EE549 - Power System Dynamics and Control

Transient Stability

S. Sivasubramani

Associate Professor
Electrical Engineering Department
Indian Institute of Technology Patna
www.iitp.ac.in/~siva

Siva (IIT P) EE549 1 / 83

Transient Stability

It is the ability of the system to stay in synchronism when subjected to a

large disturbance.

Large disturbances can be faults, large load variations, generator out-

ages and others.
Since the disturbance is large, the nonlinear differential algebraic equa-
tions that describe the system can not be linearized as the power, angles
and speeds undergo large swings..
One way of assessing the stability is through numerical solution of DAE.
The time of interest for transient stability assessment is typically 2 to
10 seconds after a disturbance.
Since the time constant of a turbine and speed governor is large, it is
assumed that they do not act during the period of study.
Therefore, the turbine and speed governor dynamics are not considered
and the input mechanical torque TM is considered constant for transient
stability analysis.
Siva (IIT P) EE549 2 / 83
Numerical Integration Methods

Let
dx
= f(x, t)
dt
where x is the state vector and f(x, t) is a vector of non linear functions.
1 Explicit Methods

1 Euler Method
2 Modified Euler Method
3 Runge-Kutta Methods
2 Implicit Methods
1 Trapezoidal Rule

Siva (IIT P) EE549 3 / 83

Euler Method
Consider the first-order differential equation.
dx
= f (x, t)
dt
with x = x0 at t = t0 .

Siva (IIT P) EE549 4 / 83

Euler Method
Consider the first-order differential equation.
dx
= f (x, t)
dt
with x = x0 at t = t0 .
x

x1
∆x ∆t
x0
t0 t1 t

Siva (IIT P) EE549 4 / 83

At x = x0 , t = t0 , the curve can be approximated by its tangent having a
slope
dx
= f (x0 , t0 )
dt x=x0
Therefore,
dx
∆x = ∆t
dt x=x0

The value of x at t = t1 = t0 + ∆t is given by

dx
x1 = x0 + ∆x = x0 + ∆t
dt x=x0

This has to be repeated till the time reaches the final simulation time.

Siva (IIT P) EE549 5 / 83

At x = x0 , t = t0 , the curve can be approximated by its tangent having a
slope
dx
= f (x0 , t0 )
dt x=x0
Therefore,
dx
∆x = ∆t
dt x=x0

The value of x at t = t1 = t0 + ∆t is given by

dx
x1 = x0 + ∆x = x0 + ∆t
dt x=x0

This has to be repeated till the time reaches the final simulation time.
Since it considers only the firs derivative of x, it is referred to as a
first order method.
∆t has to be small to achieve accuracy.
Since it uses only the first order information, it may introduce errors.

Siva (IIT P) EE549 5 / 83

Modified Euler Method

The standard Euler method produces inaccurate results because it

uses the derivative at the beginning of the interval.
The modified Euler method tries to overcome this issue by using the
average of the derivatives at the two ends.

Siva (IIT P) EE549 6 / 83

Modified Euler Method

The standard Euler method produces inaccurate results because it

uses the derivative at the beginning of the interval.
The modified Euler method tries to overcome this issue by using the
average of the derivatives at the two ends.
It consists of the following steps.
1 Predictor step
dx
x1p = x0 + ∆t
dt x=x0
2 Corrector step
!
1 dx dx
x1c = x0 + + ∆t
2 dt x=x0 dt x=x1p

This process has to be repeated until the desired accuracy or the final
simulation time.
Siva (IIT P) EE549 6 / 83
Runge-Kutta (R-K) Methods

Euler and the modified Euler method require smaller time steps.
R-K methods approximate the Taylor series solution. However they do
not need derivatives higher than the first.
R-K methods use the effectiveness of higher derivatives by several
evaluations of the first derivative.
They are classified based on the number of evaluations.

Siva (IIT P) EE549 7 / 83

Second order R- K Method

The value of x at t = t0 + ∆t is
k1 + k2
x1 = x0 + ∆x = x0 +
2
where
k1 = f (x0 , t0 )∆t
k2 = f (x0 + k1 , t0 + ∆t)∆t

Siva (IIT P) EE549 8 / 83

Second order R- K Method

The value of x at t = t0 + ∆t is
k1 + k2
x1 = x0 + ∆x = x0 +
2
where
k1 = f (x0 , t0 )∆t
k2 = f (x0 + k1 , t0 + ∆t)∆t
In general,
k1 + k2
xn+1 = xn +
2
where
k1 = f (xn , tn )∆t
k2 = f (xn + k1 , tn + ∆t)∆t

Siva (IIT P) EE549 8 / 83

Fourth order R- K Method

The value x at n + 1st step is

1
xn+1 = xn + (k1 + 2k2 + 2k3 + k4 )
6
where
k1 = f (xn , tn )∆t
k1 ∆t
k2 = f (xn + , tn + )∆t
2 2
k2 ∆t
k3 = f (xn + , tn + )∆t
2 2
k4 = f (xn + k3 , tn + ∆t)∆t

Siva (IIT P) EE549 9 / 83

 k1 = (slope at the beginning of time step)∆t
k2 = (first approximation to slope at mid step)∆t
k3 = (second approximation to slope at mid step)∆t
k4 = (slope at the end of step)∆t
1
∆x = (k1 + 2k2 + 2k3 + k4 )
6
This is equivalent to considering upto fourth derivative terms in the Taylor
series expansion.

Siva (IIT P) EE549 10 / 83

Stability of Explicit Methods

Explicit methods calculate x at any time step from the knowledge of the
values of x at previous time steps.

Siva (IIT P) EE549 11 / 83

Stability of Explicit Methods

Explicit methods calculate x at any time step from the knowledge of the
values of x at previous time steps.

They are not numerically stable.

They require smaller time steps.
For the stiff systems (Stiffness is the ratio of the largest to smallest
time constants), they blow up unless a small time step is used.
Stiffness can also be found by the ratio of the largest to smallest
eigenvalues of the linearized system.

Siva (IIT P) EE549 11 / 83

Implicit Methods

Consider the differential equation

dx
= f (x, t)
dt
with x = x0 and t = t0 .
Z t1
x1 = x0 + f (x, τ )dτ
t0

Implicit methods approximate the integral.

The simplest implicit integration method is the trapezoidal rule.
The trapezoidal rule approximates the integral by trapezoids.

Siva (IIT P) EE549 12 / 83

Trapezoidal Rule

f (x, t)

f (x1 , t1 )
f (x0 , t0 )

t0 ∆t t1 t

Siva (IIT P) EE549 13 / 83

The trapezoidal rule is
∆t
x1 = x0 + [f (x0 , t0 ) + f (x1 , t1 )]
2
In general
∆t
xn+1 = xn + [f (xn , tn ) + f (xn+1 , tn+1 )]
2

Siva (IIT P) EE549 14 / 83

The trapezoidal rule is
∆t
x1 = x0 + [f (x0 , t0 ) + f (x1 , t1 )]
2
In general
∆t
xn+1 = xn + [f (xn , tn ) + f (xn+1 , tn+1 )]
2

xn+1 appears on both sides.

Implicit methods find x using x at the previous time step as well as
current value.
Since the current value is unknown, an implicit equation must be
solved.

Siva (IIT P) EE549 14 / 83

Stability of Implicit Methods

Implicit methods are numerically stable

For the stiff systems, implicit methods suffer from accuracy but not
numerical stability.
Implicit methods work well with larger time steps.
For systems where time steps are limited by numerical stability rather
than accuracy, implicit methods are better.

Siva (IIT P) EE549 15 / 83

Transient Stability Analysis of SMIB

Let us analyze the transient stability considering different models.

1 Classical Model
2 Flux decay model - Without AVR
3 Flux decay model - With AVR
4 Flux decay model - With AVR and PSS

In all the above models, Turbine and Speed governor dynamics are
neglected. Hence TM is constant.

Siva (IIT P) EE549 16 / 83

Classical Model

The differential equations of this model are

dδ
= ω − ωs
dt
2H dω
= Tm − Te − D(ω − ωbase )
ωs dt
With resistances neglected, the air-gap power (Pe ) is equal to the terminal
power (Pt ). In per unit,

Te = Pe = Pmax sin δ

Siva (IIT P) EE549 17 / 83

Example 13.1 from P.Kundur : Let us examine the transient stability.

The initial system operating condition is as follows:

P = 0.9 Q = 0.436 Vt = 1.0∠28.34◦ V∞ = 0.90081∠0◦

The generators are modelled as a single equivalent generator represented

by the classical model.

Xd0 = 0.3 H = 3.5 sec D = 0

Siva (IIT P) EE549 18 / 83
Line 2 experiences a solid three-phase fault at point F and the fault is
cleared by isolating the faulted circuit.
1 Determine the critical clearing fault-clearing time and the critical
clearing angle using numerical integration.
2 Check the above value of critical clearing angle using the equal-area
criterion.
The steady state equivalent circuit is

0.3 It ∠γ 0.15 0.5

Vt = 1.0∠28.34◦
0.93

E 0 ∠δ 0.995∠0◦

Siva (IIT P) EE549 19 / 83

 0.9 − 0.436
Ē 0 = V̄t + I¯t × 0.3 = 1.0∠36◦ + × 0.3 = 1.1626∠41.77◦
1.0∠−28.34◦
1 Before Fault
X1 = 0.7752
1.1626 × 0.90081
Pmax1 = = 1.351
0.7752
2 During Fault
X2 = ∞
Pmax2 = 0
3 Post Fault
X3 = 0.95
1.1626 × 0.90081
Pmax3 = = 1.1024
0.95

Siva (IIT P) EE549 20 / 83

By Equal Area Criterion

Pm (δmax − δ0 ) + Pmax3 cos δmax

cos δc =
Pmax3
δc = 52.24◦
The differential equations are

dω ωs
= (Tm − Te − D(ω − ωs ))
dt 2H
dδ
= (ω − ωs )
dt
The general formula for the second order R-K method
 
k1ω + k2ω
ωn+1 = ωn +
2
 
k1δ + k2δ
δn+1 = δn +
2
Siva (IIT P) EE549 21 / 83
where
ωs
k1ω = (Tm − Pmax sin(δn ) − D(ωn − ωs ))∆t
2H
k1δ = (ωn − ωs )∆t
ωs
k2ω = (Tm − Pmax sin(δn + k1δ ) − D(ωn + k1ω − ωs ))∆t
2H
k2δ = (ωn + k1ω − ωs )∆t

Siva (IIT P) EE549 22 / 83

 Figure: Rotor angle response
Siva (IIT P) EE549 23 / 83
Flux decay model - Without AVR

The differential equations of this model are

0
dEq0
Td0 = −Eq0 − (Xd − Xd0 )Id + Efd
dt
dδ
= ω − ωs
dt
2H dω
= Tm − Te − D(ω − ωbase )
ωs dt
The electrical torque developed is

Te = (Xq − Xd0 )Iq Id + Eq0 Iq

Siva (IIT P) EE549 24 / 83

Example 13.2 from P.Kundur : Let us examine the transient stability with
manual excitation (No AVR).

The initial system operating condition is as follows:

P = 0.9 Q = 0.436 Vt = 1.0∠28.34◦ V∞ = 0.90081∠0◦

Siva (IIT P) EE549 25 / 83

A three phase fault of circuit 2 at point F cleared by isolating the faulted
circuit simultaneously at both ends. The equivalent machine parameters
are
Xd = 1.81 Xq = 1.76 Xd0 = 0.3 Xq0 = 0.65
0 0
Tdo = 8 Tqo = 1 Ra = 0.003 H = 3.5 D = 0
The steady state values are

S∗ 0.9 − 0.436
I¯t = ∗ = = 1∠2.5◦
V 1.0∠−28.34◦
Ē 0 = V̄t + I¯t × (Ra + Xq ) = 2.37∠70.14◦
δ0 = 70.14◦
Vd0 = Vt sin(δ0 − α) = 0.6666 Vq0 = Vt cos(δ0 − α) = 0.7455
Id0 = It sin(δ0 − γ) = 0.9249 Iq0 = It cos(δ0 − γ) = 0.3803
0
Eq0 = Vq0 + Iq0 Rs + Id0 Xd0 = 1.0241 Efd = Eq0
0
+ (Xd − Xd0 )Id = 2.4207

Siva (IIT P) EE549 26 / 83

The second order R-K method with the time step of 0.0001 was used.

Siva (IIT P) EE549 27 / 83

Flux decay model - With AVR

The differential equations of this model are

0
dEq0
Td0 = −Eq0 − (Xd − Xd0 )Id + Efd
dt
dδ
= ω − ωs
dt
2H dω
= Tm − Te − D(ω − ωbase )
ωs dt
dEfd
TA = −Efd + KA (Vref − Vt )
dt
The electrical torque developed is

Te = (Xq − Xd0 )Iq Id + Eq0 Iq

Siva (IIT P) EE549 28 / 83

Example 13.2 from P.Kundur : Let us examine the transient stability with
AVR.

The initial system operating condition is as follows:

P = 0.9 Q = 0.436 Vt = 1.0∠28.34◦ V∞ = 0.90081∠0◦

A three phase fault of circuit 2 at point F cleared by isolating the faulted

circuit simultaneously at both ends. The equivalent machine parameters
are
Xd = 1.81 Xq = 1.76 Xd0 = 0.3 Xq0 = 0.65
Siva (IIT P) EE549 29 / 83
 0 0
Tdo = 8 Tqo = 1 Ra = 0.003 H = 3.5 D = 0
The exciter parameters are

KA = 200 TA = 0.025

The steady state values are

S∗ 0.9 − 0.436
I¯t = ∗ = = 1∠2.5◦
V 1.0∠−28.34◦
Ē 0 = V̄t + I¯t × (Ra + Xq ) = 2.37∠70.14◦
δ0 = 70.14◦
Vd0 = Vt sin(δ0 − α) = 0.6666 Vq0 = Vt cos(δ0 − α) = 0.7455
Id0 = It sin(δ0 − γ) = 0.9249 Iq0 = It cos(δ0 − γ) = 0.3803
0
Eq0 = Vq0 + Iq0 Rs + Id0 Xd0 = 1.0241 Efd = Eq0
0
+ (Xd − Xd0 )Id = 2.4207
Vref = 1.021

Siva (IIT P) EE549 30 / 83

The second order R-K method with the time step of 0.0001 was used.

Siva (IIT P) EE549 31 / 83

Flux decay model - With AVR and PSS
The differential equations of this model are
0
dEq0
Td0 = −Eq0 − (Xd − Xd0 )Id + Efd
dt
dδ
= ω − ωs
dt
2H dω
= Tm − Te − D(ω − ωbase )
ωs dt
dEfd
TA = −Efd + KA (Vref + VPSS − Vt )
dt
1
V̇WF = − VWF + KPSS ω̇
TW
1 1 T1
V̇PSS = − VPSS + VWF + V̇WF
T2 T2 T2
The electrical torque developed is
Te = (Xq − Xd0 )Iq Id + Eq0 Iq
Siva (IIT P) EE549 32 / 83
Example 13.2 from P.Kundur : Let us examine the transient stability with
AVR and PSS.

The initial system operating condition is as follows:

P = 0.9 Q = 0.436 Vt = 1.0∠28.34◦ V∞ = 0.90081∠0◦

A three phase fault of circuit 2 at point F cleared by isolating the faulted

circuit simultaneously at both ends. The equivalent machine parameters
are
Xd = 1.81 Xq = 1.76 Xd0 = 0.3 Xq0 = 0.65
Siva (IIT P) EE549 33 / 83
 0 0
Tdo = 8 Tqo = 1 Ra = 0.003 H = 3.5 D = 0
The exciter parameters are

KA = 200 TA = 0.025

The PSS parameters are

KPSS = 9.5 Tw = 1.41 T1 = 0.154 T2 = 0.033

Siva (IIT P) EE549 34 / 83

The second order R-K method with the time step of 0.0001 was used.

Siva (IIT P) EE549 35 / 83

Numerical Solution of DAE of a multi machine system

Differential equations (51) to (61), stator algebraic equations (63) to (64)

and network power balance equations (71) to (74) together are called as
differential algebraic equations (DAE) a .
a
Refer to slide no 27 in the synchronous machine representation lecture notes

There are two methods by which DAE can be solved. They are
1 Partitioned method (Explicit Method)- first the differential equations
are solved through numerical integration and then the algebraic equa-
tions are solved and this process is repeated till the end of the simula-
tion.
2 Simultaneous method (Implicit Method) - both the differential and
algebraic equations are solved together in each iteration.

Siva (IIT P) EE549 36 / 83

The partitioned method is numerically unstable that is if the time step
of integration is not chosen properly the error will accumulate over the
iterations leading to numerical instability.
But, the partitioned method is computationally simpler than the simul-
taneous method.
Simultaneous method is numerically stable.
Though the method is stable, the computationally harder than the
partitioned method.
The time step of integration depends on the system stiffness, that is
the ratio of the highest to lowest time constant or the highest to lowest
eigenvalue of the linearized system.
The time step of integration in the partitioned method should be less
than the lowest time constant else it will lead to numerical instability.
In case of simultaneous method even if the time step is greater than
the lowest time constant, numerically it is stable but the dynamics
corresponding to that time constant are not visible in the solution.

Siva (IIT P) EE549 37 / 83

DAE

The differential equations (51) to (61) 1 can be expressed as

ẋ = f(x, Idq , V, u) (1)

where

x = [x1 T , · · · , xng T ]T , u = [u1 T , · · · , un T ]T , Idq = [Idq1 T , · · · , Idqng T ]T

xi = [δi , ωi , ψ1di , ψ2qi , Eqi0 , Edi0 , Efdi , VRi , RFi ]T i = 1, · · · ng
ui = [Vrefi , TMi ]T i = 1, · · · ng
Idqi = [Idi , Iqi ]T i = 1, · · · ng
V = [V1 , · · · , Vn ]T = [V1 e θ1 , · · · , Vn e θn ]T
(2)

1
Refer to slide no 27 in the synchronous machine representation lecture notes
Siva (IIT P) EE549 38 / 83
If saliency is neglected(Xd00 = Xq00 ), the generator current can be written as

1 
00 00 (δi − π2 ) θi

Idi + Iqi = (E di + E qi )e − Vi e i = 1, · · · ng (3)
Rsi + Xdi00

It can be written as
Idq = h(x, V) (4)

Siva (IIT P) EE549 39 / 83

Similarly, the network equations can be written as For i = 1, 2, · · · , ng
generator buses,
n
X
Pi (δi , Idi , Iqi , Vi , θi ) − Vi Vj Yij cos(αij + θj − θi ) = 0 i = 1, · · · , ng
j=1
X n
Qi (δi , Idi , Iqi , Vi , θi ) + Vi Vj Yij sin(αij + θj − θi ) = 0 i = 1, · · · , ng
j=1
Xn
PDi (Vi ) + Vi Vj Yij cos(αij + θj − θi ) = 0 i = ng + 1, · · · , n
j=1
X n
QDi (Vi ) − Vi Vj Yij sin(αij + θj − θi ) = 0 i = ng + 1, · · · , n
j=1
(5)

It can be written as
g(x, Idq , V, u) = 0 (6)
Solving the above set of equations is load flow analysis.
Siva (IIT P) EE549 40 / 83
The complete set of DAE can be expressed as

ẋ = f(x, Idq , V, u)
Idq = h(x, V) (7)
g(x, Idq , V, u) = 0

Siva (IIT P) EE549 41 / 83

Partitioned Method (Explicit Method)

The numerical integration can be done by the following methods.

1 Euler Method
2 Modified Euler Method
3 Runga-Kutta Methods

Siva (IIT P) EE549 42 / 83

Euler Method

The Euler method approximates the set of differential equations at

(n + 1)th step as follows:


xn+1 = xn + ∆t f(xn , Idqn , Vn , un ) (8)

Where ∆t is the integration time step and should be less than the least
time constant of the system in order to have a stable numerical method.

After xn+1 is found, the algebraic equations can be solved by solving the
following equations.

Idqn+1 − h(xn+1 , Vn+1 ) = 0 (9)

g(xn+1 , Idqn+1 , Vn+1 , un+1 ) = 0 (10)

Siva (IIT P) EE549 43 / 83

 Equations (9) and (10) can be solved by Newton-Raphson Method.
In case of impedance loads, the power balance equations can be
avoided.
The set of linear algebraic equations given in (9) is to be solved as
follows:

E00n+1
    
IGn+1 Y YGN
= GG T 0 (11)
0 YGN Ybus VNn+1
Where, ng is the number of generators, nt is the total number of buses and
 
V1n+1
   00 
E1n+1  .. 
IG 1n+1  . 
 ..  00  ..   
IGn+1 =  .  , En+1 =  .  , VNn+1 =  V1n g

n+1 
00 . 
IG ng n+1 E1n  .. 

g n+1
V1nt n+1

Siva (IIT P) EE549 44 / 83

Once xn+1 and E00n+1 are known,
0
VN = −(Ybus )−1 YGN
T
E00n+1 (12)
0
IGn+1 = (YGG − YGN (Ybus )−1 YGN
T
)E00n+1 (13)

This process has to be repeated until the desired accuracy or the final
simulation time.

Siva (IIT P) EE549 45 / 83

Modified Euler Method

The modified Euler method uses first a predictor step and a corrector step
to improve numerical stability.

Siva (IIT P) EE549 46 / 83

Modified Euler Method

The modified Euler method uses first a predictor step and a corrector step
to improve numerical stability.

1 The predictor step is

xpn+1 = xn + ∆t f(xn , Idqn , Vn , un )

(14)

2 The algebraic equations are solved with xpn+1 .

3 The corrector step is
∆t
xcn+1 = xn +


f(xn+1 , Idqn+1 , Vn+1 , un+1 ) + f(xn , Idqn , Vn , un ) (15)
2
4 The algebraic equations are solved with xcn+1 .
5 This process has to be repeated until the desired accuracy or the final
simulation time.
Siva (IIT P) EE549 46 / 83
Runge-Kutta Fourth Order Method
At (n + 1)th step,
1
xn+1 = xn + (k1 + 2k2 + 2k3 + k4 ) (16)
6
where
k1 = f(xn , Idqn , Vn , un )∆t
k1
x1n+1 = xn +
2
1
k2 = f(x1n+1 , I1dqn+1 , Vn+1 , u1n+1 )∆t
k2
x2n+1 = xn +
2
2
k3 = f(x2n+1 , I2dqn+1 , Vn+1 , u2n+1 )∆t
x3n+1 = xn + k3
3
k4 = f(x3n+1 , I3dqn+1 , Vn+1 , u3n+1 )∆t

This process has to be repeated until the desired accuracy or the final
Siva (IIT P) EE549 47 / 83
Simultaneous Method (Implicit Method)

The set of differential equations can be numerically integrated using the

trapezoidal rule as
Z tn+1
xn+1 = xn + f(x, Idq , V, u)dt (17)
tn

∆t

xn+1 = xn + f(xn+1 , Idqn+1 , Vn+1 , un+1 ) + f(xn , Idqn , Vn , un ) (18)
2
where ∆t = tn+1 − tn . The algebraic equations at (n + 1)th instant are

Idqn+1 − h(xn+1 , Vn+1 ) = 0 (19)

g(xn+1 , Idqn+1 , Vn+1 , un+1 ) = 0 (20)

Siva (IIT P) EE549 48 / 83

Equations (18), (19) & (20) can be rearranged as follows:
   
∆t ∆t
xn+1 − (f(xn+1 , Idqn+1 , Vn+1 , un+1 )) − xn + (f(xn , Idqn , Vn , un )) = 0
2 2
Idqn+1 − h(xn+1 , Vn+1 ) = 0
g(xn+1 , Idqn+1 , Vn+1 , un+1 ) = 0
(21)

Siva (IIT P) EE549 49 / 83

Equations (18), (19) & (20) can be rearranged as follows:
   
∆t ∆t
xn+1 − (f(xn+1 , Idqn+1 , Vn+1 , un+1 )) − xn + (f(xn , Idqn , Vn , un )) = 0
2 2
Idqn+1 − h(xn+1 , Vn+1 ) = 0
g(xn+1 , Idqn+1 , Vn+1 , un+1 ) = 0
(21)

Equation (21) can be written as

F1 (xn+1 , Idqn+1 , Vn+1 , un+1 , xn , Idqn , Vn , un ) = 0

F2 (xn+1 , Idqn+1 , Vn+1 ) = 0 (22)
F3 (xn+1 , Idqn+1 , Vn+1 ) = 0

The above equations are nonlinear.

They can be solved by Newton-Raphson method for
xn+1 , Idqn+1 , V n+1 , un+1 with initial conditions xn , Idqn , Vn , un .
This process has to be repeated until the desired accuracy or the final
simulation time.
Siva (IIT P) EE549 49 / 83
In case of a three-phase balanced fault at a bus, the row and column
of the system admittance matrix, corresponding to the faulted bus, is
removed as the voltage at that bus is zero.
The power balance equations at that bus are also removed. In case
the fault at some location on the transmission line, the bus admittance
matrix should be changed to include the changed topology with the
rest of the procedure remaining the same.
For an unbalanced fault, this procedure cannot be used. Instead sym-
metrical component theory can be used.

Siva (IIT P) EE549 50 / 83

Analysis of Unbalanced Faults

An unsymmetrical set of voltage or current phasors is resolved into

symmetrical sets of components.
The unbalanced three-phase system is resolved into three balanced
(symmetrical) systems of phasors called the positive, negative and
zero sequence components.
Let the three-phases of the original system be represented by a, b and
c, and the symmetrical components by 1, 2 and 0.

Siva (IIT P) EE549 51 / 83

Symmetrical Components
Any arbitrary set of three phasors can be written as
0 1 2
Va = Va + Va + Va (23)
0 1 2
Vb = Vb + Vb + Vb (24)
Vc1 0 1 2
Vc = Vc + Vc + V cV 2 (25)
b
Va2

Va1

Vb1 Vc2
(a) Positive Sequence (b) Negative Sequence

Va0
Vb0
Vc0
(c) Zero Sequence
Siva (IIT P) EE549 52 / 83
Let us introduce a = 1∠120◦ .
1 1 1 1
V b = a2 V a , V c = aV a (26)
2 2 2 2
V b = aV a , V c = a2 V a (27)
     0
Va 1 1 1 Va
V b  = 1 a2  1
a V a 
 (28)
Vc 1 a a2 Va
2

Vabc = TV012 , V012 = T−1 Vabc (29)

where    
1 1 1 1 1 1
1
T = 1 a2 a  , T−1 = 1 a a2 
3
1 a a2 1 a2 a

Siva (IIT P) EE549 53 / 83

In a similar way,
     0
Ia 1 1 1 Ia
I b  = 1 a 2 a  I 1a  (30)
 
Ic 1 a a2 I
2
a

The complex power is

T ∗
S = Vabc Iabc = (TV012 )T (TI012 )∗
T ∗ T ∗
(31)
= V012 TT T∗ I012 = 3V012 I012

Siva (IIT P) EE549 54 / 83

Let the self impedance of each phase be Zs and the mutual inductance
between any two phases be Zm .
    
Va Zs Zm Zs Ia
V b  = Zm Zs Zm  I b  (32)
Vc Z m Z m Z s Ic
 0    0
Va Zs Zm Zs Ia
 1 −1  1 
V a  = T Z m Z s Z m  T I a  (33)
2 Z Z Z 2
Va m m s Ia
 0    0
Va Zs + 2Zm 0 0 Ia
 1   1
V a  = 0 Zs − Zm 0  I a  (34)
2 0 0 Z − Z 2
Va s m Ia

Siva (IIT P) EE549 55 / 83

In transformer and transmission lines, the positive and negative
sequence impedances are the same.
In case of rotating machines like synchronous machines or induction
machines, the positive and negative sequence impedances are not the
same.
The zero sequence impedance is present only when there is a path for
the zero sequence currents to flow like in a star connected transformer
with neutral grounded.
If an impedance Zn is connected between the neutral and the ground
of a transformer, the zero sequence impedance will become Z0 + 3Zn .

Siva (IIT P) EE549 56 / 83

 In case a fault occurs at a particular location in the system, the rest
of the system can be represented as a pre-fault balanced Thevenin
voltage with the Thevenin equivalent impedance in series.
Let Eabc be the Thevenin voltage as seen from the fault location.
Let Zabc be the Thevenin impedance as seen from the fault location.
The voltages at the fault location are

Vabc = Eabc − Iabc Zabc (35)

Expressing the above equation using the symmetrical components,

 0      0
Va 0 Z0 0 0 Ia
 1     1
V a  = E a − 0 Z 1 0 I a   (36)
2 0 0 0 Z2 2
Va Ia

Siva (IIT P) EE549 57 / 83

Single Line to Ground Fault

Let the fault be on phase a.

V a = 0, I b = I c = 0 (37)

Therefore,
0 1 2
Va = Va + Va + Va = 0 (38)
0 1 2 Ia
Ia = Ia = Ia = (39)
3
On substituting (36) in (38) and using (39),
1 2 0
E a − I aZ 1 − I aZ 2 − I aZ 0 = 0

1 Ea
Ia = (40)
Z1 + Z2 + Z0

Siva (IIT P) EE549 58 / 83

Line to Line Fault
Let the fault between phase b and phase c.

V b = V c , I b = −I c , I a = 0 (41)

Therefore,
1 2
Va = Va (42)
0
Va =0 (43)
1 2
Ia = −I a (44)
0
Ia =0 (45)
1
Substituting (42) to (46) in (36) and solving for I a give

1 Ea
Ia = (46)
Z1 + Z2

Siva (IIT P) EE549 59 / 83

Double Line to Ground Fault
Let the fault between phase b and phase c and ground.
V b = V c = 0, I a = 0 (47)
Therefore,
0 1 2 1
Va = Va = Va = Va (48)
3
0 1 2
Ia + Ia + Ia = 0 (49)
Substituting (36) in (48) gives
1 2 0
E a − I a Z 1 = −I a Z 2 = −I a Z 0 (50)
1
Substituting (50) in (49) and solving for I a give

1 Ea
Ia = (51)
Z 2Z 0
Z1 + Z 2 +Z 0

Siva (IIT P) EE549 60 / 83

During an unbalanced fault, the power system can be represented as a
positive sequence network with appropriate effective impedance connected
at the fault location and the analysis can be done like the balanced fault.

The effective impedance for different fault-conditions is

L- G Z eff = Z 1 + Z 2 + Z 0
L- L Z eff = Z 1 + Z 2
Z 2Z 0
L- L-G Z eff =
Z2 + Z0

Siva (IIT P) EE549 61 / 83

During an unbalanced fault, the power system can be represented as a
positive sequence network with appropriate effective impedance connected
at the fault location and the analysis can be done like the balanced fault.

The effective impedance for different fault-conditions is

L- G Z eff = Z 1 + Z 2 + Z 0
L- L Z eff = Z 1 + Z 2
Z 2Z 0
L- L-G Z eff =
Z2 + Z0
Transient stability of a system is assessed by solving the DAE through nu-
merical methods mentioned above for the period of interest typically 2 to 10
seconds.
If the system does not settle down to post fault stable operating point within
the time of interest, the system is said to be unstable.
If the system settles down to a stable operating point, the system is said to
be stable.
It is necessary that both small signal and transient stability analysis should be
carried out for any system to assess the health of the system.
Siva (IIT P) EE549 61 / 83
Direct Method

Instead of solving DAE , the transient stability can be assessed directly

through Lyapunov direct method of stability.

It says
Ẋ = f(X) (52)

if there exist a positive definite continuous function V (X) whose first

partial derivative with respect to the state variables exists and if the
total derivative V (Ẋ) is negative semi definite then the system is said
to be stable.
The function V (X) is called as Lyapunov energy function.

Siva (IIT P) EE549 62 / 83

 Figure: Rolling Ball Analogy

Source: P. Kundur

Siva (IIT P) EE549 63 / 83

Initially the ball is resting at the bottom of the bowl and the state is
referred to as the stable equilibrium point (SEP).
When some kinetic energy is injected into the ball, the ball will roll up
inside the surface of the bowl along a path determined by the direction
of initial motion.
The point where the ball will stop is governed by the amount of kinetic
energy initially injected.
If the ball converts all its kinetic energy to potential energy before
reaching the rim, it will roll back and eventually settle down at the
SEP.
However, if the kinetic energy injected is high enough to cause the ball
to go over the rim, then the ball will enter the region of instability and
will not return to the SEP.
The surface inside the bowl represents the potential energy surface, and
the rim of the bowl represents the potential energy boundary surface
(PEBS).

Siva (IIT P) EE549 64 / 83

Two quantities are required to determine if the ball will enter the
instability region:
1 The initial kinetic energy injected.
2 The height of the rim at the crossing point
These can be applied to power systems, for a given fault, whether the
system kinetic energy during the fault can be dissipated or converted
into potential energy before reaching a critical point beyond which the
system will become unstable. This is called as transient energy function
We need some function which can adequately define the transient en-
ergy of the system and also the critical energy required to destabilize
the system.

Siva (IIT P) EE549 65 / 83

The swing equation of the generator connected to an infinite bus is given
as
2H d 2 δ
= Pm − Pmax sin δ (53)
ωbase dt 2
Let a potential energy function be defined as a function of the rotor angle
δ as
VPE (δ) = −Pm δ − Pmax cos δ (54)
Hence,
2H d 2 δ dVPE (δ)
= Pm − Pmax sin δ = − (55)
ωbase dt 2 dδ
Multiplying the above equation by δ̇ on both sides, rearranging and
assuming M = ω2H base
give
 2 !
d 1 dδ dVPE (δ)
M + =0 (56)
dt 2 dt dt

Siva (IIT P) EE549 66 / 83

Let ω 0 = ω − ωbase .
dδ
= ω0 (57)
  dt
d 1 0 2 dVPE (δ)
M(ω ) + =0 (58)
dt 2 dt

d
(V (δ, ω 0 )) = 0 (59)
dt

The function VPE (δ, ω 0 ) is the sum of change in kinetic energy and
the potential energy gained.
The potential energy function VPE (δ) can also be expressed as a
change in the potential energy from the steady state.

Siva (IIT P) EE549 67 / 83

If δ0 is the steady state angle, then

VPE (δ, δ0 ) = −Pm (δ − δ0 ) − Pmax (cos δ − cos δ0 ) (60)

The total energy of the system as an energy function is defined as

1
V (δ, ω 0 ) = M(ω 0 )2 − Pm (δ − δ0 ) − Pmax (cos δ − cos δ0 ) (61)
2

Infinite Bus
F

Siva (IIT P) EE549 68 / 83

 P

Pmax1

Pmax3
A2
Pm
A1

δ0 δcr δu δ
PE(δu )
Energy
Vcr (δu , ω 0 )
KE(δcr )

PE(δcr )

δ0 δcr δu δ

Siva (IIT P) EE549 69 / 83

For the system to be stable,

A1 = A2
PE(δcr ) + KE(δcr ) = PE(δu )

The transient energy at the angle δu is given as

1
Vcr (δu , ω 0 ) = M(ω 0 )2 − Pm (δu − δ0 ) − Pmax (cos δu − cos δ0 ) (62)
2

Siva (IIT P) EE549 70 / 83

For the system to be stable,

A1 = A2
PE(δcr ) + KE(δcr ) = PE(δu )

The transient energy at the angle δu is given as

1
Vcr (δu , ω 0 ) = M(ω 0 )2 − Pm (δu − δ0 ) − Pmax (cos δu − cos δ0 ) (62)
2

Vcr (δu , ω 0 ) is the maximum energy the system can have without
becoming unstable.
If the energy exceeds this critical energy then the system is unstable.
Hence, the system stability can be assessed by computing the
transient energy at the critical clearing angle and checking if it less
than Vcr (δu , ω 0 ).

Siva (IIT P) EE549 70 / 83

The transient energy at the critical clearing angle is

1
Vcl (δcr , ω 0 ) = M(ω 0 )2 − Pm (δcr − δ0 ) − Pmax (cos δcr − cos δ0 ) (63)
2

If Vcl (δcr , ω 0 ) ≤ Vcr (δu , ω 0 ), the system is stable.

If Vcl (δcr , ω 0 ) > Vcr (δu , ω 0 ), the system is unstable.

Siva (IIT P) EE549 71 / 83

Example 13.8 from P. Kundur : Transient Stability Analysis of the
Example 13.1 using the direct method.

The initial system operating condition is as follows:

P = 0.9 Q = 0.436 Vt = 1.0∠28.34◦ V∞ = 0.90081∠0◦

The generators are modelled as a single equivalent generator represented

by the classical model.

Xd0 = 0.3 H = 3.5 sec D = 0

Siva (IIT P) EE549 72 / 83
Line 2 experiences a solid three-phase fault at point F and the fault is
cleared by isolating the faulted circuit.
The post fault system parameters are as follows:

Pmax = 1.1024
 
0.9
δSEP = sin−1 = 0.9552 rad
1.1024
δUEP = π − δSEP = 2.1864 rad
The transient energy function is

1
V (δ, ω 0 ) = M(ω 0 )2 − Pm (δ − δ0 ) − Pmax (cos δ − cos δ0 )
2

Siva (IIT P) EE549 73 / 83

The critical energy, which is the system potential energy at the post fault
UEP, is

Vcr = −Pm (δUEP − δSEP ) − Pmax (cos δUEP − cos δSEP )

= −0.9(2.1864 − 0.9552) − 1.1024(cos(2.1864) − cos(0.9552))
= 0.1651

We know by Equal Area Criterion

δc = 53◦ = 0.9250 rad

The transient energy at the critical clearing angle is

1
Vcl = M(ω 0 )2 − Pm (δc − δSEP ) − Pmax (cos δc − cos δSEP )
2
1 2×H
= × (ωc − ωbase )2 − 0.9(0.9250 − 0.9552)
2 ωbase
− 1.1024(cos(0.9250) − cos(0.9552))

Siva (IIT P) EE549 74 / 83

To find ωc ,
2H dω
= Pm − Pmax sin δ
ωbase dt
At δ = δc , Pmax sin δ = 0.
ωbase 377
ωc = Pm t + ωbase = × 0.9 × 0.08 + 377 = 380.877
2H 7
Hence,
1 2 × 3.5
Vcl = × (380.877 − 377)2 − 0.9(0.9250 − 0.9552)
2 377
− 1.1024(cos(0.9250) − cos(0.9552))
= 0.1399

Since Vcl < Vcr , it is stable.

Let us find Vcl at δ = 54◦ and t = 0.09 sec.

Vcl = 0.1767

Since Vcl > Vcr , it is unstable.

Siva (IIT P) EE549 75 / 83
Practical Power Systems

The application of the direct method to practical power system is

difficult.
The analysis has been limited to power system representation with
the generators represented by the classical model and loads modelled
as constant impedances.
Let the swing equation of an i th generator be

d 2 δi
Mi = Pmi − Pmaxi sin δi i = 1, 2, · · · ng (64)
dt 2

Siva (IIT P) EE549 76 / 83

With the generator transient reactances and load admittances included in
the node admittance matrix, we may write

IG = YR EG (65)

where YR is the reduced admittance matrix with all nodes other than the
generator internal nodes eliminated, EG is the generator internal source
voltage vector, and IG is the generator current vector.

The real power output of an i th generator is given as

∗
Pei = Real(E Gi I Gi ) = Real(E Gi (YR EG )∗ )
ng !
X ∗
= Real E Gi (Gik − Bik )E Gk
k=1 (66)
ng
X
2
= EGi Gii + EGi EGk (Gik cos(δi − δk ) + Bik sin(δi − δk ))
k=1
k6=i

Siva (IIT P) EE549 77 / 83

 0 =P 2
Let Pmi mi − EGi Gii

ng
d 2 δi 0
X
Mi 2
= Pmi − EGi EGk (Gik cos(δi − δk ) + Bik sin(δi − δk )) (67)
dt
k=1
k6=i

Defining all the rotor angles and speed in terms of Centre Of Inertia (COI)
as
ng
1 X
δ0 = Mi δi
MT
i=1
ng (68)
1 X
ω0 = Mi ωi
MT
i=1
Png
where MT = i=1 Mi .

θi = δi − δ0
(69)
ω̃i = ωi − ω0

Siva (IIT P) EE549 78 / 83

Equation (67) can be represented in terms of COI variables as
ng
d 2 θi d 2 δ0 0
X
Mi 2 + Mi 2 = Pmi − (Cik sin(θi − θk ) + Dik cos(θi − θk )) (70)
dt dt
k=1
k6=i

where Cik = EGi EGk Bik and Dik = EGi EGk Gik .
ng
d 2 δ0 Mi X d 2 δi
Mi = Mi 2
dt 2 MT dt
i=1
 
ng ng
Mi X  0 X
= Pmi − EGi EGk (Gik cos(δi − δk ) + Bik sin(δi − δk ))

MT
i=1 k=1
k6=i
 
ng ng ng
Mi X 0 XX
= Pmi − EGi EGk (Gik cos(δi − δk ) + Bik sin(δi − δk ))

MT

i=1 i=1 k=1
k6=i
Mi
= PCOI
MT
(71)

Siva (IIT P) EE549 79 / 83

Equation (70) can be written as
ng
d 2 θi 0
X Mi
Mi 2 = Pmi − (Cik sin(θi − θk ) + Dik cos(θi − θk )) − PCOI (72)
dt MT
k=1
k6=i

Since sin(θi − θk ) = − sin(θk − θi ) and cos(θi − θk ) = cos(θk − θi ), PCOI

is simplified as
ng ng ng
X X X
0
PCOI = Pmi −2 EGi EGk Gik cos(δi − δk ) (73)
i=1 i=1 k=1
k6=i

The energy function for each generator is defined as

Z θi
1
Vi (θi , ω̃i ) = Mi ω̃i2 − fi (θi )dθi i = 1, 2, · · · ng (74)
2 θis

where θis is the i th generator rotor angle in COI.

Siva (IIT P) EE549 80 / 83
The function fi (θi ) is
ng
0
X Mi
fi (θi ) = Pmi − (Cik sin(θi − θk ) + Dik cos(θi − θk )) − PCOI (75)
MT
k=1
k6=i

The energy function for all the generators is

ng ng
X 1 X
0
V (θ, ω̃) = Mi ω̃i2 − Pmi (θi − θis )
2
i=1 i=1
ng ng
XX
− (Cik cos(θi − θj ) − Cik cos(θis − θjs ))
i=1 k=1
k6=i
ng ng Z θ ng Z θi
XX i X Mi
− Dik cos(θi − θk )d(θi + θk ) − PCOI dθi
s MT θis
i=1 k=1 θi i=1
k6=i
(76)

Siva (IIT P) EE549 81 / 83

The first term is the change in rotor kinetic energy of all the generators
in COI reference frame.
The rest of the terms are the potential energy of the system.
Compute the energy function given in (76) with a sustained fault till it
reaches its maximum value which is called the critical energy.
Compute the energy function given in (76) for a clearing angle and if
the energy is less than the critical energy, then the system is stable else
the system is unstable.

Siva (IIT P) EE549 82 / 83

Methods of Improving Transient Stability

1 High Speed fault clearing

2 Reduction of transmission system reactance
3 Regulated Shunt Compensation
4 Dynamic braking
5 Independent pole operation of circuit breakers
6 Single pole switching
7 Fast valve operation
8 Tripping of generator
9 Using supplementary damping controllers: Like power system
stabilizer or supplementary feedback controllers to FACTS.
10 Using HVDC links

Siva (IIT P) EE549 83 / 83