73
CHAPTER 4
MODELLING OF UPFC AND STATCOM
FORMULTIMACHINE SYSTEM STABILITY
4.1 INTRODUCTION
Stability is a condition of equilibrium among opposing forces. The
method by which interconnected synchronous machines keep on synchronism
with another machine is through restoring forces, which acts when there are
forces tending to increase or decrease its speed on one or more machines with
respect to other machines. Instability in a power system is described by
depending upon the system configuration and operating mode. Generally, the
stability problem has been maintaining of synchronous operation. This aspect
of stability is influenced by the dynamics of machine rotor angles. But,
instability may also be encountered without loss of synchronism. For
example, a system can go unstable because of the collapse of load voltage.
Maintaining synchronism is not an issue in this instance; instead, the concern
is stability and control of voltages. In this work, the discussion is restricted to
voltage and rotor angle stability.
Rotor angle stability is the ability of interconnected synchronous
machines of a power system to remain in synchronism [2]. This stability
problem is concerned with the behavior of a synchronous machine after it has
been perturbed. Under steady state conditions, there is equilibrium between
the input mechanical torque and the developed electrical torque of each
machine. This equilibrium is upset during perturbation of the system. The
torque unbalance is caused by a change in load, generation or any other
network condition.
In any case, for the system to be stable all the machines must remain
operating in parallel and at the same speed. However, the statement declaring
74
the power system to be stable is not meaningful unless the conditions under
which this stability has been examined are clearly stated. This includes the
operating conditions as well as the type of perturbation (which can be large or
small) given to the system.
Transient stability is the ability of the power system to maintain
synchronism when subjected to a large disturbance [2]. The resulting system
response involves large excursions of generator rotor angles and is influenced
by the nonlinear power-angle relationship. Stability depends on both the
initial operating state of the system and the severity of the disturbance. This
disturbance is usually so large that it alters the post disturbance equilibrium
conditions relative to those existing prior to the disturbance. The work
presented in this thesis is focused on the power system behavior when
subjected to large disturbances and the enhancement of this stability using
FACTS controller.
The most common form of instability between interconnected
generators is loss of synchronism, monotonically, in the first few seconds
following a fault due to lack of synchronizing torque and damping torque.
The first step in a stability study is to make a mathematical model of
the system. The elements included in the model are those affecting the
machine. The complexity of the model depends upon the type of stability
study. Generally, the components of the power system that influence the
electrical and mechanical torques of the machines are included in the model.
Such components are the loads and their characteristics, the network during
the disturbance and the parameters of synchronous machines (such as inertia
of the rotating mass). Thus, the basic requirements for these studies are initial
conditions of the power system prior to the start of the disturbance and the
mathematical description of the main components of the system that might
affect the behavior of synchronous machines.
75
Generally,differential equations are used to describe the various
components. The system equations for small signal stability analysis are
usually nonlinear. The behavior of any dynamic system, such as a power
system, is described by a set of n first order non-linear differential equations
of the form given by (4.1).
t~ ,u~ ,x~f=•~x (4.1)
In Equation (4.1), f is a vector of nonlinear functions. The column
vector~x is referred to as a state vector and
~u is the vector of inputs to the
system. The study of dynamic behavior of the system is based on the nature of
these differential equations.
The first step is to do a load flow study as discussed [2], to obtain the
initial steady state conditions. After establishing initial conditions, a
mathematical model of the power system is formed as discussed in [4]. In this
model the effects of AVR and PSS are taken into account except the effects of
governor. The mathematical model obtained is set of non-linear differential
equations. Solving these equations by using Runge-Kutta method the state
variables are determined. After giving a large disturbance for a particular
period the system is again restored and the behavior of the system is studied at
the same time the effect of change in load and change in mechanical input
power of the system is also studied.
4.2 SYSTEM MODELLING
This section presents the mathematical models used for the power
system components such as generator, exciter and PSS. Mathematical
modeling of STATCOM and UPFC is also discussed in this section. Feed
Back Linearizing Controller (FBLC) modeling and implementation to
76
STATCOM and UPFC is also described in this chapter. Further tuning of
FBLC is done using two intelligence techniques Bacterial Foraging Algorithm
and Differential Evolution is discussed.
4.2.1 Synchronous Machine Model
Synchronous machine is represented by means of the single-axis model
[4]. The state variables are rotor angle i , rotor angular velocity, and the
voltage proportional to main field flux linkage, 'qE . The sub transient
reactance, saturation and turbine governor dynamics are neglected.
In developing equations for a mathematical model of a multi-machine
power system, the following assumptions were made
1. Mechanical power input Pm is constant.
2. The mechanical rotor angle of a machine coincides with the angleofthe
voltage behind the transient reactance.
3. Loads are represented by passive impedance.
The resulting differential – algebraic equations for the ‘m’ machine,
‘n’ bus system with exciter model is given below as the state equations in p.u.
dididiqifdiqi ixxEE
dtdE
)'(''
T'doi (4.2)
midt
di
i ,.....,11 (4.3)
miiixxiETdt
djqididiqiqiqmi
i ,...,1/})''('{ (4.4)
77
Where j - fHs
id - direct axis current
iq - quadrature axis current
xd - direct axis synchronous reactance
xd’ - direct axis transient reactance
xq - quadrature axis synchronous reactance
xq’ - quadrature axis transient reactance
Eq’ - Voltage proportional to main field flux linkage
T’do - direct axis open circuit time constant
Efd - Equivalent stator emf corresponding to field voltage
Hs - inertia constant of synchronous machine
Tm - Mechanical torqueof synchronous machine
Equation (4.4) has dimensions of torque in per-unit.When the stator
transients were neglected, the electrical torque became equal to the per-unit
power associated with the internal voltage source. The dynamic performances
of STATCOM and UPFC have been analyzed with different types of
disturbance thus damping has not been included in equation 4.4, because
STACOM and UPFC can improve the damping of the system. The system
data have been given in Appendix 3.
78
4.2.2 Algebraic Equation The stator algebraic equation can be written in the form of
m1,...,i
ejEix-xE-
)eji)(ijx(ReV0
)2
-j'qiqi
'di
'qi
'di
)2
-j
qidi'disii
i
ii
(4.5)
Where 'diE = 0, since one axis model has been considered.
4.2.3 Excitation System and PSS Model
IEEE Type 1S [4] excitation system model is considered in the sample
system. The block diagram of the excitation system with PSS is shown in
Figure 4.1. The state equations are given below.
1V•
= (Vt-V1)/TR (4.6)
Efd = (KAVe- Efd)/TA (4.7)
3
•
V = {[KF(KAVe-Efd)/TA]-V3}/TF (4.8)
Where Ve - VREF+VF-V1-V3+ upss (4.9)
Vt - terminal voltage of synchronous machine
V1 - output signal of filter
V3 - output signal of stabilizing circuit
VR - regulator output signal
VF - supplementary stabilizing signal
VREF - regulator reference voltage
Ve - error voltage
79
KA - gain of amplifier excitation system
KF - gain of stabilizing feedback
TF - time constant of stabilizing feedback
TA - time constant of amplifier of excitation system
TR - regulator time constant
upss - PSS output signal.
The PSS is represented by a washout filter and a cascade of lead lag
controllers of the following form
2
1
11
1 sTsT
sTwsTwKU stabpss (4.10)
Where, the rotor speed deviation is taken as the input to the PSS. Kstab, T1
and T2 are stabilizer gain and time constants respectively.
The parameters Kstab, T1 and T2 are to be determined to enhance the
system damping for the electromechanical mode.Practically, the washout
block has little phase compensation effect and its time constant Tw is fixed to
ten seconds in advance. The Exciter, PSS data for 3 machine, 9-bus system
are given in Table A 3.5 and Table A 3.6 of Appendix 3.
Figure 4.1Block diagram of IEEE Type-1s Excitation model with PSS
80
VjQ-P
=y 2i
LiLiLi
4.2.4 Internal Node Model
This is a widely used reduced-order multi-machine model in first-
swing transient stability analysis. In this model, the loads are assumed to be
constant impedances and converted to admittances as [41]
n1,....,i (4.11)
Where Liy load admittance at ith bus, Vi – Voltage at ith bus
LiP - real power load at ith bus and LiQ - reactive power load at ith bus.
There is a negative sign for Liy , since loads are assumed as injected
quantities.
(4.12)
Where,
'
1
dijXDiagy i = 1,2,….., m (4.13)
m -no of machines
and 000
1yYY NN (4.14)
y y 0
y1NY
0
augY
81
If transmission line resistances are neglected, then the network
admittance matrix is
ijN BjY (4.15)
Where Bij=suspectance betweenith and jth bus.
Adding Liy to the diagonal elements of the N1Y matrix and makes it )(12 LiNN yDiagYY .
The modified augmented Y matrix becomes
(4.16)
Figure 4.2 Augmented Y matrix with constant impedance
y y 0
y 2NY
0
augnewY
82
The passive portion of the network is shown in Figure 4.2. The
network equations for the new augmented network can be written as
B
A
DC
BAA
VE
YYYY
0I (4.17)
where yYA ,
0YB y ,
0y-YC and 2DY NY
The n network buses can be eliminated, since there is no current
injection at these buses. Thus
ACD1
BAA EYYYYI (4.18)
AintA EI Y
where ,the elements of AI and AE are respectively,
QiDi2
j
qidii jIIejIIIi
and iii EE . (4.19)
Where i = 1,…,m.
intY = CD1
BA YYYY (4.20)
The elements of intY are ijijij jBGY . Since the network buses have
been eliminated, the internal nodes are such as 1,.., m, for ease of notation.
j
1i EI
m
jijY
83
i= 1, ……, m (4.21)
Real electrical power out of the internal node ‘i’ from Figure 4.2 is given by
i*
iei IEReP
m
1j
*j
*ijiei EYeEReP i
m
1j
-jijijiei
ji eEBGeEReP j
m
1jjiijijei sincosEEBGReP jiji jj
(4.22)
4.3 INTERFACING OF STATCOM AND UPFC
For 3 machines, 9-bus system the STATCOM has been connected at
8th bus, which is a load bus and UPFC has been connected between line 7-8
which is shown in Figure 4.3 and 4.4
84
Figure 4.3 Line diagrams of 3 machines, 9-bus system with STATCOM
Figure 4.4 Line diagrams of 3 machines, 9-bus system with UPFC
4.3.1 Modelling of UPFC for Multi Machine System
The mathematical model of the UPFC is derived here in the d-q
(synchronously rotating at the system angular frequency ) frame of
reference. This is followed by a detailed description of the conventional PI
control strategy employed for active and reactive power control using UPFC.
The equivalent circuit model of a power system equipped with a UPFC
is shown in figure A3.2 inAppendix 3. The series and shunt VSIs are
represented by controllable voltage sources Vc and Vp, respectively.Rp and Lp
represent the resistance and leakage reactance of the shunt transformer
respectively. Leakage reactance and resistance of series transformer have
been neglected.
The mathematical model of UPFC is derived by performing standard
d-q transformation [4] of the current through the shunt transformer and series
85
transformer. They are as given below ( is the angular frequency of the
voltages and currents).
4.3.2 Modeling of Shunt Converter
The dynamic equations governing the instantaneous values of the
three-phase voltages across the two sides of STATCOM and the current
flowing into it are given by [40]
R + L i = V V (4.23)
= R + V V (4.24)
Where:i = i i i , V = [V V V ] , V = V V V
= 0 0
0 00 0
and = 0 0
0 00 0
Under the assumption that the system has no zero sequence
components, all currents and voltages can be uniquely represented by
equivalent space phasors and then transformed into the synchronous d-q-o
frame by applying the following transformation (q is the angle between the d-
axis and reference phase axis):
=
cos cos +
sin sin sin + (4.25)
Thus, the transformed dynamic equations are given by,
= R + V V + i (4.26a)
86
= R + V V i (4.26b)
Where, is the angular frequency of the AC bus voltage.
4.3.3 Cascade Control Strategy for Shunt Converter
The conventional control strategy for this inverter concerns with the
control of ac-bus and dc-link voltage. The dual control objectives are met by
generating appropriate current reference (for d and q axis) and, then, by
regulating those currents. PI controllers are conventionally employed for both
the tasks while attempting to decouple the d and q axis current
regulators.In this study, the strategy adopted in [40] [41] for shunt current
control has been taken.The inverter current ( pi ) is split into real (in phase with
ac-bus voltage) and reactive components.The reference value for the real
current is decided so that the capacitor voltage is regulated by power
balance.The reference for reactive component is determined by ac-bus voltage
regulator.As per the strategy, the original currents in d-q frame )i ,i( pqpd are
now transformed into another frame, qd frame, where d axis coincides
with the ac-bus voltage (Vs), as shown in Figure 4.5.
Figure 4.5 Phasor diagram showing d-q and d’-q’ frame
Thus, in qd frame, the currents dpi and qpi represent the real and
reactive currents and they are given by:
87
= cos + sin (4.27)
= cos sin (4.28)
Now, for current control, the same procedure as outlined in [57] has
been adopted by re-expressing the above differential equations as:
= R + V V + i (4.29)
= R + V i (4.30)
Where
= cos + sin (4.31)
= cos sin (4.32)
= +
The VSI controlled voltages are as follows:
V = L i + L u (4.33)
V = L i + V L u (4.34)
By putting the above expressions for dpV and qpV in equations (4.29)
and (4.30) the following set of decoupled equations are obtained.
= i + u (4.35)
= i + u (4.36)
88
Conventionally, the control signals du and qu are determined by
linear PI controllers. The complete cascade control architecture is shown
below in Figure 4.6, where dpqiqpicpcitpt K,K,K,K,K,K,K and diK are the
respective gains of the PI controllers.
In this study, the above design has been used for demonstration of
STATCOM control.This approach leads to good control as illustrated by the
simulation results.
The final and important stage of the design of PI based STATCOM
involves tuning of parameters of STATCOM, which is posed as an
optimization problem. In this problem the optimal output gain K0 are
determined by maximizing the damping out of transient voltage oscillations of
the load bus voltage and dc capacitor voltage being controlled. This is in
effect carried out by minimizing Sum Squared Deviation (SSD) of the load
bus voltage and dc capacitor voltage being controlled from the desired value
through non-linear simulation of power system under typical operating
condition and disturbance. The non-linear simulation is carried out using a
Transient Stability Algorithm [4] employing a Runge-Kutta fourth order
method. To get the original currents it is again transformed to d -q
Figure 4.6 PI-Control Structure of STATCOM
4.3.4 Modeling of Series Converter
= + ( sin ) (4.37)
89
= + cos (4.39)
For fast voltage control, the net input power should instantaneously
meet the charging rate of the capacitor energy. Thus, by power balance,
= + + + ( + )
= Vdcidc= +
= + + + ( ) + (4.40)
An appropriate series voltage (both magnitude and phase) should be
injected for obtaining the commanded active and reactive power flow in the
transmission line, i.e., uu Q ,P in this control. The current references are
computed from the desired power references and are given by,
= (4.41)
= (4.42)
The power flow control is then realized by using appropriately
designed controllers to force the line currents to track their respective
reference values. Conventionally, two separate PI controllers are used for this
purpose. These controllers output give the amount of series injected voltages
)V ,V( cqcd . The corresponding control system diagram is shown in Figure 4.7.
P I cqi
Equations (4.39) & (4.40)
sK
K idpd
sK
K iqpq
P I
refP
refQ
refcqi
refcdi
cdi
+
_
_
+
cdV
cqV
90
Figure 4.7 PI Control of Series Converter
4.4 IMPLEMENTATION OF FBLC FOR UPFC IN SMIB
ANDMULTI MACHINE
4.4.1 Implementation of FBLC for UPFC in SMIB
In this section, the design steps for the feedback linearizing control of
UPFC have been presented followed by simulation results under various
transient disturbances. A brief review of nonlinear control using feedback
linearization is presented in the Appendix 2.
4.4.2 FBLC design
In UPFC control, there are four objectives. They are (i) active power
control, (ii) reactive power control, (iii) ac-bus voltage (Vs) control and (iv)
dc link voltage (Vdc) control. Tracking of active and reactive power are
indirectly translated to tracking of line currents to their respective reference
values computed from Pref and Qref.The differential equations of the line
currents (ibd, ibq) and Vdcare already derived. Thus, for control of Vs, its
differential equation need to be derived, i.e, Vsis taken as an additional state
in this control design.
Now, for the control design, the complete state space model is
expressed in the form of Equations. (A.1) and (A.2) in Appendix 1 as follows:
91
X=
123456
= (4.41)
= =
=
=
=
=
= =
Thus,
= ( ) + (4.42)
= ( ) + (4.43)
= ( ) + (4.44)
= ( ) + (4.45)
= ( ) (4.46)
= ( ) + + + + ( ) (4.47)
Where:
1( ) +1
Lp(Vsd) (4.48)
92
2( ) +1
Lp(Vsq) (4.49)
=1
=
3( ) +1
Le(Vsd Vbd) (4.50)
4( ) +1
Le)(Vsq Vbq) (4.51)
=1
=
5( ) = 1 + 2 (4.52)
51( ) =1
C[ Vsd + Vsq + (Vsd Vud) + (Vsq Vuq) )]
1 =1
C[Vsd 1 + Vsq 2 + ( Vud + Vsd) 3f3 + ( Vuq + Vsq) 4)
+ ( dVsd + dVsq + (dVsd dVud)+ (dVsq dVuq)]
C2 = 51[(1/C( ))[( (Vsd + Vsq + (Vsd Vud)+ (Vsq Vuq) )]
= ( 1)
= ( 2)
=1
( + ) 3
=1
( + ) 4
93
6( ) =VsdVs
(dvtd Re f1 + xt1 f2 Re f3 + xt1 f4)VsqVs
(dVtq + Re f2 + xt1 f1 + xt1 f3 + Re f4)(4.53)
1( 1 + 1 1)
=1
( 2)
1( 3 + 3)
=1
( 4)
The outputs of the system are: , , ,and Vdc..That is
Proceeding with the exact steps as outlined in the Appendix, the following
can be derived: =
3( )4( )6( )5( )
+
0 0 00 0 0
(4.54)
= A(x) + E(x) u
94
= ( ) + ( ) (4.55)
Thus:
= ( ) ( ) + (4.56)
The non-singularity of E(x) can be observed by computing the
determinant of E(x).
E(x) is nonsingular in the operating ranges of and Vdc. For tracking
of and Vdc, the new control inputs v1, v2, v3 and v4 are selected as (by both
proportional and integral control):
=
+ +
+ +
+ +
+ + +
(4.57)
Where y1ref is the ,y2ref is the ,y3ref isthe , ac-bus reference
voltage (Vsref) and y4ref is the dc-bus reference voltage (Vdref) and e1 and e2
are error variables defined by
= (4.58)
= (4.59)
95
= (4.60)
= (4.61)
The gain parameters K11, K12, K21, K22,K23are determined by assigning
desired poles on the left half s-plane and, thus, asymptotic tracking control to
the reference can be achieved.
From u1 u2 u3 and u4, the control signals in d'-q' frames are determined by:
(4.62)
(4.63)
(4.64)
(4.65)
Again, from Vpd' and Vpq', the control signals in d-q frame isVpdand
Vpq. In the computer simulation studies presented in the followings, the
derivative dvtd’/dtappearing in the control design, is neglected in the control
computation. This leads to the assumption that the generator bus voltage Vtis
treated as a constant only for the control design.
In this section, the design steps for the feedback linearizing control of
UPFC have been presented followed by simulation results under various
transient disturbances. A brief review of nonlinear control using feedback
linearization is presented in the Appendix 2.
4.4.3 Implementation of FBLC for STATCOM in Multi Machine
As mentioned earlier, in the STATCOM control, there are two broad
objectives, i.e., ac-bus voltage )V( s and dc-bus voltage )V( dc control.In the
following control design,
96
sV is taken as an additional state in addition to the other three states
,i,i( qpdp and )Vdc in the STATCOM modeling.
8VVs (4.66)
Now, for the control design, the complete state space model is
expressed in the form of equations (A.1) and (A.2) as follows.
qp
dp
dc
qp
dp
VV
uu
VVii
xxxx
2
1
84
3
2
1
,
ux (4.67)
= ( ) + (4.68)
= ( ) + (4.69)
(x)fx 33 (4.70)
= ( ) + + (4.71)
where,
( )= + + (4.72)
= = (4.73)
( ) (4.74)
( ) = (4.75)
8(x)f4 V (4.76)
97
, (4.77)
The outputs of the system are sV and Vdc
Thus, y1and y2 =Vdc
Proceeding with the exact steps as outlined in the Appendix 2, the
following can be derived.
= ( )
( ) ( ) + ( ) + ( )+
+ +
=A(x) +E(x) (4.78)
=1
2
where,
= (4.79)
(4.80)
= ( + ) (4.81)
= (4.82)
Thus, = ( ) ( ) + (4.83)
The no singularity of )x(E can be observed by computing the
determinant of )x(E , which is:
98
| | (4.84)
It is known that the magnitude of current 'pdi is very small such that
2 'pdpiR2 is almost negligible compared to Vs.Now it is readily seen that
E(x)is nonsingular in the operating ranges of sV and dcV .
For tracking of and , the new control inputs and are selected
as (by both proportional and integral control):
=+ +
+ + + (4.85)
Where y1refis the Ac bus reference voltage ( ) and y2ref is the dc bus
reference voltage ( )
and e1 and e2are error variables defined by :
= and
= (4.86)
From equation (4.86), the error dynamics are given by :
+ + = 0 (4.87)
+ + + = 0 (4.88)
The gain parameters ,K,K,K,K 22211211 and 23K are determined by
assigning desired poles on the left-half s-plane and, thus, asymptotic tracking
control to the reference can be achieved.From 1u and 2u , the control signals in
qd frame are determined by,
99
` and
` (4.89)
Again, from dpV and qpV , the control signals in qd frame i.e., pdV and
pqV , are obtained by making use of equations (4.31) and (4.32).
4.5 SAMPLE SYSTEM AND RESULTS FOR SMIB USING UPFC
The performance of the UPFC with PI controller and FBLC for
synchronous generator stabilization is evaluated by MATLAB simulation
studies. In this simulation UPFC has been connected to load bus of SMIB.
The system data are provided in Appendix 3. The following case studies are
undertaken for evaluating the performance of the proposed controller.
Case 1
The synchronous generator is assumed to operate at P = 1.2 p.u. and Q
= 0.85 p.u.A 3-phase fault occurs near the infinite bus for duration of 100 ms.
Figure 4.8 shows the transient response of the system with
conventional PI based UPFC and FBLC based UPFC. The transient
performances of the rotor angle, rotor speed deviation are shown for three
phase fault when the generator is operating at P = 1.2 p.u. and Q = 0.85 p.u. A
comparison of the system responses for a 3-phase fault at infinite bus which is
cleared after 100 ms is also shown. This study clearly indicates better
stabilizing properties of UPFC, particularly the restoration of bus voltages to
the pre-disturbance value and the performance of dc capacitor voltage Vdc for
the above case. The transient oscillations in rotor angle and speed
deviationexhibit good damping behaviour for FBLC compared to cascade PI
controllers which has been presented. This is possible because of non-linear
100
control of bus voltage, resulting in better power modulation, by FBLC
controller for stabilizing the synchronous generator.
Fig 4.8 Comparison of transient performances for Case 1 with PI based
UPFC ( ) and FBLC based UPFC ( )
Case 2
101
For this case The synchronous generator is assumed to operate at a
load of P=1.0 p.u. and Q = 0.5 p.u.The load values have been reduced by 50%
for duration of 100 ms.
The transient oscillations in rotor angle and speed exhibit good
damping behaviour for FBLC compared to cascade PI controllers which has
been presented in Figure 4.9. This is possible because of non-linear control of
bus voltage, by FBLC controller for stabilizing the synchronous generator.
The single line diagram of this system and the data of this system is given in
Appendix 3.
Figure 4.9 Comparison of transient performances for Case 2 with PI based UPFC () and FBLC based UPFC ( )
102
4.6 SAMPLE SYSTEM AND RESULTS FOR MULTI
MACHINEUSING STATCOM AND UPFC
A multi-machine system namely 3 machines, 9-bus system [4] is
considered for analysis. The single line diagram of this system is shown in
Figure 4.3 and 4.4 and the data of this system is given in
Appendix 2. In this system number of machines m = 3 and number of busn =
9. The system is subjected to various disturbances in order to identify the
critical cases. From the investigations the following cases are considered to be
critical cases. Initially base case load flow was run for the sample 9-bus
system, which results are given in Table 2.8 of Appendix 3. After getting the
initial values from load flow, the non-linear simulation is carried using
MATLAB.
Case 1 A three phase fault of 0.25 second duration is created at the middle of
the transmission line connecting bus 5 and bus 7. The performance of the
conventional PI controller with STATCOM and without STATCOM in the
network isanalysed.
The Figure 4.10 represents voltage at 8th bus with PI based STATCOM
and FBLC based STATCOM for the variation of dc capacitor voltage of
STATCOM. The dc capacitor voltage is a very important factor for successful
operation of the shunt converter.
Figure 4.10 Comparison of transient response of voltage at8th bus for case 1, With PI STATCOM ( ), with FBLC STATCOM ( )
103
Case 2
The load values have been reduced by 50% for duration of 100
ms.. The performance of the conventional PI controller with UPFC and
without UPFC in the network is shown in Figure 4.11,Figure 4.12 and Figure
4.13 respectively and represents the relative rotor angle oscillation,the speed
deviation and changes in voltage at 8th bus, the variation of dc capacitor
voltage of UPFC. The dc capacitor voltage is a very important factor for
successful operation of the shunt converter.
Figure 4.11 Comparison of transient response of relative rotor angle for case 2, without UPFC ( ), with PI based UPFC ( )
104
Figure 4.12 Comparison of transient response of voltage at 8th bus for case 2, without UPFC ( ), with PI based UPFC ( )
Figure 4.13 Comparison of transient response of speed deviation for case 3, without UPFC ( ), with PI controller based UPFC ( )
105
Case 3:The mechanical torque input of generator 1 is suddenly
increased by 50% for 100 ms duration. The oscillations are presented in the
figure below respectively for without with STATCOM PI control and FBLC
control schemes. The superiority of FBLC over PI is well marked in damping
oscillations in Figures 4.14 and 4.15
Figure 4.14 Comparison of transient response of relative rotor angle for case 3, FBLC based STATCOM ( ), With PI based STATCOM ( )
Figure 4.15 Comparison of transient response of speed deviation For case 3, FBLC based STATCOM ( ), With PI based STATCOM ( )
106
4.7 BACTERIAL FORAGING OPTIMIZATION ALGORITHM
In this sectionBacteria Foraging Optimization Algorithm, is used for
tuning the parameters of FBLC controller. The simulation is done for FBLC
controller with STATCOM and the results are discussed. BFOA [68]
proposed by Passino, is a new comer to the family of nature-inspired
optimization algorithms. For over the last five decades, optimization
algorithms like Genetic Algorithms, Evolutionary Programming and
Evolutionary Strategies which draw their inspiration from evolution and
natural genetics, have been dominating the realm of optimization algorithms.
Recently natural swarm inspired algorithms like Particle Swarm
Optimization, Ant Colony Optimization have found their way into this
domain and proved their effectiveness. Following the same trend of swarm-
based algorithms, Passino[71] [72] proposed the BFOA.
Bacterialforaging optimization algorithm has been widely accepted as
a global optimization algorithm of current interest for distributed optimization
and control. BFOA is inspired by the social foraging behavior of Escherichia
coli. BFOA has already drawn the attention of researchers because of its
efficiency in solving real-world optimization problems arising in several
application domains. The biology behind the foraging strategy of ‘E, coli’is,
emulated in an extraordinary manner and used as a simple optimization
algorithm.
Application of group foraging strategy of a swarm of ‘E, coli’bacteria
in multi-optimal function optimization is the key idea of the new algorithm.
Bacteria search for nutrients in a manner to maximize energy obtained per
unit time. Individual bacterium also communicates with others by sending
signals. A bacterium takes foraging decisions after considering two previous
107
factors. The process, in which a bacterium moves by taking small steps while
searching for nutrients, is called chemotaxis and key idea of BFOA is
mimicking chemotactic movement of virtual bacteria in the problem search
space.
It possess four process
Figure 4.16Four Process In BFOA
4.7.1 Chemotaxis
The process, in which a bacterium moves by taking small steps while
searching for nutrients, is called chemotaxis. This process simulates the
movement of an ‘E, coli’cell through swimming and tumbling via flagella.
Biologically an ‘E, coli’bacterium can move in two different ways. It can
swim for a period of time in the same direction or it may tumble, and alternate
between these two modes of operation for the entire lifetime.
CHEMOTAXIS
SWARMING
REPRODUCTION
ELIMINATION & DISPERSAL
108
Figure 4.17 SWIM & TUMBLE OF A BACTERIUM
During foraging of the real bacteria, locomotion is achieved by a set of
tensile flagella. Flagella help an ‘E, coli’bacterium to tumble or swim, which
are two basic operations performed by a bacterium at the time of foraging.
When they rotate the flagella in the clockwise direction, each flagellum pulls
on the cell. That results in the moving of flagella independently and finally
the bacterium tumbles with lesser number of tumbling whereas in a harmful
place it tumbles frequently to find a nutrient gradient. Moving the flagella in
the counterclockwise direction helps the bacterium to swim at a very fast rate.
In the above-mentioned algorithm the bacteria undergoes chemotaxis, where
they like to move towards a nutrient gradient and avoid noxious environment.
Figure 4.17 depicts how clockwise and counter clockwise movement of a
bacterium take place in a nutrient solution.
4.7.2 Swarming
An interesting group behavior has been observed for several motile
species of bacteria including ‘E, coli’andS. typhimurium, where intricate and
stable spatio-temporal patterns (swarms) are formed in semisolid nutrient
medium. A group of ‘E, coli’ cells arrange themselves in a traveling ring by
SWIM TUMBLE
109
moving up the nutrient gradient when placed amidst a semisolid matrix with a
single nutrient chemo-effecter. The cells when stimulated by a high level of
succinate, release an attractant aspertate, which helps them to aggregate into
groups and thus move as concentric patterns of swarms with high bacterial
density.
4.7.3 Reproduction
When they get food in sufficient, they are increased in length and in
presence of suitable temperature they break in the middle to from an exact
replica of itself. This phenomenon inspired Passino to introduce an event of
reproduction in BFOA. The least healthy bacteria eventually die while each of
the healthier bacteria (those yielding lower value of the objective function)
asexually split into two bacteria, which are then placed in the same location.
This keeps the swarm size constant.
4.7.4 Elimination and Dispersal
Due to the occurrence of sudden environmental changes or attack, the
chemotactic progress may be destroyed and a group of bacteria may move to
some other places or some other may be introduced in the swarm of concern.
This constitutes the event of elimination-dispersal in the real bacterial
population, where all the bacteria in a region are killed or a group is dispersed
into a new part of the environment.
Gradual or sudden changes in the local environment where a bacterium
population lives may occur due to various reasons e.g. a significant local rise
of temperature may kill a group of bacteria that are currently in a region with
a high concentration of nutrient gradients. Events can take place in such a
fashion that all the bacteria in a region are killed or a group is dispersed into a
new location. To simulate this phenomenon in BFOA some bacteria are
110
liquidated at random with a very small probability while the new
replacements are randomly initialized over the search space.
In a three machine nine bus system, the 3 phase symmetrical fault
occurs in the line connecting the buses 5 and 7, for the duration of 0.25s. The
fault is created at 1s and removed at 1.25s. STATCOM is connected at the 8th
bus. The controllers used are PI controller and FBLC. First PI controller is
used and then FBLC is used to improve the output. The following combined
output waveforms show the comparison of transient responses of STATCOM
in presence of two different controllers.
In order to get better output, the control parameters of FBLC are tuned
by using Bacterial Foraging Optimization Algorithm (BFOA). There are five
control parameters namely K11, K12, K21, K22 & K23.
These control parameters are determined by using two poles namely
Pole1 and Pole2. The formula for calculating the control parameters are given
below:
K11 = 2 x Pole1
K12 = Pole1 x Pole1
K21 = 3 x Pole2
K22 = 3 x Pole2 x Pole2
K23 = Pole2 x Pole2 x Pole2
These two poles are determined by using BFOA. The value for Pole1
is 0.03 and the value for Pole2 is -28.
For the same duration of 0.25s three phase symmetrical fault is created
at multi-machine system as that of previous case and STATCOM is connected
111
at 8th bus. Here FBLC controller is only used and is tuned by using BFOA to
get better output. The simulations are carried out using MATLAB software.
The following combined output waveforms show the comparison of
transient responses of STATCOM in presence of generalized FBLC and
FBLC tuned by BFOA.
The figure 4.18 shows the comparison of rotor angle curves of
generator 1. This output waveforms clearly show that the critical clearing time
(CCT) is 4.5s in presence of STATCOM with FBLC and CCT is 4.3s in
presence of STATCOM with FBLC when tuned by BFOA. Thus the CCT got
reduced by using BFOA and hence the transient stability of the multi-machine
power system got enhanced.
Similarly the comparison of rotor angle curves of generator 2 and
generator 3 are shown in the figures 4.19 and 4.20 respectively
Then the voltage curves at the 8th bus are compared and are shown in
the Figure 4.21 Finally the DC voltage curves of STATCOM with FBLC and
then tuned by BFOA are compared.
From all the above waveforms it is clearly understood that by using
STATCOM three phase symmetrical fault gets cleared at faster rate and also
CCT in presence of FBLC tuned by BFOA is less than that of generalized
FBLC.
Hence STATCOM with FBLC tuned by BFOA damp out the rotor
oscillations quickly and enhance the transient stability of the multi-machine
power system at a faster rate.
112
Figure 4.18 Comparisons of Rotor Angle Curves of Generator 1
Figure 4.19 Comparisons of Rotor Angle Curves of Generator 2
----- : STATCOM with FBLC
___ : STATCOM with FBLC tuned by BFOA
----- : STATCOM with FBLC
___ : STATCOM with FBLC tuned by BFOA
113
Figure 4.20 Comparisons of Rotor Angle Curves of Generator 3
Figure 4.21 Comparisons of Voltage Curves at 8th Bus
----- : STATCOM with FBLC
___ : STATCOM with FBLC tuned by BFOA
----- : STATCOM with FBLC
___ : STATCOM with FBLC tuned by BFOA
114
4.8 DIFFERENTIAL EVOLUTION
The constants of the controller for STATCOM can be tuned by
different evolutionary algorithms for the purpose of improved performance.
Differential evolution (DE) algorithm is arguable one of the most powerful
stochastic real parameter optimization algorithm in current use. DE operates
through similar computational steps as employed by a standard evolutionary
algorithm (EA). How- ever, unlike traditional EAs, the DE-variants perturb
the current- generation population members with the scaled differences of
randomly selected and distinct population members.
Therefore, no separate probability distribution has to be used for
generating the offspring. Since its inception in 1995, DE has drawn the
attention of many researchers all over the world resulting in a lot of variants
of the basic algorithm with improved performance. This section presents a
detailed review of the basic concepts of DE and a survey of its major variants,
its application tomulti objective, constrained, large scale, and
uncertainoptimization problems,
andthetheoreticalstudiesconductedonDEsofar.Also, it provides an overview of
the significant engineering applications that have benefited from the powerful
nature of DE.
Differential evolution is an attractive optimization tool because of the
following reasons
Compared to most other EAs, DE is much simpler to implement end
easy to code in any programming language.
DEexhibitsmuch better
The number of control parameters in DE is very few (Cr, F, and NP in
classical DE).
115
The space complexity of DE is low and this feature helps in extending
DE for handling large scale and expensive optimization problems.
Differential evolution uses three main control parameters to optimize
the objective function used. They are Cr which is the crossover ratio, F which
is the mutation constant and NP which denotes the number of population
members.
4.8.1 Stages of Differential Evolution
The main stages of DE are as follows and shown in Figure 4.23
Initialization of vectors
Difference vector based mutation
Crossover/Recombination
Selection
Figure 4.22 Stages of Differential Evolution Algorithm
4.8.2 Initialization of the Parameter Vectors
DE searches for a global optimum point in a D-dimensional population
of NP D dimensional real-valued parameter vectors. Each vector, also known
as genome/chromosome, forms a candidate solution to the multidimensional
optimization problem.
116
4.8.3 Mutation with Difference Vectors
Biologically, “mutation” means a sudden change in the gene
characteristics of a chromosome. In the context of the evolutionary computing
paradigm, however, mutation is also seen as a change or perturbation with a
random element. In DE-literature, a parent vector from the current generation
is calledtargetvector,amutantvectorobtained through the differential mutation
operation is known as donor vector and fin0 ally an offspring formed by
recombining the donor with the target vector is called trial vector.
4.8.4 Crossover
To enhance the potential diversity of the population, a crossover
operation comes into play after generating the donor vector through mutation.
There are two kinds of crossover strategies usually used
Exponential Crossover
Binary Crossover
4.8.5 Selection
To keep the population size constant over subsequent generations, the
next step of the algorithm calls for selection to determine whether the target or
the trial vector survives to the next generation
4.9 IMPLEMENTATION OF DIFFERENTIAL EVOLUTION TO
DETERMINE FBLC CONSTANTS
The FBLC controller uses pole tracking algorithm to determine the
error constants. The two poles used here are poledc and polev. These are
determined using the differential evolution algorithm. From the poles
determined the error constants for the FBLC constants are calculated. The
117
ranges for both the pole values are specified in the DE algorithm. An initial
value may be assigned. The range for each pole is determined using trial and
error method. Each pole value is determined using differential evolution
following the above algorithm. The poledc value is for the DC part of the
STATCOM while the polev value is for the AC part of the STATCOM. The
tuned FBLC values result in better settling time than the normal pole values.
In a three machine nine bus system, the 3 phase symmetrical fault
occurs in the line connecting the buses 5 and 7, for the duration of 0.25s. The
fault is created at 1s and removed at 1.25s. FACTS device is connected at the
8th bus. The controller used is FBLC. The following combined output
waveforms show the comparison of transient responses with and without
FACTS devices. The simulations are carried out using MATLAB software.
The figure 4.23 shows the comparison of rotor angle curves of
generators 1 and 2. This output waveforms clearly shown that the critical
clearing time (CCT) is 6.0s in the absence of any FACTS devices and CCT is
5.5s in presence of STATCOM. Thus the CCT get reduced by using FACTS
devices and hencethe transient stability of the multi-machine power system
got enhanced.
Then the comparison of speed curves of the generators 1&2, 2&3 and
3&1 are shown in the figures 4.23, 4.24 and 4.25 respectively. The voltage
curves at the 8th bus are compared and are shown in the figure 4.26.From all
the above curves it was clearly understood that by using FACTS devices three
phase symmetrical fault get cleared at faster rate. Hence STATCOM with PI
controller damp out the rotor oscillations quickly and enhance the transient
stability.
118
Figure 4.23 Comparisons of Rotor Angle Curves of Generators 1 & 2
Figure 4.24 Comparisons of Rotor Angle Curves of Generators 1 & 3
119
Figure 4.25 Comparisons of Rotor Angle Curves of Generators 2 & 3
Figure 4.26 Comparisons of Dc Voltage Curves
120
4.10 SUMMARY
Modeling of STATCOM and UPFC is successfully done in this
chapter. A simple mathematical model of a STATCOM and UPFC has been
systematically derived. To study in detail the effectiveness of STATCOM and
UPFC, a Single Machine Infinite Bus system and standard 3 machine 9 bus
system is considered. As illustrated by simulation studies, the damping of the
electromechanical oscillations of the synchronous generator in a simple power
system is better with FBLC based STATCOM and UPFC over conventional
PI controller.hm.
The proposed FBLC for UPFC is proved to be very effective and
robust in damping power system oscillations and thereby enhancing system
transient stability. As illustrated by computer simulation studies, the superior
damping of the electromechanical oscillations of the synchronous generator
provided by this proposed control strategy over the conventional cascade
control approach has been established for a variety of severe transient
disturbances.
Both the STATCOM and the UPFC using the proposed control scheme
provides significant improvement in damping out the electromechanical
oscillations of the generators. The oscillations show significant reductions in
their first and subsequent swings for this controller, in comparison to
conventional PI controllers. The FBLC tuned by BFOA and DE gives better
performance over the other controllers. The electromechanical oscillations
have damped quickly. The settling time for the generators 1, 2 and 3 has
reduced from 5.5, 5 and 5 seconds to 4.5, 5.5 and 4.5 seconds respectively for
FBLC controller and 4.3, 5.2 and 4.2seconds for FBLC tuned by BFOA.
Finally several faults and load disturbance results have been presented to
highlight the effectiveness of proposed FACTS controllers.