one hand, for fast evaluation of contingencies,
sensitivity based methods need to be used whereas, for accurate estimation of the effects of any
outage, slower, full AC power flow analysis needs to be carried out. To break this dilemma, a middle
path is followed. Initially, all the outage studies are carried out using the sensitivity factors. Based on
the results of the sensitivity analysis, all the outage cases are ranked according to a suitably chosen
performance index (PI). Once the outage cases are ranked and sequentially arranged in decreasing
order of the performance index, the top few outage cases are analyzed further in detail using the AC
power flow analysis.
Therefore, for contingency ranking, the choice of the performance index is very important. The
PI should be such that it should satisfy the following criteria.
a. It should adequately reflect the severity of any particular contingency.
b. The final list of contingencies for which full AC power flow analysis is to be carried out, should
246
5.6.1
MW ranking methods
P I = Wj [
j=i
Pj
Pj max
(5.65)
In equation (5.65), L is the number of lines in the system, Pj and Pj max are MW flow and MW
capacity of the line j respectively and n is a suitable index. However, the PI given in equation
(5.65) is prone to masking phenomenon in which a contingency causing many lines to be heavily
loaded with no lines being overloaded is ranked higher than a contingency causing few lines to be
overloaded with the remaining lines being lightly loaded. However, this masking phenomenon can be
removed if the summation in equation (5.65) is taken over only the set of overloaded lines as shown
in equation (5.66).
P I = Wj [
jSL
Pj
Pj max
(5.66)
P I = Hd1 + Wj [
n1
jSL
n2
Pj
Pj max
(5.67)
In equation (5.67), Hd1 is the change in power flow in the highest overloaded line while n1 and
n2 are suitable indices. In case the highest overload in two cases of contingency are same, then the
effect of second highest overloaded line is also taken into account as shown in equation (5.68).
P I = Hd1 + Hd2 + Wj [
n1
n2
jSL
Pj
Pj max
n3
(5.68)
In equation (5.68), Hd2 denotes the change in power flow in the second highest overloaded
247
line. Similarly, if both the highest and second highest overloading conditions are same for two
contingencies, then the third highest overloaded line is taken into account separately.
n2
n3
jSL
Pj
Pj max
n4
(5.69)
As before,Hd3 denotes the change in power flow in third highest overloaded line. Of course, this
same philosophy can be extended further to include more number of lines for the calculation of PI,
if necessary.
5.6.2
These are several PIs suggested for properly ranking the voltage /reactive power contingencies. Some
of them are:
i)
2
i Vi
]
P Iv = [
lim
i=1 2 Vi
N
Where,
(5.70)
1
Vi = Vi Visp ;
Vilim = (Vimax Vimin ) ;
2
Vi = post-contingency voltage magnitude of bus i;
Visp = Nominal or specified voltage of bus i;
Vimax , Vimin = Maximum and minimum limit of voltage magnitudes of bus i
i = User selected weighting factor
P Iv = Performance index corresponding to voltage security
N = number of buses in the system
ii)
Vi Vilim
P Iv = Wvi
Vilim
iS1
(5.71)
where,
(5.72)
Wvi in the weighting factor for bus i and S1 in the set of all buses at which the voltage limits
have been violated.
iii)
P Iv = Wvi (Vi )2
iS1
where,
Vi =
Vinom
1;
Vilim
Vinom = Vi Vinom ;
248
(5.73)
1
Vinom = (Vimax + Vimin );
2
1
Vilim = (Vimax Vimin )
2
iv)
2
N
dmax
dmin
i
P Iv = [ max ] + [ imin ]
i=1 ai
i=1 ai
N
(5.74)
Where,
Vi V amax
if Vi > Viamax
Vinom
= 0 otherwise
(5.75)
Vimin Vi
if Vi < Viamin
nom
Vi
= 0 otherwise
(5.76)
dmax
=
i
dmin
=
i
amax
=
i
1
V
nom
i
(Vimax Viamax );
amin
=
i
1
Vinom
(Viamin Vimin )
Viamax and Viamin are the higher and lower volatage alarm limits for bus i.
v)
N o
nom
V
+
V
i
i
P Iv = i max
min
V
i=1
i
i
(5.77)
m
Vi
]
P Iv = [ Wi
Vimax
i=1
(5.78)
The value of m is taken to be very large ( 20) to avoid the masking effect.
One popular way of combining the above discussed real power and reactive power ranking method
is to use the 1P1Q method. In this method, a decoupled power flow is used and after one iteration,
(one P-Q computation and one Q-V computation), the bus voltages are noted and with these bus
voltages, the line power flows are calculated. It generally appears that there is sufficient information
available (in the bus voltages) to arrive at a reasonable values of the performance indices. After
calculating the real power and reactive power performance indices separately, the combined PI is
calculated by adding the appropriate real power and reactive power indices together.
With these descriptions of contingency ranking methods, we conclude our discussion of contingency analysis. From the next lecture, we will start our discussion of stability analysis.
249
Module 6
Power system stability
6.1
Introduction
In general terms, power system stability refers to that property of the power system which enables
the system to maintain an equilibrium operating point under normal conditions and to attain a
state of equilibrium after being subjected to a disturbance. As primarily synchronous generators
are used for generating power in grid, power system stability is generally implied by the ability
of the synchronous generators to remain in synchronism or in step. On the other hand, if the
synchronous generators loose synchronism after a disturbance, then the system is called unstable.
The basic concept of synchronism can be explained as follows.
In the normal equilibrium condition, all the synchronous generators run at a constant speed and
the difference between the rotor angles of any two generators is constant. Under any disturbance, the
speed of the machines will deviate from the steady state values due to mismatch between mechanical
and electrical powers (torque) and therefore, the difference of the rotor angles would also change.
If these rotor angle differences (between any pair of generators) attain steady state values (not
necessarily the same as in the pre-disturbance condition) after some finite time, then the synchronous
generators are said to be in synchronism. On the other hand, if the rotor angle differences keep on
increasing indefinitely, then the machines are considered to have lost synchronism. Under this out
of step condition, the output power, voltage etc. of the generator continuously drift away from the
corresponding pre-disturbance values until the protection system trips the machine.
The above phenomenon of instability is essentially related with the instability of the rotor angles
and hence, this form of instability is termed as rotor angle instability. Now, as discussed above,
this instability is triggered by the occurrence of a disturbance. Depending on the severity of the
disturbance, the rotor angle instability can be classified into two categories:
Small signal instability: In this case, the disturbance occurring in the system is small. Such
kind of small disturbances always take place in the system due to random variations of the loads
and the generation. It will be shown later in this chapter that under small perturbation (or
disturbance), the change in the electrical torque of a synchronous generator can be resolved into
two components, namely, a) synchronizing torque (Ts ) - which is proportional to the change in
the rotor angle and b) damping torque (Td ), which is proportional to the change in the speed
250
of the machine. As a result, depending on the amounts of synchronizing and damping torques,
small signal instability can manifest itself in two forms. When there is insufficient amount of
synchronizing torque, the rotor angle increases steadily. On the other hand, for inadequate
amount of damping torque, the rotor angle undergoes oscillations with increasing amplitude.
These two phenomena are illustrated in Fig. 6.1. In Fig. 6.1(a), both the synchronizing
and damping torques are positive and sufficient and hence, the rotor angle comes back to a
steady state value after undergoing oscillations with decreasing magnitude. In Fig. 6.1(b), the
synchronizing torque is negative while the damping toque is positive and thus, the rotor angle
envelope is increasing monotonically. Fig. 6.1(c) depicts the classic oscillatory instability in
which Ts is positive while Td is negative.
a. Local mode: In this type, the units within a generating station oscillate with respect to
the rest of the system. The term local is used because the oscillations are localized in a
particular generating station.
b. Inter-area mode: In this case, the generators in one part of the system oscillate with
respect to the machines in another part of the system.
c. Control mode: This type of instability is excited due to poorly damped control systems
such as exciter, speed governor, static var compensators, HVDC converters etc.
d. Torsional mode: This type is associated with the rotating turbine-governor shaft. This
type is more prominent in a series compensated transmission system in which the mechanical system resonates with the electrical system.
Transient instability: In this case, the disturbance on the system is quite severe and sudden
and the machine is unable to maintain synchronism under the impact of this disturbance. In
this case, there is a large excursion of the rotor angle (even if the generator is transiently stable).
Fig. 6.2 shows various cases of stable and unstable behavior of the generator. In case 1, under
the influence of the fault, the generator rotor angle increases to a maximum, subsequently
decreases and settles to a steady state value following oscillations with decreasing magnitude.
In case 2, the rotor angle decreases after attaining a maximum value. However, subsequently,
it undergoes oscillations with increasing amplitude. This type of instability is not caused by
the lack of synchronizing torque; rather it occurs due to lack of sufficient damping torque in the
post fault system condition. In case 3, the rotor angle monotonically keeps on increasing due
to insufficient synchronizing torque till the protective relay trips it. This type of instability, in
which the rotor angle never decreases, is termed as first swing instability.
Apart from rotor angle instability, instability can also occur even when the synchronous generators
are maintaining synchronism. For example, when a synchronous generator is supplying power to an
induction motor load over a transmission line, the voltage at the load terminal can progressively
252
reduce under some conditions of real and reactive power drawn by the load. In this case, loss of
synchronism is not an issue but the challenge is to maintain a stable voltage. This type of instability
is termed as voltage instability or voltage collapse. We will discuss about the voltage instability issue
later in this course.
Now, for analysing rotor angle stability, we have to first understand the basic equation of motion
of a synchronous machine, which is our next topic.
6.2
The equation of motion of a synchronous generator is based on the fact that the accelerating torque
is the product of inertia and its angular acceleration. In the MKS system, this equation can be
written as,
d2 m
= Ta = Tm Te
dt2
(6.1)
In equation (6.1),
t Time in seconds
Ta The net accelerating torque, in N-m
Tm The mechanical or shaft torque supplied by the prime mover less retarding torque due
to rotational losses, in N-m
dm
dm
= sm +
dt
dt
or,
253
dm dm
=
sm
dt
dt
(6.3)
d2 m d2 m
=
(6.4)
dt2
dt2
dm
represents the deviation of the actual rotor speed
Equation (6.3) shows that the quantity
dt
from the synchronous speed in mechanical radian per second. Substituting equation (6.4) into
equation (6.1) one gets,
d2 m
= Ta = Tm Te
dt2
(6.5)
Jm
dm
. From equation (6.5) we get,
dt
d2 m
= m Ta = m Tm m Te
dt2
Or,
Jm
d2 m
= Pa = Pm Pe
dt2
(6.6)
In equation (6.6), Pa , Pe and Pm denote the accelerating power, electrical output power and the
input mechanical power (less than the rotational power loss) respectively.
The quantity Jm is the angular momentum of the rotor and at synchronous speed, it is known
as the inertia constant and is denoted by M . Strictly, the quantity Jm is not constant at all
operating conditions since m keeps on varying. However, when the machine is stable, m does not
differ significantly from sm and hence, Jm can be taken approximately equal to M . Hence, from
equation (6.6) we obtain,
d2 m
= Pa = Pm Pe
dt2
(6.7)
Now, in machine data, another constant related to inertia, namely H-constant is often encountered. This is defined as;
H=
Or,
1 Jsm 2 1 M sm
H=
=
2 Smc
2 Smc
MJ/MVA =
1 M sm
2 Smc
sec.
(6.8)
In equation (6.8), the quantity Smc is the three phase MVA rating of the synchronous machine.
Now, from equation (6.8),
M=
2HSmc
sm
MJ/mech. rad
(6.9)
2H d2 m Pa Pm Pe
=
=
sm dt2
Smc
Smc
(6.10)
In equation (6.10), both m and sm are in mechanical units. Now, the corresponding quantities
254
s =
P
sm ;
2
P
m ;
2
(6.11)
In equation (6.11), P is the number of pole in the generator, s is the synchronous speed of the
machine in electrical radian/sec. and (in electrical radian) is the angular displacement of the rotor
from the synchronously rotating reference axis. Substituting equation (6.11) in equation (6.10) we
get,
2H d2
= Pa = Pm Pe per unit
s dt2
(6.12)
Equation (6.12) is known as the swing equation of the synchronous machine. As this is a
second order differential equation, it can be written as a set of two first order differential equations
as below.
2H d
= Pm Pe per unit
s dt
(6.13)
d
= s
dt
(6.14)
In equations (6.13) - (6.14), the quantity is the speed of the synchronous machine and is
expressed in electrical radian per second. Now, in the above two equations, no damping of the
machine is considered. If damping is considered (which opposes the motion of the machine), a term
proportional to the deviation of the speed (from the synchronous speed) is introduced in equation
(6.13). Therefore, the modified equation becomes;
2H d
= Pm Pe d( s ) per unit
s dt
(6.15)
In equation (6.15), d is called the damping co-efficient. However, in the presence of damping, equation
(6.14) does not change. Therefore, in the presence of damping, this pair of equations ((6.14) and
(6.15)) describe the motion of the synchronous machine.
With this introduction of motion of synchronous machine, we are now ready to address the various
stability issues. From the next lecture we will start with transient stability analysis.
255
6.3
Before solving the differential equations to determine the transient stability or instability of the
system, it is necessary to compute the initial conditions.
6.3.1
Let us consider an n bus power system with m generators (m < n). Without loss of generality, it
is assumed that the m generators are located at first m buses of the system. Towards computation
of initial conditions, initially the load flow solution of the system is computed. From the load flow
solution, following quantities are availbale:
a. Vi i
for i = 1, 2, n
b. PLi , QLi
for i = m + 1, m + 2, n
for i = 1, 2, m
In the above, PLi and QLi denote the real and reactive power load at bus i respectively. Similarly,
PGi and QGi denote the real and reactive power generation at bus i respectively. Further, the
quantities Vi and i denote the voltage magnitude and angle of it h bus respectively. With these
information, following calculations are carried out:
(i) At any bus i, the loads are converted to equivalent admittance as;
yLi =
PLi jQLi
Vi 2
for i = 1, 2, n
(6.16)
for i = 1, 2, n
(6.17)
old
BUS
BUS matrix of the system used in load flow calculation.
where, Y
is the original Y
(iii) At the generator buses, the generators are represented as equivalent voltage sources behind
the direct axis transient reactances as,
PGi jQGi
Ei = Vi + jxdi Ii = jxdi
= Ei i
Vi
for i = 1, 2, m
(6.18)
In equation (6.18), the quantity xdi denotes the direct axis transient reactance of the ith machine.
While performing the transient stability analysis, the magnitude Ei is held constant. The equivalent
diagram of the n bus power system is shown in Fig. 6.3.
With the initial conditions computed as above, we are now ready to solve the transient stability
problem. Basically, the transient stability problem is solved by two techniques: i) partition explicit
(PE) method and ii) simultaneous implicit (SI) method. In the PE method, the network algebraic
equations and the generator differential equations are solved separately. In the SI method, these
256
algebraic and differential equations are solved together. In this course, however, we are going to
discuss PE method only. Before, discussing the PE method, it is necessary to describe the network
algebraic equations.
6.3.2
IBUS = Y
BUS V
BUS
(6.19)
(6.20)
vk = jxdiik
ik = yk vk
(6.21)
In equation (6.21), ik and vk are the current through and voltage across the generator reactance
1
= j/xdi . Now, at the generator terminals, Ik = ik = yk vk ; or, Ik =
jxdi
yk (Ek Vk ) as (vk = Ek Vk ). Or,
Ik + yk Vk = yk Ek
(6.22)
respectively and yk =
(Y11 + y1 ) V1
Y21 V1
Ym1 V1
Y(m+1),1 V1
Yn1 V1
+
Y12 V2
+ (Y22 + y2 ) V2
+
+
Ym2 V2
+ Y(m+1),2 V2
+
+
Yn2 V2
+
+
+
+
+
+
+
+
Y1m Vm
+
Y2m Vm
+
+ (Ymm + ym ) Vm
+
+
+
+
+
+
+
+
+
Y1n Vn
+
Y2n Vn
+
+
Ymn Vn
+ Y(m+1),n Vn
+
+
Ynn Vn
= y1 E1
= y2 E2
=
= ym Em
=
0
=
=
0
(6.23)
1 ,
From equation (6.23), the voltages V1 , V2 Vn can be solved, for known values of E
E2 En . With these known terminal voltages, the electrical power output of each generator
can be calculated as;
for i = 1, 2, m
Pei = Re (EI Ii )
(6.24)
Where,
Ii = yi (Ei Vi )
for i = 1, 2, m
(6.25)
With the above equations, we are now in a position to discuss the PE method.
6.3.3
In the PE method, the numerical integration of the generator differential equations are carried
out separately from the solution of the network algebraic equations. For numerical integration of
differential equations, the total simulation time (tT ) is divided into N intervals, each interval being
tT
i obtained above, solve the equation set (6.23) to obtain the terminal
2. With the values of E
voltages at all the buses (Vi , for i = 1, 2, n). Subsequently, the electrical power output of all
the generators (Pei , for i = 1, 2, m) are computed from equations (6.24) - (6.25).
258
3. Under steady state condition, mechanical power input to each generator is equal to its
generator electrical output power (neglecting losses). Therefore, Pmi = Pei , for i = 1, 2, m.
Also, under steady state, all the generators are assumed to operate at synchronous speed (s ).
4. Thus, after the above three steps, at t = 0, the variables pertaining to generators (Ei , i ,
i , Pmi for i = 1, 2, m) and the network bus voltages (Vi , for i = 1, 2, n) are all known.
5. The simulation process advances to t = t. At this instant, first the network equations
given in the equation set (6.23) are solved to compute the bus voltages and subsequently, the output
electrical power of each generator is calculated by using equations (6.24) - (6.25). Now, if the
steady state condition is still maintained, i.e. if there is no change in the network (as compared
to the network condition at t = 0), then the calculated values of Pei would be again equal to the
corresponding value of Pmi .
6. With the solution of the network equations at hand, we should now solve the solve the
generator differential equations for calculating the values of i and i at t = t. Towards that end,
let us first re-write the swing equations of ith machine for convenience in equations (6.26) - (6.27)
below.
2Hi di
= Pmi Pei
(6.26)
s dt
di
= i s
(6.27)
dt
di
Now, from equation (6.26), as Pmi = Pei ,
= 0. In other words, there is no change in the
dt
di
speed of the generator and hence i = s . As a result, from equation (6.27),
= 0 and hence, the
dt
angle i would also be maintained at the value calculated at t = 0. Therefore, under steady state
condition, at t = t, both i and i would be maintained at the values calculated at t = 0.
7. Assume that the steady state condition continues from t = 0 to t = (k 1)t, where, k is
a positive integer. Following the arguments described at steps 5 and 6, it can be easily seen that at
the end of t = (k 1)t sec., both i and i would still be maintained at the values calculated at
t = 0.
8. Now, let us assume that a three phase to ground short circuit fault occurs in the system
at t = to = kt at the `th bus. To accommodate this fault condition in the network equations,
the element Y`` is increased manyfold to reflect very high admittance from bus ` to ground. With
this imposed condition, the network bus voltages are calculated from equation set (6.23). Subsequently, the output electrical power of each generator is calculated by using equations (6.24) - (6.25)
corresponding to time t = to .
9. With the values of Pei calculated in step 8, the swing equations (6.26) - (6.27) are integrated
to obtain the values of i and i at t = to . Now, for integrating the swing equations, initial values of
i and i are required. These initial values are taken to be equal to the values of i and i obtained
at the end of t = (k 1)t. For brevity, the value of i (i = 1, 2, m) calculated at t = to is
denoted as i (to ). With this value of i (to ), the voltage behind the transient reactance is updated
i = Ei i (to ), for i = 1, 2, m, where the magnitude Ei has been taken to be equal to the
as E
constant value calculated at step 1.
259
10. The simulation advances to t = to + t. If the fault still persists, the values of Pei are
i in equation set (6.23)
calculated as described in step 8. However, for this purpose, the values of E
are taken to be equal to the latest values calculated at the end of t = to sec.
11. With the values of Pei calculated at t = to + t, step 9 is repeated to update the variables
i , i and Ei at the end of t = to + t. Again, please note that for integrating the differential
equations, the latest values of i and i (calculated at t = to ) have been taken as the initial values.
12. Steps 10 and 11 are repeated to update the variables at t = to + 2t, to + 3t, to +
4t, till the fault clears.
13. At t = tcl (tcl = pt, p being a positive integer and p > k ), the fault clears. This condition
is imposed in the network equations by restoring Y`` to its original pre-fault value and subsequently,
steps 8 - 12 are repeated to obtain the variations of i and i . By observing the variation of i , the
stability of the system is assessed.
In step 9, the generator differential equations are numerically integrated. In the next lecture, we
will look into two such numerical integration techniques, namely, i) modified Eulers method and ii)
4th order Runga-Kutta technique.
260
6.3.4
Before discussing the application of Eulers method for solving the swing equations, let us first review
the basic Eulers method of numerical integration. Let the general from of a differential equation is
given by;
dy
= f (x, y);
dx
y(xo ) = yo ;
(6.28)
In equation (6.28), x and y are independent and dependent quantities respectively and xo and yo
are initial values of x and y respectively. For the purpose of numerical integration, the independent
axis (x-axis) is divided into intervals of length h such that discrete points on the independent axis
are xo , xo + h, xo + 2h, etc. As indicated in equation (6.28), the value of y at x = xo is yo .
The task is to calculate the values y1 , y2 , corresponding to the x co-ordinates xo + h, xo + 2h,
respectively. Once these values are obtained, the smooth curve representing the solution of the
differential equation given in equation (6.28) can be plotted.
In the modified Eulers method, the values y1 , y2 , are calculated in two steps:
Predictor
(1)
y1(1) = yo + h
dy
= yo + hf (xo , yo )
dx x=xo
Corrector
(1)
dy
= f (xo + h, y1(1) )
dx x=xo +h
b) With this updated value of
(6.29)
dy
at x = xo + h as;
dx
(6.30)
dy
at x = x1 = xo + h, the final value of y1 is calculated as;
dx
dy
h dy
[
+
]
2 dx x=xo dx x=xo +h
h
= yo + [f (xo , yo ) + f (xo + h, y1(1) )]
2
y1 = yo +
or,
y1
(6.31)
With this final value of y1 obtained at x = xo + h = x1 , the above two steps are repeated to
calculate y2 at x = xo + 2h = x2 and subsequently, this process is repeated to obtain the complete
solution of the differential equation.
Now, for our application, let us note that the independent axis (x-axis) denotes time. Therefore,
as already discussed, for solving the differential equations, the time axis is divided into intervals of duration t sec. (i.e. h = t). Further, let us also assume that the values of i and i (i = 1, 2, m)
have already been obtained at t = to and these values are denoted as io and io respectively. More(o)
over, the initial values of Pei (denoted as Pei ) are also assumed to be calculated utilising the values
261
of io . With thse known values, the values of i and i at t = to + t are calculated as follows.
Predictor step
(1)
di
= io + t(io s )
dt t=to
(6.32)
di
s
(Pmi Pei(o) )
= io + t
dt t=to
2Hi
(6.33)
i(1) = io + t
i(1) = io + t
(1)
and i
Corrector step
(1)
With the new values of i (i = 1, 2, m) obtained in the predictor step, the values of Pei
(i = 1, 2, m) are updated using equations (6.23)-(6.25) (after appropriately incorporating the
(1)
network conditions in the equation set (6.23)). Let these updates values of Pei be denoted as Pei .
Thereafter, the derivatives at the end of the present time step are calculated as follows:
di
= i(1) s
dt t=to +t
(6.34)
di
s
=
(Pmi Pei(1) )
dt t=to +t 2Hi
(6.35)
With the above new derivative values obtained, the final values of i and i at t = to +t (denoted
as i1 and i1 respectively) are calculated as;
i1 = io +
t di
di
[ +
]
2 dt t=to dt t=to +t
(6.36)
i1 = io +
t di
di
[
+
]
2 dt t=to dt t=to +t
(6.37)
Proceeding further, for calculating i and i at t = to +2t, the quantities io and io are replaced
by i1 and i1 respectively and equations (6.32)-(6.37) are followed again.
6.3.5
Let us again consider the same general form of a differential equation as in equation (6.28):
dy
= f (x, y);
dx
y(xo ) = yo ;
(6.38)
Again, the meanings of different notations used in equation (6.38) are same as those in equation
262
h
y1 = yo + (k1 + 2k2 + 2k3 + k4 )
6
(6.39)
k1 = hf (xo , yo )
(6.40)
In equation (6.39),
k1
h
k2 = hf (xo + , yo + )
2
2
h
k2
k3 = hf (xo + , yo + )
2
2
k4 = hf (xo + h, yo + k3 )
(6.41)
(6.42)
(6.43)
Now, for solving the transient stability problem with RK 4th order method, let us again assume
that the value of i and i (i = 1, 2, m) are known at t = to (denoted as io and io respectively). Moreover, the values of Pei are also assumed to be known (calculated utilising the values
of io ). From these initial values, the procedure of calculation of i1 and i1 (values of i and i at
t = to + t) is as follows.
di (1)
= io s
dt
(6.44)
di (1) s
[Pmi Pei(o) ]
=
dt
2Hi
(6.45)
With these first estimates of the derivatives, the values of i and i (i = 1, 2, m) are updated
as:
(1)
i
1 di (1)
= io + t
2
dt
(6.46)
(1)
i
1 di (1)
= io + t
2
dt
(6.47)
(1)
di (2)
= i(1) s
dt
(6.48)
di (2) s
[Pmi Pei(1) ]
=
dt
2Hi
(6.49)
With these second estimates of the derivatives, the values of i and i (i = 1, 2, m) are
updated as:
1 di (2)
i(2) = io + t
2
dt
(6.50)
1 di (2)
= io + t
2
dt
(6.51)
(2)
i
(2)
With these values of i , the values of Pei are again updated from equations (6.23) - (6.25). Let
(2)
these newly calculated values of Pei be denoted as Pei (i = 1, 2, m). We now proceed to the
next step.
di (3)
= i(2) s
dt
(6.52)
di (3) s
[Pmi Pei(2) ]
=
dt
2Hi
(6.53)
With these third estimates of the derivatives, the values of i and i (i = 1, 2, m) are updated
as;
(3)
i
di (3)
= io + t
dt
(6.54)
(3)
i
di (3)
= io + t
dt
(6.55)
(3)
With these values of i , the values of Pei are again updated from equations (6.23) - (6.25). Let
(3)
these newly calculated values of Pei be denoted as Pei (i = 1, 2, m). We now proceed to the
next step.
di (4)
= i(3) s
dt
(6.56)
di (4) s
[Pmi Pei(3) ]
=
dt
2Hi
(6.57)
After the fourth estimates are obtained, we are now in a position to calculate i1 and i1
(i = 1, 2, m).
Calculation of final values
The final, updated values are calculated as:
t di (1)
di (2)
di (3) di (4)
i1 = io +
[ +2 +2 +
]
6 dt
dt
dt
dt
(6.58)
di (2)
di (3) di (4)
t di (1)
[
+2
+2
+
]
i1 = io +
6 dt
dt
dt
dt
(6.59)
Proceeding further, for calculating i and i at t = to +2t, the quantities io and io are replaced
by i1 and i1 respectively and equations (6.44)-(6.59) are followed again.
In the next lecture, we will illustrate the applicaion of Modified Eulers method for transient
stability calculation.
265
6.3.6
As an illustration of the Modified Eulers method, let us consider a three machine, 9 bus system.
The schematic diagram of this system is shown in Fig. 6.4. The bus data of this system is given
in Table A.7 while the line data are given in Table A.8. Further, the values of xdi , di (damping
constant) and Hi (inertia constant) for all the three generators are given in Table 6.1.
With the data given in Tables A.7 and Table A.8, the load flow solution of this system has been
carried out and the load flow results are given in Table 6.2. With the help of load flow results and
and ) of the internal voltages of all the generators
values of xdi , the magnitudes and angles (E
have been calculated by utilising equations (6.16)-(6.18) and are also shown in Table 6.2. It is to
i (i = 1, 2, 3) are to
be noted that, throughout the transient stability simulation, the values of E
be kept constant at the values given in Table 6.2. Also, at steady state, the speed of all generators
are assumed to be equal to s (i.e. io = s ; i = 1, 2, 3). Now, following the arguments given in
step 7 of sub-section 6.3.3, under steady state, the values of i (i = 1, 2, 3) will remain constant
at those values given in Table 6.2. Similarly, under steady state, the values of i (i = 1, 2, 3) will
all be equal to s (= 376.9911184307752 rad./sec for a 60 Hz. system). Moreover, for transient
stability simulation, a time step of 0.001 sec. (t = 0.001) has been taken. Further, to start with,
the damping of the generators have been neglected.
no.
(MW)
(MVAR)
(p.u)
(deg.)
(p.u)
(deg.)
1
71.64102147 27.0459235334 1.04
0.0
1.0566418430 2.2716458404
2
163.0
6.6536603184 1.025 9.2800054816 1.0502010147 19.7315857693
3
85.0
-10.8597090709 1.025 4.6647513331 1.0169664112 13.1664110346
values of Vi (i = 1, 2, 3). These values are shown in second column of Table 6.3. With these
(o)
calculated values of the generator terminal voltages, the values of Pei (i = 1, 2, 3) have been
calculated using equations (6.24)-(6.25) and are shown in third column of Table 6.3. Now, from Fig.
6.4, bus 7 is the terminal bus of generator 2 (just after the transformer) and therefore, for any short
circuit fault at bus 7, the real power output of generator 2 is expected to fall drastically. Indeed,
(o)
from Table 6.3, after the fault, Pe2 is indeed very low (0.0003691143 p.u.). Also, note that the
values of io and io (i = 1, 2, 3), which are to be used for calculation at this time instant, are all
equal to the corresponding steady state values as there was no fault prior to this time instant.
Table 6.3: Calculations with Eulers method for predictor stage at t = 0.5 sec. (damping = 0)
Gen.
no.
1
2
3
V (p.u)
0.8515307492 - 0.0053320553i
0.3391142785 + 0.1215867116i
0.6169489129 + 0.0743553183i
(o)
Pe(o)
(p.u)
0.6791748939
0.0003691143
0.3821503052
di
di
and
(i = 1, 2, 3) are
dt
dt
calculated from equations (6.27) and (6.26) respectively and are shown in columns 2 and 3 of Table
(1)
(1)
6.4 respectively. Finally, the values of i and i (i = 1, 2, 3) are calculated by using equations
(6.32) - (6.33) and are shown in columns 4 and 5 of Table 6.4 respectively. From Tables 6.1 - 6.4 it
is observed that the steady state values of i (for i = 1, 2, 3; shown in the last column of Table 6.1)
(1)
are equal to the corresponding values of i (for i = 1, 2, 3; shown in the fourth column of Table
di
(i = 1, 2, 3) are all equal to zero.
dt
(1)
However, due to fall in Pei (i = 1, 2, 3), from equation (6.33), the values of i (i = 1, 2, 3) are all
6.4). This is due to the fact that at this predictor stage,
267
greater than s .
Table 6.4: Calculations with Eulers method for predictor stage at t = 0.5 sec. (damping = 0)
Gen.
no.
1
2
3
d
dt
0
0
0
(1)
d
dt
(1)
(deg.)
(rad/sec.)
0.2968990107 2.2716458404 376.9914153297
47.9965914231 19.7315857693 377.0391150221
29.2982026019 13.1664110346 377.0204166333
Once the calculations pertaining to the predictor stage at t = 0.5 sec. are over, we move on
to the calculations pertaining to the corrector stage. Towards this goal, initially the values of Pei
(1)
(i = 1, 2, 3) are updated using the values of i (i = 1, 2, 3) in equations (6.23) - (6.25). As
(1)
(1)
discussed earlier, the updated value of Pei is denoted as Pei . Now, as the values of i (i = 1, 2, 3)
(1)
are equal to the corresponding steady state values, the values of Pei (i = 1, 2, 3) are also equal
(1)
(1)
to the values given in Table 6.3. Subsequently, with the values of Pei and i (i = 1, 2, 3), the
values of
di
di
and
(i = 1, 2, 3) at the end of the present time step are calculated from equations
dt
dt
(6.34) - (6.35) and are shown in columns 2 and 3 of Table 6.5 respectively. Lastly, the final values
of i and i (i = 1, 2, 3) at t = 0.5 sec. are calculated by using equations (6.36) - (6.37), which are
shown in columns 4 and 5 of Table 6.5 respectively.
Table 6.5: Calculations with Eulers method for corrector stage at t = 0.5 sec. (damping = 0)
Gen.
d
d
no.
(deg.)
(rad/sec.)
dt
dt
1
0.0002968990 0.2968990107 2.2716543459 376.9914153297
2
0.0479965914 47.9965914231 19.7329607703 377.0391150221
3
0.0292982026 29.2982026019 13.1672503663 377.0204166333
With these final values of i and i (i = 1, 2, 3) corresponding to t = 0.5 sec. at hand, we now
increment the time by t (= 0.001 sec.) and repeat the calculations for predictor and corrector
stages for t = 0.501 sec. The detailed calculations are shown in Tables 6.6 - 6.9. Initially, with the
values of i and i (i = 1, 2, 3) just obtained, the generator terminal voltages and the generator
output electrical powers are calculated and are shown in Table 6.6. With these newly calculated
di di (1)
(1)
,
, i and i (i = 1, 2, 3) corresponding to the
dt dt
(1)
(1)
predictor stage are calculated and are shown in Table 6.7. With these updated values of i and i
(i = 1, 2, 3), the values of the generator terminal voltages and Pei (i = 1, 2, 3) are re-calculated
values of Pei (i = 1, 2, 3), the values of
corresponding to the corrector stage and are shown in Table 6.8. Lastly, using these updated values
di
di
and
corresponding to the corrector stage are calculated and
dt
dt
are shown in Table 6.9. Finally, the values of i and i (i = 1, 2, 3) at t = 0.501 sec. are calculated
by using equations (6.36)-(6.37), which are shown in columns 4 and 5 of Table 6.9 respectively.
268
Table 6.6: Caculations with Eulers method for predictor stage at t = 0.501 sec. (damping = 0)
Gen.
no.
1
2
3
V (p.u)
0.8515306823 - 0.0053313633i
0.3391113605 + 0.1215948479i
0.6169476981 + 0.0743625981i
Pe
(p.u)
0.6791650241
0.0003691288
0.3821597463
Table 6.7: Calculations with Eulers method for predictor stage at t = 0.501 sec. (damping = 0)
Gen.
(1)
(1)
d
d
no.
(deg.)
(rad/sec.)
dt
dt
1
0.0002968990 0.2969777087 2.2716713570 376.9917123074
2
0.0479965914 47.9965909948 19.7357107724 377.0871116131
3
0.0292982026 29.2976113769 13.1689290296 377.0497142447
Table 6.8: Caculations with Eulers method for corrector stage at t = 0.501 sec. (damping = 0)
Gen.
no.
1
2
3
V (p.u)
0.8515305484 - 0.0053299791i
0.3391055238 + 0.1216111204i
0.6169452683 + 0.0743771578i
Pe
(p.u)
0.6791452844
0.0003691579
0.3821786281
Table 6.9: Calculations with Eulers method for corrector stage at t = 0.501 sec. (damping = 0)
Gen.
d
d
no.
(deg.)
(rad/sec.)
dt
dt
1
0.0005938767 0.2971351045 2.2716798648 376.9917123861
2
0.0959931824 47.9965901381 19.7370857735 377.0871116127
3
0.0585958139 29.2964289348 13.1697683443 377.0497136535
For subsequent time instants, the calculations proceed exactly in the same way as described
above. The fault is assumed to be cleared at t = 0.6 sec. To simulate this event (clearing of fault),
the value of Y77 is restored to its pre-fault value and subsequently, the values of Pei (i = 1, 2, 3)
are calculated from equations (6.23)-(6.25). Please note that, while doing so, the latest values of i
(i = 1, 2, 3) obatained at t = 0.599 sec. are used in equation set (6.23). For subsequent instances
(beyond t = 0.6 sec.), the calculations proceed in the identical manner and finally, the simulation
study is stopped at t = 5.0 sec. The variations of i (i = 1, 2, 3) with respect to the center of inertia
(COI) are shown in Fig. 6.5. For calculating the value of i with respect to the COI (denoted as
269
iCOI ) at each time step, the following expression has been used:
iCOI = i COI ;
for i = 1, 2, 3;
(6.60)
where,
m
COI =
Hi i
i=1
m
(6.61)
Hj
j=1
In equation (6.61), m denotes the number of generators in the system (in our present case, m
= 3). At each time step, with the final calculated values of i (i = 1, 2, 3) calculated at the end
of corrector step, the value of COI is computed and thereafter, each value of iCOI (i = 1, 2, 3)
is computed with the help of equation (6.60). With the values of iCOI thus obtained for all time
steps, the plots shown in Fig. 6.5 are obtained. These plots show that the generators experience
sustained oscillations. This is due to the fact that in this study, the damping of the generators have
been neglected.
Figure 6.5: Variation of 1COI (with no damping) obtained with Eulers method
Let us now consider the damping of the generators. As discussed earlier (in the context of
equations (6.14) and (6.15)), when damping is considered, an extra term (representing damping)
is introduced in the differential equation corresponding to rate of change of speed. However, there
would be no change in the differential equation representing the rate of change of generator angle.
Therefore, the set of differential equations for ith generator is given by;
di
= i s
dt
(6.62)
2Hi di
= Pmi Pei di (i s )
s dt
(6.63)
In equation (6.63), the extra term di (i s ) represnts the damping of the generator. There
would be, of course, no change in the algebraic equations and the set of algebraic equations would
270
still be represented by equation set (6.23). With these sets of differential and algebraic equations,
the calculations proceed in identically the same way as described above and the variations of iCOI
(i = 1, 2, 3) for the same fault considered above are shown in Fig. 6.6. Comparison of these
three plots with those shown in Fig. 6.5 shows that when damping of the generators are taken into
consideration, the oscillations in all the three generators reduce gradually with time, which, indeed
should be the case. As the generator oscillations are decreasing with time, the generators will remain
in synchronism and therefore, the system is stable.
Figure 6.6: Variation of 1COI (with damping) obtained with Eulers method
We will now discuss the application of Runga Kutta (RK) 4th order method of integration for
solving the transient stability problem in the next lecture.
271
6.3.7
Again, as an example, 3 machine, 9 bus system shown in Fig. 6.4 is again considered. Initially, the
damping of the generators are neglected (i.e. di = 0 for i = 1, 2, 3). The load flow results and the
initial values of the magnitudes and angles (E and ) of the internal voltages of all the generators
are same as those already given in Table 6.2. As before, it is again assumed that at t = 0.5 sec.,
a three phase to ground short circuit fault takes place at bus 7. The faulted generator terminal
voltages and generator output powers are same as those shown in Table 6.3. With these values of
(o)
ei
di (1)
di (1)
and io , the estimates
and
(i = 1, 2, 3) are calculated from equations (6.44)
dt
dt
and (6.45) respectively and are shown in columns 2 and 3 of Table 6.10 respectively. With these first
(1)
(1)
estimates of the derivatives, the values of i and i (i = 1, 2, 3) are calculated from equations
(6.46) - (6.47) and are shown in columns 4 and 5 of Table 6.10 respectively.
Table 6.10: Calculations with RK method for first estimate at t = 0.5 sec. (damping = 0)
Gen. no.
di (1)
dt
1
2
3
0
0
0
di (1)
dt
(1)
(1)
(deg.)
(rad/sec.)
0.2968990107 2.2716458404 376.99126688
47.9965914231 19.7315857693 377.01511672
29.2982026019 13.1664110346 377.00576753
(1)
With the values of i (i = 1, 2, 3) obtained above, the generator terminal voltages and output
powers are updated from equations (6.23) - (6.25) and are shown in Table 6.11. Please note that
(1)
following the notations used earlier, the output powers calculated at this stage are denoted as Pei
(i = 1, 2, 3).
Table 6.11: Caculations with RK method for second estimate at t = 0.5 sec. (damping = 0)
Gen.
no.
1
2
3
(1)
V (p.u)
0.8515307492 - 0.0053320553i
0.3391142785 + 0.1215867116i
0.6169489129 + 0.0743553183i
(1)
Pe(1)
(p.u)
0.6791748939
0.0003691143
0.3821503052
di (2)
di (2)
and
dt
dt
(i = 1, 2, 3) are calculated from equations (6.48) - (6.49) and are shown in columns 2 and 3 of Table
(2)
(2)
6.12 respectively. Subsequently, the values of i and i (i = 1, 2, 3) are calculated from equations
(6.50) - (6.51) and are shown in columns 4 and 5 of Table 6.12 respectively.
(2)
Again, with the values of i (i = 1, 2, 3) obtained above, the generator terminal voltages and
output powers are updated from equations (6.23) - (6.25) and are shown in Table 6.13. Please note
272
Table 6.12: Calculations with RK method for second estimate at t = 0.5 sec. (damping = 0)
Gen. no.
1
2
3
di (2)
dt
di (2)
dt
(2)
(2)
(deg.)
(rad/sec.)
0.0001484495 0.2968990107 2.2716500931 376.9912668802
0.0239982957 47.9965914231 19.7322732698 377.0151167264
0.0146491013 29.2982026019 13.1668307004 377.0057675320
that following the notations used earlier, the output powers calculated at this stage are denoted as
Pei(2) (i = 1, 2, 3).
Table 6.13: Caculations with RK method for third estimate at t = 0.5 sec. (damping = 0)
Gen.
no.
1
2
3
Pe(2)
V (p.u)
0.8515307158 - 0.0053317093i
0.3391128195 + 0.1215907798i
0.6169483055 + 0.0743589582i
(2)
(p.u)
0.6791699590
0.0003691216
0.3821550258
(2)
di (3)
di (3)
and
(i = 1, 2, 3) are calculated from equations (6.52) - (6.53) and are shown in
dt
dt
(3)
(3)
columns 2 and 3 of Table 6.14 respectively. Subsequently, the values of i and i (i = 1, 2, 3) are
calculated from equations (6.54) - (6.55) and are shown in columns 4 and 5 of Table 6.14 respectively.
Table 6.14: Calculations with RK method for third estimate at t = 0.5 sec. (damping = 0)
Gen. no.
1
2
3
di (3)
dt
di (3)
dt
(3)
(3)
(deg.)
(rad/sec.)
0.0001484495 0.2969383597 2.2716543459 376.9914153691
0.0239982957 47.9965912089 19.7329607703 377.0391150219
0.0146491013 29.2979069891 13.1672503663 377.0204163377
(3)
Again, with the values of i (i = 1, 2, 3) obtained above, the generator terminal voltages and
output powers are updated from equations (6.23) - (6.25) and are shown in Table 6.15. Please note
that following the notations used earlier, the output powers calculated at this stage are denoted as
Pei(3) (i = 1, 2, 3).
Lastly, with the values of
and
(3)
i
and P
(3)
ei
di (4)
(i = 1, 2, 3) calculated above, the quantities
dt
di (4)
(i = 1, 2, 3) are calculated from equations (6.56) - (6.57) and are shown in columns 2
dt
273
Table 6.15: Calculations with RK method for final estimate at t = 0.5 sec. (damping = 0)
Gen.
no.
1
2
3
V (p.u)
0.8515306823 - 0.0053313633i
0.3391113605 + 0.1215948479i
0.6169476981 + 0.0743625981i
Pe(3)
(p.u)
0.6791650241
0.0003691288
0.3821597463
and 3 of Table 6.16 respectively. Finally, the values of i and i (i = 1, 2, 3) at the end of t = 0.5
sec. are calculated from equations (6.58) - (6.59) and are shown in columns 4 and 5 of Table 6.16
respectively.
Table 6.16: Calculations with RK method for final estimate at t = 0.5 sec. (damping = 0)
di (4)
dt
Gen. no.
1
2
3
di (4)
dt
(deg.)
(rad/sec.)
0.0002969383 0.2969777087 2.2716543463 376.9914153560
0.0479965912 47.9965909948 19.7329607703 377.0391150220
0.0292979069 29.2976113769 13.1672503634 377.0204164363
After the final values of i and i (i = 1, 2, 3) corresponding to t = 0.5 sec. are obtained, we
increment the time by t (= 0.001 sec.) and repeat the calculations for t = 0.501 sec. Towards
this goal, the quantities i and i (i = 1, 2, 3) shown in Table 6.16 are substituted for io and io
(i = 1, 2, 3) in equations (6.44) - (6.59). With these values of io (i = 1, 2, 3), equations (6.23) (6.25) are solved to calculate the initial values of Pei (i = 1, 2, 3) at t = 0.501 sec. The results are
shown in Table 6.17.
Table 6.17: Calculations with RK method for initial estimate at t = 0.501 sec. (damping = 0)
Gen.
no.
1
2
3
V (p.u)
0.8515306823 - 0.0053313633i
0.3391113605 + 0.1215948479i
0.6169476981 + 0.0743625981i
Pe(o)
(p.u)
0.6791650241
0.0003691288
0.3821597462
(o)
With the values of Pei (i = 1, 2, 3) thus obtained, calculations pertaining to the first estimate
are carried out by using equations (6.44) - (6.47) and the results are shown in Table 6.18. Sub(1)
sequently, the values of Pei (i = 1, 2, 3) are calculated and the results are shown in Table 6.19.
(1)
(1)
(1)
With the values of Pei , i and i (i = 1, 2, 3) obtained as above, the calculations pertaining
(2)
(2)
(2)
to second estimate are performed to obtain Pei , i and i (i = 1, 2, 3). The results are shown
274
Table 6.18: Calculations with RK method for first estimate at t = 0.501 sec. (damping = 0)
di (1)
dt
Gen. no.
1
2
3
di (1)
dt
(1)
(1)
(deg.)
(rad/sec.)
0.0002969252 0.2969777084 2.2716628526 376.9915638448
0.0479965912 47.9965909948 19.7343357714 377.0631133175
0.0292980055 29.2976113792 13.1680896895 377.0350652419
Table 6.19: Calculations with RK method for first estimate at t = 0.501 sec. (damping = 0)
Gen.
no.
1
2
3
Pe(1)
V (p.u)
0.8515306154 - 0.0053306712i
0.3391084422 + 0.1216029842i
0.6169464833 + 0.0743698779i
(p.u)
0.6791551544
0.0003691434
0.3821691871
Gen. no.
1
2
3
di (2)
dt
di (2)
dt
(2)
(2)
(deg.)
(rad/sec.)
0.0004454140 0.2970564057 2.2716671065 376.9915638842
0.0719948867 47.9965905664 19.7350232719 377.0631133173
0.0439468112 29.2970201614 13.1685093468 377.0350649463
Table 6.21: Calculations with RK method for second estimate at t = 0.501 sec. (damping = 0)
Gen.
no.
1
2
3
Pe(2)
V (p.u)
0.8515305819 - 0.0053303251i
0.3391069830 + 0.1216070523i
0.6169458758 + 0.0743735177i
(2)
(2)
(2)
(p.u)
0.6791502196
0.0003691506
0.3821739075
Proceeding further, using the values of Pei , i and i (i = 1, 2, 3), the calculations pertaining
(3)
(3)
(3)
to third estimate are performed to obtain Pei , i and i (i = 1, 2, 3). The results are shown in
Tables 6.22 and 6.23.
(3)
(3)
(3)
Using the values of Pei , i and i (i = 1, 2, 3), the fourth estimates of the derivatives are
computed and subsequently, the final values of i and i (i = 1, 2, 3) corresponding to t = 0.501
sec. are obtained. The calculations are shown in Table 6.24.
275
Table 6.22: Calculations with RK method for third estimate at t = 0.501 sec. (damping = 0)
Gen. no.
1
2
3
di (3)
dt
di (3)
dt
(3)
(3)
(deg.)
(rad/sec.)
0.0004454534 0.2970957537 2.2716798689 376.9917124517
0.0719948865 47.9965903523 19.7370857735 377.0871116124
0.0439465156 29.2967245580 13.1697683133 377.0497131608
Table 6.23: Calculations with RK method for third estimate at t = 0.501 sec. (damping = 0)
Gen.
no.
1
2
3
V (p.u)
0.8515304815 - 0.0053292869i
0.3391026052 + 0.1216192565i
0.6169440533 + 0.0743844372i
Pe(3)
(p.u)
0.6791354153
0.0003691725
0.3821880684
Table 6.24: Calculations with RK method for final estimate at t = 0.501 sec. (damping = 0)
Gen. no.
1
2
3
di (4)
dt
di (4)
dt
(deg.)
(rad/sec.)
0.0005940209 0.2972137971 2.2716798685 376.9917124386
0.0959931816 47.9965897098 19.7370857735 377.0871116124
0.0585947300 29.2958377565 13.1697683161 377.0497132593
For subsequent time instants, the calculations proceed exactly in the same way as described
above. As in the case with Eulers method, in this case also, the fault is assumed to be cleared at t =
0.6 sec. and finally, the simulation study is stopped at t = 5.0 sec. The variations of i (i = 1, 2, 3)
with respect to the center of inertia (COI) are shown in Fig. 6.7 below. Please note that in this
figure, no damping of the generators has been considered.
The simulation studies have also been carried out by considering the damping of the generators.
The variations of i (i = 1, 2, 3) with respect to the center of inertia (COI) for this case are shown
in Fig. 6.8 below. Comaprison of Figs. 6.7 - 6.8 with Figs. 6.5 - 6.6 reveals that the responses
obtained with these two methods are almost identical to each other.
With this example, we are now at the end of discussion of transient stability analysis. From the
next lecture, we will start the discussion of small signal stability analysis.
276
Figure 6.7: Variation of 1COI (with no damping) obtained with Runga-Kutta method
Figure 6.8: Variation of 1COI (with damping) obtained with Runga-Kutta method
277
6.4
For small signal analysis of a multimachine power system, we need to linearise the differential equations of the machines and form the system A matrix. Thereafter, by computing the eigenvalues of
the A matrix, the system stability can be assessed. Now, let us consider an n bus power system
having m generators. Each generator is represented by its classical model. Further, without any
loss of generality, it is assumed that the generators are connected at the first m buses of the system.
Now, for linearising the differential equations, let us recall the swing equations of each generator
here for ready reference.
di
= i s
for i = 1, 2, m
dt
2Hi di
= Pmi Pei
for i = 1, 2, m
s dt
(6.64)
(6.65)
Linearising equation (6.64) for the ith generator, one can get,
di
= i
dt
(6.66)
= [1 , 2 , m ]
= [1 , 2 , m ]
(6.67)
(6.68)
In equations (6.67) and (6.68), the vectors and denote the vectors of perturbed values
of rotor angle and machine speed respectively. Please note that the size of each of these two vectors
is (m 1).
Now, collecting equation (6.66) for all the m generators and re-writing them in matrix notation
we can obtain,
d
= F1
dt
(6.69)
In equation (6.69), F1 is a (m m) identity matrix. Now, linearising equation (6.65) for the ith
generator, one can get,
2Hi di
= Pei
s dt
(6.70)
For performing linearisation of Pei , its expression its required. This expression can be derived as
follows. From equation (6.25), one can write,
Or,
E i Vi
Ei
Ei Ii = Ei eji Ii = ej/2 ej(/2+i i )
xdi
xdi
Therefore,
E i Vi
Pei = Re (EI Ii ) = sin(i i )
xdi
(6.71)
(6.72)
Where,
k1i =
Ei Vi
cos(i i );
xdi
k2i =
Ei Vi
cos(i i );
xdi
k3i =
Ei
sin(i i );
xdi
(6.73)
In equation (6.73), the constants k1i , k2i and k3i are evaluated using the values of Ei , Vi , i and
i at the current operating point for i = 1, 2, m.
Now, again let us define,
g = [1 , 2 , m ]
Vg = [V1 , V2 , Vm ]
(6.74)
(6.75)
(6.76)
In equations (6.74) and (6.75), the vectors g and Vg denote the vectors of perturbed values
of generator terminal voltage angles and magnitudes respectively. Similarly, the vector Pe denotes
the perturbed values of the generator real powers. Please note that the size of each of these three
vectors is also (m 1).
Now, collecting equation (6.72) for all the m generators and re-writing them in matrix notation
we can obtain,
Pe = K1 + K2 g + K3 Vg
(6.77)
In equation (6.77), the size of each of the matrices K1 , K2 and K3 is (m m). Moreover, all
these three matrices are diagonal matrices and are given by;
(6.78)
6.4.1
To illustrate the procedure for linearisation of the network equations, let us first consider the first
equation of the equation set (6.23) below.
(6.79)
Now, let,
Vi = Vi exp (ji );
for i, j = 1, 2, n;
(6.80)
yk = yk exp (jk );
Ek = Ek exp (jk );
for k = 1, 2, m;
(6.81)
and
(6.82)
Taking the real part of both the sides of equation (6.82), we get,
(6.83)
(6.84)
(6.85)
In equation (6.85),
(6.86)
Again, taking the real part of both sides of the second equation of the equation set (6.23) (after
280
(6.87)
(6.88)
In equation (6.88),
(6.89)
Similarly, continuing with linearisation of the real parts of first m equations (corresponding to
the generator buses) of the equation set (6.23), we get,
[A1
Vg
L
= [G] []
B1 ]
(6.90)
In equation (6.90),
is a (m n) matrix
A1 =
(6.91)
B1 =
is a (m n) matrix
(6.92)
281
g1 0 0 0
0 g2 0 0
is a (m m) diagonal matrix
G=
0 0 0 gm
(6.93)
i = 1, 2, m; j = 1, 2, n; i j;
i = 1, 2, m;
i = 1, 2, m;
(6.94)
i = 1, 2, m; j = 1, 2, n; i j;
gi = yi Ei sin (i + i );
i = 1, 2, m;
L = [m+1 , m+2 , n ]
VL = [Vm+1 , Vm+2 , Vn ]
(6.95)
(6.96)
Please note that the size of each of the above two vectors is ((n m) 1). So far, we have
considered only the algebraic equations at the generator buses. However, for completing the small
signal model, the algebraic equations at the load buses all need to be linearised. We will discuss this
issue in the next lecture.
282
6.4.2
Now, let us consider the algebraic equations at the load buses. For this, please recollect that the
loads have been assumed to be connected at the last (nm) buses. The real part of the pth equation
(p = (m + 1), (m + 2), n) of the equations set (6.23) is given by,
(6.97)
Yp1 cos (p1 + 1 )V1 Yp1 V1 sin (p1 + 1 )1 + Yp2 cos (p2 + 2 )V2
Yp2 V2 sin (p2 + 2 )2 + + Ypn cos (pn + n )Vn Ypn Vn sin (pn + n )n = 0
(6.98)
Or,
(6.99)
In equation (6.99),
(6.100)
Please note that equation (6.99) can be written for all the n m load buses (by varying p from
m + 1 to n). In that case, in equation (6.100) also, p will vary from m + 1 to n. Collecting all
these n m equations and putting them in a matrix form, we get,
[R1
Vg
L
= [1 ] []
S1 ]
(6.101)
In equation (6.101), the matrix 1 is a (n m) m null matrix and the matrices R1 and S1
are given by,
r11
r12
r1n
r21
r22
r2n
R1 =
is a ((n m) n) matrix
283
(6.102)
s11
s12
s1n
s21
s
22
2n
is a ((n m) n) matrix
S1 =
(6.103)
rij = a(i+m),j ;
sij = b(i+m),j ;
i = 1, 2, (n m); j = 1, 2, n;
i = 1, 2, (n m); j = 1, 2, n;
(6.104)
In equation (6.104), the expressions of the co-efficients a(i+m),j and b(i+m),j are given by equation
(6.100). Now, taking the imaginary part of both the sides of equation (6.82), we get,
(6.105)
(6.106)
(6.107)
In equation (6.107),
(6.108)
Again, taking the imaginary part of both sides of the second equation of the equation set (6.23)
(after substituting equations (6.80) and (6.81) into it), we get,
(6.109)
(6.110)
In equation (6.110),
(6.111)
Continuing with linearisation of the imaginary parts of first m equations (corresponding to the
generator buses) of the equation set (6.23), we get,
[C1
Vg
L
= [H] []
D1 ]
(6.112)
In equation (6.112),
C1 =
is a (m n) matrix
is a (m n) matrix
D1 =
h1 0 0 0
0 h2 0 0
H=
is a (m m) diagonal matrix
0 0 0 hm
285
(6.113)
(6.114)
(6.115)
i = 1, 2, m; j = 1, 2, n; i j;
i = 1, 2, m;
i = 1, 2, m;
(6.116)
i = 1, 2, m; j = 1, 2, n; i j;
hi = yi Ei cos (i + i );
i = 1, 2, m;
Now, let us consider the imaginary parts of the algebraic equations at the load buses. The real
part of the pth equation (p = (m + 1), (m + 2), n) of the equations set (6.23) is given by,
(6.117)
Yp1 sin (p1 + 1 )V1 + Yp1 V1 cos (p1 + 1 )1 + Yp2 sin (p2 + 2 )V2
+Yp2 V2 cos (p2 + 2 )2 + + Ypn sin (pn + n )Vn + Ypn Vn cos (pn + n )n = 0
(6.118)
Or,
(6.119)
In equation (6.119),
(6.120)
Again, equation (6.119) can be written for all the n m load buses (by varying p from m + 1
to n). In that case, in equation (6.120) also, p will vary from m + 1 to n. Collecting all these
n m equations and putting them in a matrix form, we get,
[U1
Vg
L
= [2 ] []
V1 ]
(6.121)
In equation (6.121), the matrix 2 is a (n m) m null matrix and the matrices U1 and V1
286
u11
u12
u1n
u21
u22
u2n
U1 =
is a ((n m) n) matrix
(6.122)
v11
v12
v1n
v21
v22
v2n
V1 =
is a ((n m) n) matrix
(6.123)
uij = c(i+m),j ;
vij = d(i+m),j ;
i = 1, 2, (n m); j = 1, 2, n;
i = 1, 2, (n m); j = 1, 2, n;
(6.124)
In equation (6.124), the expressions of the co-efficients c(i+m),j and d(i+m),j are given by equation
(6.120).
Now, combining equations (6.90), (6.101), (6.112) and (6.121), one can write,
A1
R
1
C1
U1
Or,
Where,
B1 Vg G
S1 VL 1
= []
D1 g H
V1 L 2
Vg
L
[J1 ]
= [J2 ] []
g
A1
R
1
[J1 ] =
C1
U1
B1
S1
;
D1
V1
G
1
[J2 ] =
H
2
(6.125)
(6.126)
(6.127)
Now, let us recollect that the dimensions of various sub-matrices are as follows: A1 (m n),
B1 (m n), G (m m), R1 ((n m) n), S1 ((n m) n), 1 ((n m) m),
C1 (m n), D1 (m n), H (m m), U1 ((n m) n), V1 ((n m) n),
287
2 ((n m) m). Therefore, the size of the matrix J1 is (2n 2n) and that of the matrix J2 is
(2n m). Hence, matrix J1 is a square matrix and hence invertible. Thus, from equation (6.126),
A2
Vg
B
V
1
2
L
= [J1 ] [J2 ] [] = []
C2
g
D2
L
(6.128)
[Vg ] = [A2 ] [] ;
[g ] = [C2 ] [] ;
(6.129)
Please note that the dimension of both the matrices A2 and C2 is (mm). We are now ready to
form the system state matrix which we will discuss in the next lecture. Further, in the next lecture,
we will also look at an example of small signal stability analysis.
288
6.4.3
From equation (6.70), let us recall the following linearised equation for ith generator,
2Hi di
= Pei
s dt
(6.130)
Collecting equation (6.130) for all the m generators, the following matrix equation can be
written;
d
2
HM = Pe
s
dt
(6.131)
HM = diag (H1 , H2 , Hm )
(6.132)
2
d
HM = (K1 + K2 g + K3 Vg )
s
dt
(6.133)
Further, substituting equation (6.129) into equation (6.133) one can write,
d
2
HM = (K1 + K2 C2 + K3 A2 )
s
dt
(6.134)
d
= K4
dt
(6.135)
Or,
Where,
K4 =
s 1
H (K1 + K2 C2 + K3 A2 )
2 M
(6.136)
d
= F1
dt
(6.137)
O F1
d
[
]=[
][
]
dt
K4 O
(6.138)
O F1
]
K4 O
Equation (6.138) is the required state space equation of the system and the matrix A = [
is the corresponding state matrix. The eigenvalues of the matrix A gives the necessary information
about the small signal stability of the system under study.
Now, let us consider the damping of the generators. Towards this goal, let us now linearize
289
2Hi di
= Pei di i
s dt
(6.139)
Collecting equation (6.139) for all the m generators, the following matrix equation can be
written;
d
2
HM = Pe DM
s
dt
(6.140)
DM = diag (d1 , d2 , dm )
(6.141)
Following the same procedure as in equations (6.133) and (6.134), we can get,
d
2
HM = (K1 + K2 C2 + K3 A2 ) DM
s
dt
(6.142)
d
= K4 + K5
dt
(6.143)
Or,
Where,
K5 =
s 1
H DM
2 M
(6.144)
O F1
d
[
]=[
][
]
dt
K4 K5
(6.145)
Equation (6.145) is the required state space equation of the system when damping of the gener-
O F1
] is the corresponding state matrix.
K4 K5
6.5
As an illustration of small signal stability, let us consider the three machine, 9 bus system (m = 3
and n = 9) shown in Fig. Fig. 6.4. The complete data of this system are given in Tables A.7, A.8
and 6.1. Further, the different quantities pertaining to the initial condition are given in Table 6.2.
The load flow solution of the system is shown in Table 6.25.
Initially, the damping of the machines are neglected (di = 0; i = 1, 2, 3). With these initial
values, the different matrices are calculated as follows:
18.0599
0
0
K1 = 0
8.8364
0
0
0
5.6864
290
(6.146)
Voltage (p.u.)
1.0400
1.0116 + 0.1653i
1.0216 + 0.0834i
1.0250 - 0.0397i
0.9932 - 0.0693i
1.0106 - 0.0651i
1.0236 + 0.0665i
1.0158 + 0.0129i
1.0317 + 0.0354i
18.0599
0
0
K2 =
0
8.8364
0
0
0
5.6864
0.6889
0
0
K3 = 0
1.5902
0
0
0
0.8293
(6.147)
(6.148)
0.0000
0
0
0.6715 0 0
0
0
0
A1 = 0
(6.149)
3.9262
0
0
0 0 1.0380 0
0
0
0
1.8364
0
0 0
0
0 0.5856
35.1608
0
0
17.7955 0 0
0
0
0
B1 = 0
(6.150)
24.6293
0
0
0 0 16.3777 0
0
0
0
23.0684
0
0 0
0
0 17.6066
17.3653
0
0
G = 0
(6.151)
8.2516
0
0
0
5.4618
0.0000
0
0
1.7844 0.5547 1.2622
0
0
0
0
0
0
0.9153 2.5634
0
1.5727
0
0
0
0
0
1.5342
0
3.0558
0
0
1.4730
R1 =
2.5802
0
0
0.7691
0
5.0984 1.7909
0
0
0
0
0
0
0
0
2.5024 4.0411 1.4902
0
0
1.3878
0
0
0.9200
0
1.2792 3.5391
(6.152)
The matrix S1 is given in parts in equations (6.153) and (6.154) below, as the width of the page
is not sufficient enough to accomodate the complete matrix S1 .
291
The matrix S1 (, 1 5) (comprising of the elements in columns 1 to 5 for all rows) is given by;
18.0556
0
0
40.4237 11.6200
0
0
0
11.9486 17.9858
0
0
0
10.8507
0
S1 (, 1 5) =
0
16.1854
0
0
6.0169
0
0
0
0
0
0
0
17.4335
0
0
(6.153)
The matrix S1 (, 6 9) (comprising of the elements in columns 6 to 9 for all rows) is given by;
10.7481
0
0
0
0
6.0372
0
0
16.5710
0
0
5.7202
S1 (, 6 9) =
0
36.0958 13.8936
0
0
13.9138 23.9677 10.0539
5.7307
0
9.9240 33.0882
33.8085
0
0
17.3481 0 0
0
0
0
C1 =
0
24.0286
0
0
0 0 15.9663 0
0
0
0
22.5058
0
0 0
0
0 17.0548
0.0000
0
0
0.6889 0 0
0
0
0
D1 = 0
4.0244
0
0
0 0 1.0648 0
0
0
0
1.8823
0
0 0
0
0 0.6046
0.6889
0
0
H = 0
2.9596
0
0
1.2777
(6.154)
(6.155)
(6.156)
(6.157)
The matrix U1 is given in parts in equations (6.158) and (6.159) below, as the width of the page
is not sufficient enough to accomodate the complete matrix U1 .
The matrix U1 (, 1 5) (comprising of the elements in columns 1 to 5 for all rows) is given by;
17.3611
0
0
39.4074 11.6710
0
0
11.6482
18.0647
0
0
10.5779
0
U1 (, 1 5) =
15.7906
0
0
6.0433
0
0
0
0
0
0
17.0083
0
0
292
(6.158)
The matrix U1 (, 6 9) (comprising of the elements in columns 6 to 9 for all rows) is given by;
10.6138
0
0
0
0
5.8855
0
0
16.3639
0
0
5.5410
U1 (, 6 9) =
0
35.1890 13.6763
0
0
13.5642 23.5930 9.7389
5.6591
0
9.7688 32.0513
(6.159)
0.0000
0
0
1.8304 0.5523 1.2782
0
0
0
0
0
0
0.9389 2.5522
0
1.6133
0
0
0
0
0
1.5737
0
3.0944
0
0
1.5207
V1 =
2.6447
0
0
0.7657
0
5.2297 1.8194
0
0
0
0
0
0
0
0
2.5669 4.1053 1.5384
0
0
1.4225
0
0
0.9316
0
1.2995 3.6536
(6.160)
The matrix J1 is given in parts in equations (6.161) - (6.163) below, as the width of the page is
not sufficient enough to accomodate the complete matrix J1 .
The matrix J1 (, 1 6) (comprising of the elements in columns 1 to 6 for all rows) is given by;
0.0000
0
0
0.6715
0
0
0
3.9262
0
0
0
0
0
0
1.8364
0
0
0
0.0000
0
0
1.7844
0.5547
1.2622
0
0
0
0.9153
2.5634
0
0
0
0
1.5342
0
3.0558
0
2.5802
0
0
0.7691
0
0
0
0
0
0
0
0
0
1.3878
0
0
0.9200
J1 (, 1 6) =
0
0
17.3481
0
0
33.8085
0
24.0286
0
0
0
0
0
0
22.5058
0
0
0
0
0
39.4074 11.6710 10.6138
17.3611
0
0
0
11.6482
18.0647
0
0
0
0
10.5779
0
16.3639
0
15.7906
0
0
6.0433
0
0
0
0
0
0
0
0
0
17.0083
0
0
5.6591
(6.161)
The matrices J1 (, 7 12) (comprising of the elements in columns 7 to 12 for all rows) and
293
J1 (, 13 18) (comprising of the elements in columns 13 to 18 for all rows) are given by;
0
0
0
35.1608
0
0
1.0380
0
0
0
24.6293
0
0
0
0.5856
0
0
23.0684
0
0
0
18.0556
0
0
1.5727
0
0
0
0
0
0
0
1.4730
0
0
0
5.0984
1.7909
0
0
16.1854
0
2.5024
4.0411
1.4902
0
0
0
0
1.2792
3.5391
0
0
17.4335
J1 (, 7 12) =
0
0
0
0.0000
0
0
15.9663
0
0
0
4.0244
0
0
0
17.0548
0
0
1.8823
0
0
0
0.0000
0
0
5.8855
0
0
0
0
0
0
0
5.5410
0
0
0
35.1890 13.6763
0
0
2.6447
0
0
9.7688 32.0513
0
0
1.4225
(6.162)
17.7955
0
0
0
0
0
0
0
0
16.3777
0
0
0
0
0
0
0
17.6066
0
0
0
11.9486 17.9858
0
6.0372
0
0
10.8507
0
16.5710
0
0
5.7202
0
6.0169
0
36.0958 13.8936
0
0
0
0
13.9138
23.9677
10.0539
0
0
5.7307
0
9.9240
33.0882
J1 (, 13 18) =
0
0
0
0
0
0.6889
0
0
0
1.0648
0
0
0
0
0
0
0
0.6046
0.5523 1.2782
0
0
0
1.8304
0.9389
2.5522
0
1.6133
0
0
1.5737
0
3.0944
0
0
1.5207
0
0.7657
0
5.2297
1.8194
0
0
0
0
2.5669
4.1053
1.5384
0
0
0.9316
0
1.2995
3.6536
(6.163)
294
17.3653
0
0
0
8.2516
0
0
0
5.4618
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
J2 =
0
0
0.6889
0
2.9596
0
0
0
1.2777
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
HM
23.6400
0
0
= 0
6.4000
0
0
3.0100
24.1837 13.5322
10.6514
1 0 0
F1 = 0 1 0
0 0 1
295
(6.164)
(6.165)
(6.166)
(6.167)
(6.168)
(6.169)
With all the above calculated matrices, the system state matrix is finally computed as;
0
0
0
1.0000
0
0
0
0
0
0
1.0000
0
0
0
0
0
0
1.0000
A=
10.6514
0
0
0
24.1837 13.5322
(6.170)
The eigenvalues of the system state matrix are lastly computed and are shown in Table 6.26.
Table 6.26: Eigenvalues of the 3 machine system with damping neglected
No.
Eigenvalue
1
0.0000 +13.3602i
2
0.0000 -13.3602i
3
0.0000 + 8.6898i
4
0.0000 - 8.6898i
5
-0.0000 + 0.0000i
6
-0.0000 - 0.0000i
0.2025
0
0
K5 = 0
0.1944
0
0
0
0.1628
296
(6.171)
0
0
0
1.0000
0
0
0
0
0
0
1.0000
0
0
0
0
0
0
1.0000
A=
10.6514 0.2025
0
0
24.1837 13.5322
(6.172)
Eigenvalue
-0.0844 +13.3599i
-0.0844 -13.3599i
-0.0970 + 8.6893i
-0.0970 - 8.6893i
-0.0000
-0.1970
297
6.6
Voltage stability
Voltage collapses usually occur on power system which are heavily loaded or faulted or have shortage
of reactive power. Voltage collapse is a system instability involving many power system components.
In fact, a voltage collapse may involve an entire power system.
Voltage collapse is typically associated with reactive power demand of load not being met due to
shortage in reactive power production and transmission.
Voltage collapse is a manifestation of voltage instability in the system. The definition of voltage
stability as proposed by IEEE/CIGRE task force is as follows:
Voltage stability refer to the ability of power system to maintain steady voltages at all buses in
the system after being subjected to a disturbance from a given initial operating point. The system
state enters the voltage instability region when a disturbance or an increase in load demand or
alteration in system state results in an uncontrollable and continuous drop in system voltage.
A system is said to be in voltage stable state if at a given operating condition, for every bus in
the system, the bus voltage magnitude increases as the reactive power injection at the same bus is
increased.
A system is voltage unstable if for at least one bus in the system, the bus voltage magnitude
decreases as the reactive power injection at the same bus is increased. It implies that if, V-Q
sensitivity is positive for every bus the system is voltage stable and if V-Q sensitivity is negative
for at least one bus, the system is voltage unstable. The term voltage collapse is also often used for
voltage instability conditions. It is the process, by which, the sequence of events following voltage
instability leads to abnormally low voltages or even a black out in a large part of the system.
The driving force for voltage instability is usually the loads and load characteristics, hence, voltage
stability is sometimes also called load stability. In response to a disturbance, the power consumed
by the loads tends to be restored by load dynamics. This in turn increases the stress on the high
voltage network by increasing the reactive power consumption and further reducing the voltage.
A major factor contributing to voltage instability is the voltage drop in the line impedances when
active and reactive powers flow through it. As a result, the capability of the transmission network
for power transfer and voltage support reduces. Voltage stability of a system is endangered when
a disturbance increases the reactive power demand beyond the sustainable capacity of the available
reactive power resources.
The voltage stability has been further classified into four categories: Large disturbance voltage
stability, small disturbance voltage stability, short term voltage satiability and long term voltage
stability. A summary of these classifications is as follows:
Large disturbance voltage stability: It refers to the systems ability to maintain steady voltage
following large disturbances such as, system faults, loss of generation or circuit contingencies.
This ability is determined by the system load characteristics and interaction of both continuous
and discrete controls and protections. The study period of interest may be from few seconds
to tens of minutes. This requires long term dynamic simulation study of the system to capture
the interactions of under-load tap changer and generator field current limiter.
298
If following a large disturbance and subsequent system control actions, voltages at all the buses
in the system settle down at acceptable levels, the system is said to be large disturbance voltage
stable.
Small-disturbance voltage stability: This stability is concerned with the ability of the system
to maintain acceptable level of steady voltages, when subjected to small perturbations such as
incremental changes in system load. This form of stability is also influenced by the characteristics of loads, continuous controls, and discrete controls at a given instant of time. The basic
processes contributing to small disturbance stability are essentially of a steady state nature.
Therefore, static analysis can be effectively used to estimate stability margins.
Short term voltage satiability: It involves dynamics of fast acting load components such as
induction motors, electronically controlled loads and HVDC converters. The study period of
interest is in the order of several seconds and the analysis requires solution of appropriate
system differential equations.
Long term voltage stability: The study of long term voltage stability involves the dynamics of
slower acting equipment such as tap changing transformers, thermostatically controlled loads
and generator current limiters. The study period of interest may extend to several or many
minutes, and requires long term dynamics system simulation.
Voltage instability may arise due many reasons, but some significant contributors are:
Increase in loading
Generators, synchronous condensers, or SVC reaching reactive power limits
Action of tap changing transformers
Load recovery dynamics
Line tripping or generator outages
Most of these changes have a significant impact on the reactive power production, consumption
and transmission in the system.
Some counter measures to prevent voltage collapse are:
I =
=
=
E
ZL + ZD
E
ZL + ZD
E
(ZL cos + ZD cos ) + j(ZL sin + ZL sin )
(6.173)
E
I=
(ZL cos + ZD cos )2 + (ZL sin + ZL sin )2
which may be written as:
I=
ZL FL
300
(6.174)
where,
ZD
ZD 2
FL = 1 + [ ] + 2 [ ] cos ( )
ZL
ZL
Where
The magnitude of the receiving and voltage is given by:
V = ZD I
E ZD
= [ ]
FL ZL
(6.175)
PL = V Icos
ZD
E 2
PL = ( ) [ ] cos
FL ZL
The plots of I,V and PL are shown in Fig. 6.10 as a function of
and .
(6.176)
ZL
ratio for a specific value of
ZD
Figure 6.10: Receiving end voltage, Current and Power as a function of Load
An explanation of the chracteristics of Fig. 6.10 is as follows:
As the load demand is increased by reducing ZD , the load power PL increases rapidly at first
and then slowly, before reaching a maximum value and then starts decreasing. There is thus, a
maximum value of active power that can be transmitted through an impedance from a constant
voltage source.
301
The transmitted power reaches a maximum when the voltage drop in the line is equal to the
load voltage V. This occurs, when the load impedance ZD is equal to line impedance ZL . As
ZD is gradually reduced, current in the line I increases and load voltage V decreases. Initially,
for high values of ZD , the enhancement in I is more than the reduction in V, and hence load
power PL increases rapidly with reduction in ZD . As ZD approaches ZL , the effect of the
enhancement in I is only slightly greater than that of the reduction in V, hence increase in PL
is slow. Finally, when ZD is quite less than ZL the reduction in V dominates over increase in
I and hence, PL decreases.
The critical operating condition, corresponding to maximum power, is the limiting point of
satisfactory operation. For higher load demand, control of power by varying load would be
unstable, as a reduction in load impedance will reduce power. The load characteristics decides
whether the system voltage decreases progressively and the system will become unstable. With
a constant impedance static load characteristic, the system stabilizes at power and voltage levels
lower than the desired values. For a constant power load characteristic, the system becomes
unstable through collapse of load bus voltage. If the load is supplied by transformers with
automatic under load-tap- changing (ULTC), the tap changer will try to raise the effective
load impedance ZD as seen from the system. This will lower the load bus voltage still further
and lead to a progressive reduction of voltage. This is the ease of simple voltage instability.
From the study of voltage stability the relationship between PL and V is important and this will
be discussed in the next lecture.
302
6.7
Neglecting the resistance of generator transformer and transmission line, the equivalent circuit of
the system and its phasor diagram are shown in Fig. 6.11 (a) and (b).
Figure 6.11: Equivalent circuit of the system(a) and the phasor diagram (b)
From the phasor diagram of Fig. 6.11 (b), it can be observed that IX cos = E sin and
IX sin = E cos V
IX cos EV
=
sin
X
X
PL (V ) = V I cos = V
(6.177)
QL (V ) = V I sin = V
IX sin EV
V
=
cos
X
X
X
EV 2
V2 2
2
[
] = [PL (V )] + [QL (V ) +
]
X
X
(6.178)
This static power-voltage equation determines all the possible network solutions when the voltage
characteristics PL (V ) and QL (V ) are taken into account.
For an ideally stiff load, the power demand of the load is independent of voltage and is constant
PL (V ) = PL and QL (V ) = QL , where, PL and QL are the real and reactive power demand of the
load at the rated voltage V.
For stiff load the equation (6.178) can now be written as:
EV 2
V2 2
2
[
] = [PL ] + [QL +
]
X
X
(6.179)
EV 2
V2 2
V2
=[
] [ ]
P + P tan + 2 PL tan
X
X
X
2
L
Substituting tan =
2
L
(6.180)
sin
and noting that sin2 + cos2 = 1 equation (6.180) can be further
cos
simplified as
V2
V2 2
P + 2 PL
sin cos = 2 (E V 2 ) cos2
X
X
2
L
(6.181)
2
V2
Adding and subtracting (
sin cos ) to the right hand side of equation (6.181), gives
X
2
V2
V2 2
V2
(PL +
sin cos ) ( ) sin2 cos2 = ( 2 ) (E 2 V 2 ) cos2
X
X
X
PL +
V
V2
sin cos =
cos E 2 V 2 cos2
X
X
(6.182)
The voltage at the load bus can be expressed in per unit as V/E. Equation (6.182) can be
expressed as:
E2 V 2
E2 V
PL = ( ) ( ) sin cos + ( ) ( ) cos
X
E
X
E
Or
Where p =
V 2
1 ( ) cos2
E
(6.183)
PL
V
, and v =
2
(E /X)
E
Equation (6.183) describes a family of curves with as a parameter. One such P-V curve is
shown, for a particular value of power factor cos in Fig. 6.12.
The Power-Voltage curve (PV-curve) presents load voltage as a function of load real power. For
304
Figure 6.15: Maximum deliverable power and loadabilty limit for polynomial load
V
This is shown in Fig. 6.15 for a polynomial load P = P0 ( ) , where, P0 represents the base
V0
value of load active power at rated voltage V0 = 1 p.u. and represents the voltage exponent. = 0
represents a constant power load.
Equation (6.183) when plotted for different values of gives a family of PV curves. Because of
306
From equation (6.183), when QL = 0 ( and hence = 0), p = v 1 v 2 . To find the value of PL
dp
= 0 and the solution of the resulting equation gives the values
dv 2
1
1
E
E2
of v = and p = . Hence, PL =
. Note that
is the short circuit power at the load bus,
2
2
2X
X
at the peak of the nose curve set
as it is the product of no load voltage E and the short circuit current (E/X). The maximum power
limit for a lossless line, with unity power power, thus corresponds to half the short circuit power.
PV-curves (nose curves), illustrate the dependency of the voltage on real power of a composite
load assuming that the power factor is a parameter. The QV curves discussed next are derived
assuming that the voltage is a parameter.
307
For given value of V, equation (6.179) describes a circle in the (PL QL ) plane as shown in Fig.
6.17 (a). The centre of the circle lies on the QL -axis and is shifted vertically down from the origin
by (V 2 /X) , the radius of the curve is (EV /X) . Increasing the voltage V produces a family of
circles of increasing radius and increasing downward shift, bounded by an envelope as shown in Fig.
6.17 (b).
Figure 6.17: QP curves for stiff load (a) one circle for a given V (b) family of curves for different
voltages and their envelope
For each point inside envelope, for example, point A, there are two possible solutions to equation
(6.179), at voltages V1 and V2 , as it lies on both the circles. For any point on the envelope, say point
B, there is only one value of V for which the equation (6.179) is satisfied. By determining the values
of PL and VL for which only one solution to equation (6.179) exists, the equation of the envelope
can be developed. Rearranging equation (6.179), one gets:
V2 2
E2 V 2
( ) + (2QL
) ( ) + (PL2 + Q2L ) = 0
X
X
X
(6.184)
E2 2
D = (2QL
) 4 (PL2 + Q2L ) = 0
X
(6.185)
QL = (
P2
E2
)[ 2L ]
4X
(E /X)
(6.186)
This represents the equation of an inverted parabola that crosses PL axis at E 2 /2X and has its
maximum at
308
PLmax = 0; QLmax =
E2
4X
(6.187)
Thus the maximum reactive power supplied to a load with PL = 0 i.e. a purely reactive load
is equal to one fourth of the short-circuit power. Also a point with coordinates PL = E 2 /2X and
QL = 0 corresponds to the peak of the nose curve for = 0 and QL = 0.
The parabola as described equation (6.186) defines the shape of envelope in Fig. 6.17 (b) that
encloses all the possible solutions to the network equation (6.179). Every point (PL , QL ) inside the
parabola satisfies two possible network solutions corresponding to two distinct values of load voltage
V, and each point on the parabola satisfies only one network solution corresponding to only one
possible value of voltage. Further, there are no network solutions outside the parabola implying that
it is not possible to deliver power for a (PL , QL ) point lying outside the parabola.
Various stability criteria to assess the system voltage stability are discussed in the next lecture.
309
6.8
With reference to Fig. 6.17 (b), for each point inside the envelope of network solution, there are
two voltage solutions, one with a higher value of voltage and the other with a lower value. It then
becomes necessary, to find out which of these two solutions represents a stable operating point. This
can be done by employing voltage stability criteria which are discussed next.
(a) The dQ/dV criterion
The classical voltage stability criterion is based on the capability of the system to supply reactive
power for a given amount of load real power demand. For the explanation of this criterion, it is
convenient to notionally separate the real and reactive power demands of the load as represented in
Fig. 6.18.
Figure 6.18: Equivalent circuit for determining the reactive power characteristics of the system
Let PL (V ) and QL (V ) be the load real and reactive power demands respectively.
Also, let PS (V ) and QS (V ) be the real and reactive powers supplied by the source to the load.
As the real power demand is always connected to the transmission link, PL (V ) = PS (V ) also
during normal operation QL (V ) = QS (V ) , but for the purpose of stability analysis, this link
between QL (V ) and QS (V ) is theoretically separated, hence, QS (V ) is treated as the reactive
power supplied to the load and is assumed to be independent of the load reactive power demand
QL (V ) .
The real and reactive load powers are given as
PL (V ) = PS (V ) =
EV
sin
X
(6.188)
and
QS (V ) =
EV
V2
cos
X
X
(6.189)
Squaring and adding the above two equations and using the identity sin2 + cos2 = 1, and
then solving for QS (V ) one can obtain,
310
EV 2
V2
2
QS (V ) = {
} {PL (V )} ( )
X
X
(6.190)
This equation determines the reactive power-voltage characteristic of a system, and shows how
much reactive power will be supplied by the source if the system is loaded only with the real power
PL (V ) and the load voltage is treated as a variable. For a constant power load PL (V ) = PL =
constant equation (6.186) takes the form of an inverted parabola as shown in Fig. 6.18. The first
term of the equation (6.186) depends on the equivalent system reactance X and the load real power
PL and has the effect of shifting the parabola downwards and towards the right.
EV
V2
)
and the parabola crosses the horizontal axis at V = E and
X
X
dQS (V )
V = 0. For finding the maximum value Qmax , set the derivative
= 0. On solving the
dV
E
E2
resulting equation, the value of V at which Qmax occurs is
and Qmax is equal to
.
2
4X
2
E 2
PL (V )X
Similarly, for PL > 0,the maximum values of QS (V ) occurs at voltage V =
( ) +(
)
2
E
E
which is greater than
2
For PL = 0, QS (V ) = (
Next, the QS (V ) and QL (V ) characteristics can be drawn on the same diagram as shown in
Fig. 6.20 (a). At equilibrium the supply must equal the demand i.e., QL (V )=QS (V ) and the two
possible equilibrium points VS and VU are obtained. This situation is similar to the one shown in
Fig. 6.17 (b).
The stability of the two equilibrium points can be evaluated using small perturbation method and
the fact that an excess of reactive power produces an increase in voltage while a deficit of reactive
power decreases the voltage.
Now, consider the equilibrium points of Fig. 6.20(a), and assume a small reduction in voltage
V . At point S, this reduction will result in the supplied reactive power QS (V ) being greater
than the reactive power demand QL (V ). This excess reactive power will try to increase the voltage
311
Figure 6.20: QS (V ) and QL (V ) Characteristics for (a) two equilibrium points S and U (b) the
classical stability representation
and therefore, force the voltage to return to point S. If there is a small increase in V , then at
point S the supplied reactive power QS (V ) becomes smaller than the load reactive power QL (V ).
This deficit in reactive power brings the voltage back to point S and thus, it can be concluded that
the equilibrium point S is a stable operating point.
At point U, the other equilibrium point, a small reduction in voltage results in QL (V ) becoming
greater than QS (V ). This deficit of reactive power, further reduces the voltage, and thus, the system
fails to return to the equilibrium point U. Hence, the system is unstable at point U. Similarly,
an increase in the voltage in the vicinity of U results in QS (V ) becoming greater than QL (V ).
This excess of reactive power increases the voltage further. Again the system fails to come back to
equilibrium point, and hence, the equilibrium point U is unstable.
From Fig. 6.20(b), it can be observed that at the two equilibrium points S and U the derivative
d (QS QL )
d
dQ
=
(QS QL ) < 0
dV
dV
or
dQS dQL
<
dV
dV
From Fig. 6.20(b), it can be clear that at point U,
(6.191)
dQS
dQL
>
, and hence, point U is
dV
dV
not stable as per the criterion established in equation (6.191). While at point S the condition of
equation (6.191) is satisfied and hence, it is a stable point.
For the simple system of Fig. 6.18, the supplied real and reactive powers are functions of two
variables V and
312
PS (V ) = PL (V ) = fP (V, )
(6.192)
QS (V ) = fQ (V, )
Hence, the incremental values of PS and QS can be expressed as :
PS
PS
V +
PL = PS =
(6.193)
QS =
QS
QS
V +
(6.194)
Substituting in the expression of QS of equation (6.193) and dividing both sides of the
resulting expression by V one gets
QS QS QS PS 1 PL PS
=
+
[
] [
]
V
V
V
V
(6.195)
dQS QS QS PS 1 dPL PL
+
[
] [
]
dV
V
dV
V
(6.196)
The partial derivatives are calculated from equation (6.188) and equation (6.189)
PS
PS
V
QS
QS
V
EV
cos
X
E
=
sin
X
EV
=
sin
X
E
V
=
cos 2
X
X
2V EV
X
dPL E
dQS E
cos
sin
[
sin ]
dV
X
X
X
EV cos dV
X
313
(6.197)
Figure 6.21: QS (V ) and QL (V ) Characteristics indicating stable and unstable operating points
E
2V dPL
QS
[
+
tan ]
V
Xcos
X
dV
(6.198)
dQL
E
2V dPL
>[
{
+
tan }]
dV
Xcos
X
dV
(6.199)
dPL
dQL
E(V ) =
PL (V )X
QL (V )X
(V +
) +(
)
V
V
2
314
(6.200)
QL (V )X
In this equation,
represents the voltage drop component in phase with V and
V
represents the voltage drop component in quadrature with V.
PL (V )X
V
A typical E(V) characteristic is shown in Fig. 6.22 with the normal operating point of the load
as S. V is large at this point and is much greater than both in phase and quadrature components of
the voltage drop. In this case then, a drop in voltage V will result in a drop in the emf E(V). As V
is reduced further, the in phase and quadrature components of the voltage drop become prominent
and below a certain value of V, they will force E(V) to rise. As a result each value of E(V) may
correspond to two possible network solutions of voltage V. The stability of these two solutions can
be examined using small perturbation method.
dE
dV
Let us examine the system behavior at point S as shown in Fig. 6.22 with the assumption
that source emf is maintained at a constant value. A reduction in the load voltage by V will
cause a reduction in the required value of the emf E(V). As the available E is constant and greater
than the required value of E(V), it will force the voltage to return to its initial value V. Thus, the
system returns to the initial equilibrium point after the disturbance. Similarly, when the voltage is
increased by V , the required emf E(V) to maintain the enhanced voltage (V + V ) is larger than
the available source emf E. Hence, the voltage is again forced to return to its initial value V by the
constant emf E. Thus, it can be safety established that the point S is a stable equilibrium point.
Next, consider the other equilibrium point U. As the voltage is reduced, a higher value of emf
E(V) is required to maintain it. But as E is constant and less than the required value of emf E(V),
this will result in further reduction in voltage. As a result, the voltage further reduces and moves
away from the equilibrium point. Similarly, if the voltage is increased by V , then, the required emf
E(V) is smaller than the available source emf E. This larger available emf E will cause the voltage
to increase further and move further away from the initial equilibrium point U. Hence, it can be
concluded that point U is an unstable equilibrium point.
From the above arguments, it is apparent that the system is stable if the equilibrium point lies
on the right hand side of the characteristics, that is when :
dE
>0
dV
315
dE
criterion of volatge stability
dV
Vcr is the minimum system voltage at which the system can be operated stably. The usual
system operation voltage is greater than Vcr .
(c) The dQS /dQL criterion
Let QS (V ) be the reactive power generation by the source and QL (V ) be the load demand.
Then, QS (V ) = QL (V ) +line reactive power loss. From Fig. 6.11(b) one can write:
(6.201)
Substituting IXsin = E cos V and IX cos = E sin in equation (6.201), one gets:
E
E
(E cos V ) cos + (E sin ) sin
X
X
E 2 EV
QS (V ) =
cos
X
X
QS (V ) =
(6.202)
Substituting the expression for QL (V ) from equation (6.177) in the above equation, we get:
QS (V ) =
E2 V 2
QL (V )
X X
Or
V 2 E2
=
QL (V ) QS (V )
X
X
(6.203)
Substituting this expression into equation (6.178) and rearranging the terms gives,
Q2S (V )
PL2 (V )
QL (V ) = 2
+ QS (V )
(E /X)
(E 2 /X)
316
(6.204)
dQS
>0
dQL
(6.205)
Further, at the maximum loading point at the nose of QS QL characteristic of Fig. 6.24, the
derivative dQG /dQL tends to infinity.
It is worth noting that the QG QL characteristic is a parabola only for ideally stiff real power
load PL (V ) = P =constant. For voltage dependant loads an explicit expression for QL (QG ) with
PL (V ) cannot be obtained, and an iteractive procedure is required.
317
318
Appendix A
The system data
Type
1
2
2
3
3
PG
QG
PL
QL
QM IN
QM AX
(MVAR)
0
-500
-500
-500
-500
(MVAR)
0
500
500
500
500
319
X (p.u)
0.168
0.126
0.126
0.126
0.210
0.252
B/2 (p.u)
0.041
0.031
0.031
0.031
0.051
0.061
Tx. Tap
0
0
0
0
0
0
Gsh
Bsh
(p.u) (p.u)
0
0
0
0
0
0
0
0
0
0
Type
1
2
3
3
3
2
3
3
3
3
3
3
3
3
PG
QG
PL
QL
QM IN
QM AX
(MVAR)
0
-500
0
0
0
-500
0
0
0
0
0
0
0
0
(MVAR)
0
500
0
500
0
500
0
0
0
0
500
0
0
0
320
X (p.u)
0.0592
0.223
0.1979
0.1763
0.1738
0.171
0.0421
0.209
0.5562
0.2522
0.1989
0.2557
0.1302
0.1762
0.011
0.0845
0.2703
0.192
0.1999
0.3479
B/2 (p.u)
0.0528
0.0492
0.0438
0.0374
0.0339
0.0346
0.0128
0
0
0
0
0
0
0
0
0
0
0
0
0
Tx. Tap
0
0
0
0
0
0
0
1
1
1
0
0
0
1
1
0
0
0
0
0
Gsh
Bsh
(p.u)
(p.u)
0
0
0
0
0.0002 0.502
0
0
0
0
0
0
0
0
0.0023 0.1325
0
0.0633
0
0
0
0
0
0
0
0
0
0
Type
1
2
3
3
2
3
3
2
3
3
2
3
2
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
PG
QG
PL
QL
321
QM IN
QM AX
(MVAR)
0
-500
0
0
-500
0
0
-500
0
0
-500
0
-500
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
(MVAR)
0
500
0
0
500
0
0
500
0
0
500
0
500
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
Gsh
Bsh
(p.u) (p.u)
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0.19
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0.04
0
0
0
0
0
0
0
0
0
0
0
0
322
X (p.u)
0.0575
0.1852
0.1737
0.0379
0.1983
0.1763
0.0414
0.116
0.082
0.042
0.208
0.556
0.208
0.11
0.256
0.14
0.2559
0.1304
0.1987
0.1997
0.1932
0.2185
0.1292
0.068
0.209
0.0845
0.0749
0.1499
0.0236
0.202
0.179
0.27
0.3292
0.38
0.2087
0.396
0.4153
0.6027
0.4533
0.2
0.0599
B/2 (p.u)
0.0528
0.0408
0.0368
0.0084
0.0418
0.0374
0.009
0.0204
0.017
0.009
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0.0418
0.013
Tx. Tap
0
0
0
0
0
0
0
0
0
0
1
1
0
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
Type
1
2
2
3
3
3
3
3
3
PG
QG
PL
QL
QM IN
QM AX
(MVAR)
0
-500
-500
-500
-500
-500
-500
-500
-500
(MVAR)
0
500
500
500
500
500
500
500
500
323
X (p.u)
0.0625
0.0576
0.0586
0.092
0.085
0.161
0.17
0.1008
0.072
B/2 (p.u)
0.0
0.0
0.0
0.079
0.088
0.153
0.179
0.1045
0.0745
Tx. Tap
1
1
1
0
0
0
0
0
0
Gsh
Bsh
(p.u) (p.u)
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0