Anda di halaman 1dari 95

EE 6501 POWER SYSTEM

ANALYSIS

NUMERICAL INTEGERATION METHODS


FOR STABILITY ANALYSIS

1
OBJECTIVE
To introduce numerical integration
methods - Euler method, Modified Euler
method, and Runge-Kutta methods for
solving Transient stability problem in
Power System.

2
OBJECTIVE (Cont)

To apply Modified Euler method to a


Single-Machine Infinite Bus System
and to obtain Swing curve.

3
FIG 15.1 AN SMIB SYSTEM
EB = 1.0 p.u.
Vt = 1.0 p.u.
Xd
xL1
xT
Infinite Bus
G

xL2
Pe = 1.0 p.u.

4
FIG 15.1 AN SMIB SYSTEM (Cont)
Xd = 0.2 p.u, xT = 0.1 p.u, xL1=xL2=0.4 p.u.
A 3-phase to ground fault occurs at the the left
end of line L2. (Fig 15.1). The fault is cleared at a
time of 0.05 sec. Is the system stable?
Determine also the critical clearing angle and
critical clearing time.

5
PRE FAULT CONDITION
Determine initial conditions for the
stability analysis:
xL1
E| Vt|

xL2 EB |0o

Xd xT

Pe Pe
Pe

FIG 15.2 EQUIVALENT CIRCUIT -1


6
PRE FAULT CONDITION (Cont)

E| Vt|

xe
EB |0o

jxd

Pe , It
Pe , It
xL1 xL2
x e =x T +
(xL1 + xL2 )

FIG 15.3 EQUIVALENT CIRCUIT -2


7
PRE FAULT CONDITION (Cont)

Pe

xe

EB |0o
Vt|

Vt EB
Pe = sin ;
Xe
= sin-1 {Pe X e / VtEB } 8
PRE FAULT CONDITION (Cont)
Compute the following :

It = (V t - EB ) / (jX e )

E = V t + jx d It
' '

Thus the rotor angle


at t = 0 is computed. 9
SWING EQUATION
The SwingEquation for the SMIB systemis :

H d2
=Pm - Pe p.u.MW (15.1)
f dt 2

'
E EB
Pe = sin = Pmax sin (15.2)
X

X = x 'd + X e (15.3)
10
SWING EQUATION (Cont)

During the fault period, it is assumed


that the mechanical power Pm, (Turbine
Power) remains constant.
The electrical power, Pe changes during
fault period

11
SWING EQUATION (Cont)

Pre fault period : Pe = Pmax(pre) sin


Fault period : Pe = Pmax(f) sin
Post-fault period : Pe = Pmax(post) sin
In Pmax ,E and EB remain the same. Only the
transfer reactance X changes.

12
CHANGE IN TRANSFER REACTANCE

Xpre = xd + Xe (refer eqn. 15.3)


Xpost = xd +xT + xL1
Xf , the reactance during the fault period is
to be computed using the nature of the
fault.

13
CHANGE IN TRANSFER REACTANCE
(Cont)
C
A B xL1
E|
(xd +xT)

EB |0o

xL2

End of line fault in L2

14
CHANGE IN TRANSFER REACTANCE
(Cont)
xAC

A
C
xL1
(xd +xT)
B

xL2
xCo
xAo o

15
CHANGE IN TRANSFER REACTANCE
(Cont)

UsingStar - Delta transformation


' '
DefineNr =(x + x T ).xL1 + x L1 (0)+ (0)(x + x T )
d d

Nr Nr Nr
x AC = ;x A0 = ;x C0 = '
0 xL1 (x d + x T )
X f = x AC =
Pmax(f) = 0
16
EQUAL AREA CRITERION
P

Pmax(pre)

Pre fault curve


Pmax(post)
F
E
A2 Post fault curve
A H
Pm D G

A1

0 B C cl m *
o

17
EQUAL AREA CRITERION (Cont)

Pre fault operating point A; Pm = Pe ; = o .


Fault occurs at t=0; Pa = Pm - Pe = Pm 0= Pm.

2
H d
= Pm ;
f dt 2

increases from 0 ; B C
At = cl , fault is cleared;
Point C jumps to Point E 18
EQUAL AREA CRITERION (Cont)
Now Pe (given by point E) is greater than Pm.
Deceleration acts on rotor. However rotor
angle increases further since speed of rotor is
still greater than synchronous speed.
Rotor swings up to m . This corresponds to
area A2 becoming equal to A1.

19
EQUAL AREA CRITERION (Cont)
Area A1 is proportional to the excess K.E.
pumped to the rotor making it run at a speed
> synch. Speed.
Area A2 is proportional to the K.E. drawn
back from rotor.
When A2=A1, all the excess K.E. pumped into
the rotor is taken back. = s . No more
increase in . Now decrease because of
deceleration. 20
EQUAL AREA CRITERION (Cont)
S.S. solution, under post fault situation is given
by point H on the post fault curve.
Rotor angle swings around this point until
damping brings it to equilibrium point.
It should be noted that the negative area A2 can
be generated only if is less than * (Fig 15.4).

21
EQUAL AREA CRITERION (Cont)

Pm
* = - sin {
-1
}
Pmax(post)
If goes beyond * ,Pe < Pm then rotor
starts accelerating again contributing
to + ve area

22
EQUAL AREA CRITERION (Cont)
System will be stable if the fault is cleared
giving an area A1 for which we can generate
an area A2 before the rotor reaches *.
(the point of no return).

Hence critical clearing angle can be derived


as follows.

23
EQUAL AREA CRITERION (Cont)
P

Pmax(pre)

Pmax(post)
A2
Pm

A1

o s cc *

Fig 15.4 Equal Area Criterion 24


CRITICAL CLEARING ANGLE

A1 = A 2
*

Pm (cc - 0 ) = P
cc
max(post) sin d -Pm ( * - cc )

*
Pm ( * - 0 ) = - Pmax(post) cos cc

= Pmax(post) [coscc - cos * ]

25
CRITICAL CLEARING ANGLE (cont)

coscc =[Pm / Pmax(post) ]( * - 0 )+cos *

cc = cos {cos * +[Pm / Pmax(post) ] ( * - 0 )}


-1

* = - sin {Pm / Pmax(post) }


-1

26
CRITICAL CLEARING TIME
Under fault:
d
2
f
2
= Pm
dt H
d f
= Pm t
dt H
cc t cc
f

0
d =
H
Pm
0
t dt
27
CRITICAL CLEARING TIME (Cont)

f
( cc - o ) = Pm (t cc / 2)
2

H
t cc = Critical clearing time
1/2
2H
t cc = (cc - o )
f Pm
28
NUMERICAL EXAMPLE
E| xL1
Vt|

xL2
EB |0o

xd xT

Pe ,It Pe ,It

Vt =1.0, EB =1.0, Pe =1.0


xd = 0.2pu, xT = 0.1pu,
xL1 = xL2 =0.4p.u 29
NUMERICAL EXAMPLE (Cont)

EB =1.0 0 ;V t =1.0 ;xE =0.1+ 0.2 = 0.3


o o

Vt EB 1.0 * 1.0
Pe = sin = sin = 1.0
xE 0.3
=17.458 ;V t = 1.0 17.458
o o

=0.954 + j0.3p.u

30
NUMERICAL EXAMPLE (Cont)

(0.954 + j0.3)- 1.0


It = =1.0 + j0.01535
j0.3
o
=1.02 8.729
E' = (0.954 + j0.3) + j0.2(1.0 + j0.1535)
o
=0.923 + j0.5 =1.05 28.44

31
CRITICAL CLEARING ANGLE

Pre fault:
Xpre =0.2 + 0.1+ 0.2 = 0.5pu
E'EB 1.05 * 1.0
Pmax(pre) = = = 2.10pu
X 0.5
pre
Pm = Pe =1.0;0 = 28.44 o

32
CRITICAL CLEARING ANGLE (Cont)
During fault :
X (f) = ;Pmax(f) =0
Post fault :
X (post) =0.2 + 0.1+ 0.4 = 0.7pu
1.05 * 1.0
Pmax(post) = =1.5pu
0.7
* = - sin {Pm / Pmax(post) } = - sin {1.0/1.5}
-1 -1

33
CRITICAL CLEARING ANGLE (Cont)
* = - 0.7297 = 2.4119rad

cc = cos {cos* +[Pm / Pmax(post) ](* - 0 )}


-1

-1
= cos {cos(2.4119) +[1.0/1.5](2.4119 - 0.4957)}

o
=1.0097 rad or 57.9
34
CRITICAL CLEARING TIME
2H
t cc = { (cc - 0 )} 1/2

f Pm
2*5
={ (1.0097 - 0.4957)}
* 50 * 1.0
=0.181sec
t cc = 0.05 sec .
Hence the system is stable
35
MID LINE FAULT

We have derived expression for critical clearing


angle and critical clearing time for end of line
fault in Line L2. For this case we found that
during fault period, transfer reactance, x(f) was
found to be infinity and Pmax(f) = 0.

36
MID LINE FAULT (Cont)

Let us now consider a mid-line fault in line


L2. In this case, Pmax(f) will be non zero and
it will be lesser than both Pmax(pre) and
Pmax(post).
Let us derive expression for critical
clearing angle for a mid-line fault.
37
MID LINE FAULT (Cont)
P
Pmax(pre)

Pmax(post)

A2
Pm
A1
Pmax(f)

o cc *

38
MID LINE FAULT (Cont)

In figure A1 = A 2
cc

A1 =
0
[Pm -Pe(f) ] d

cc
= Pm (cc - 0 ) + Pmax(f) cos 0

=Pm (cc - 0 ) + Pmax(f) [cos cc - cos 0 ]

39
MID LINE FAULT (Cont)
*

A2 = [P
cc
max(post) sin -Pm ]d

*
=- Pmax(post) cos cc -Pm ( * - cc )

A 2 =Pm (cc - * )-Pmax(post) [cos * - cos cc ]

40
MID LINE FAULT (Cont)
A1 = A 2

Pm [cc - 0 - cc + * ]-Pmax(f) cos 0 +Pmax(post) cos *

= Pmax(post) coscc - Pmax(f) coscc

Pm [ * - 0 ] - Pmax(f) cos0 + Pmax(post) cos *

= coscc [ Pmax(post) - Pmax(f) ]


41
MID LINE FAULT (Cont)

Pm [ * - 0 ] +Pmax(post) cos * - Pmax(f)cos0


cc =cos {
-1
}
[Pmax(post) - Pmax(f) ]

42
NUMERICAL EXAMPLE
Same example, but mid-line fault in line L2
0.4
E|

EB |0o

0.2 0.1s
0.2 0.2

A B

43
NUMERICAL EXAMPLE (Cont)
xAC

A
C
0.4
0.3
B

0.2

0.2
xAo xC0

o
44
NUMERICAL EXAMPLE (Cont)

(0.3 * 0.4) + (0.4 * 0.2) + (0.2 * 0.3)


x AC =
0.2
0.12 + 0.08 + 0.06 0.26
= = = 1.3p.u
0.2 0.2
other limbs are not of interest

45
NUMERICAL EXAMPLE (Cont)

Transfer reactance during fault


X (f) = X AC = 1.3p.u
E' * EB 1.05 * 1.0
Pmax(f) = = = 0.8077
X (f) 1.3
Other data may be borrowed from
previous example.

46
NUMERICAL EXAMPLE (Cont)

Pm = 1.0 p.u; 0 =28.44o or 0.4957 rad


* = 2.4119 rad
Pmax(post) = 1.5 p.u.
Let us compute cc using the above data.

47
NUMERICAL EXAMPLE (Cont)
cc =
-1 1.0[2.4119 - 0.4957]+1.5cos(2.4119)- 0.8077cos(0.4957)
cos { }
[1.5- 0.8077]

= cos-1(0.12698)

o
= 82.7

Compare with the cc of the severe fault = 57.9o


48
STATE VARIABLE MODEL

The first step in applying a numerical


method to solve Swing Equation is to
convert it into state variable form which
comprises only 1st order differential
equation.

49
STATE VARIABLE MODEL (Cont)
The state variable model enables
systematic application of any numerical
integration technique for solving Swing
Equation thereby helping development of
computer software which can be applied to
problems involving Power System of any
size.
50
SWING EQUATION

S wing E quation for a S MIB system is


H d2

=Pa =Pm -Pe


f dt2

=Pm -Pmax sin, p.uMW ( 16.1)


:in electricalradians ; H :in sec
f: inHz ; t:in sec

51
SWING EQUATION (Cont)

d
=-s ( 16.2)
dt
d
dt gives the relative rotor speed
in elect.radian per sec.

It is the rotor angular velocity withreference to


synchronously rotating reference frame.

52
SWING EQUATION (Cont)
Differentiating( 16.2)w.r.ttime t
d d
2

= ( 16.3)
dt 2
dt
S ubstituting( 16.3)in( 16.1)anddefining
x1 = andx 2 =,
.x = =
. f P
-Pmax sin ( 16.4)
.x = .=-
H
1 m

1 s
( 16.5)
53
SWING EQUATION (Cont)

f
x1 = Pm -Pmax sinx 2 =f(1 x )

H ( 16.6)

x1 = x1 -2 f = f(2 x )

x = (f x ) ( 16.7)

x = ( x1 x 2 ) =( )
t t
( 16.8)

54
SWING EQUATION (Cont)

Problem Statement:
Given: Initial Conditions, (0) and (0)
and the fault sequence.
To det: the solution of (16.6) or (16.7)
(t) and (t) ; t 0

55
NUMERICAL INTEGERATION
METHODS
The most practical method of transient
stability analysis is time-domain
simulation.
In this method the non-linear differential
equations are solved using step-by-step
numerical integration technique.

56
EULER METHOD
It is a simple method and can serve as
a good introduction to numerical
integration.
Later advanced methods used in
commercial software will be taken up.

57
EULER METHOD (Cont)

Consider a first order differential equation

dx
x= = f (x, t) (16.9)
dt
with x = x0 at t = t 0
Fig(16.1) illustrates the principle underlying
the Euler method

58
EULER METHOD (Cont)
x True solution

x1
Tangent
x
xo

to t1 t

FIG. 15.1 BASIC PRINCIPLE 59


EULER METHOD (Cont)
* At t = to, x = xo we can approximate the curve
representing the true solution by the tangent
with the following slope.

dx
| = f(x 0 ,t 0 )
dt (xo ,to )
60
EULER METHOD (Cont)

Hence the change in x is


dx
x = |( x0 ,t0 ) . t
dt
The values of x at t = t1 = t0 +t is
dx
x1 = x 0 +x = x 0 + |( x0 ,t0 ) . t ( 16.10)
dt

61
EULER METHOD (Cont)

E uler method is equivalent to using


the first two terms of the Taylor series
expansion for x around the point( x o,to )

t2 t3
x1 = x o + t( x o ) + ( x o ) + ( x o )+..... ( 16.11)



2! 3!

62
ALGORITHM FOR EULER METHOD

dx
x n = x n-1 + |( xn-1,tn-1 ) . t ( 16.12)
dt
n =1, 2,.......Nmax

where Nmax =( Tmax /t )


Tmax = Time upto which simulation is required

63
LIMITATION OF EULER METHOD

Euler method is a first order method since it


considers only the first derivative of x.
To get adequate accuracy at each step, t has to
be chosen small.
This increases the computational effort very
much.

64
NUMERICAL STABILITY
Numerical stability of a numerical technique
depends on the propagation of error.
If early errors cause other large error later, the
method is said to be numerically unstable.
Euler method has poor numerical stability.

65
MODIFIED EULER METHOD
The inaccuracy of the Euler method is due to the
reason that the derivative at the beginning of the
interval is used for the entire interval.
The modified Euler method overcomes this
problem by using the average of the derivatives at
the two ends.

66
MODIFIED EULER METHOD
Predictor Step:
Predict the value at the end of the step using
the derivative at the beginning of the step.

dx
x1 = x0 +
p
|( x0 ,t0 ) . t ( 16.13)
dt
67
MODIFIED EULER METHOD (Cont)
Corrector step:
The corrected value is found by using the
average of the derivative at the beginning of
the step and the derivative at the end of the
step computed using the predicted value x1p

1 dx dx
x 1 = x 0 + |( x o,t ) + |( xp,t ) .t
c
( 16.14)
2 dt o
dt 1 1
68
NUMERICAL EXAMPLE
A 25 MVA, 50 HZ water wheel generator delivers 20
MW over a double circuit transmission line to a
metro system (Infinite Bus)

H = 2.76 MJ/MVA ; xd = 0.30 p.u.

Xline = 0.20 p.u. (Base 25 MVA) per circuit.

69
NUMERICAL EXAMPLE (Cont)

E=1.03 p.u ; EB = 1.0 p.u

Three-phase solid fault occurs at the middle of one


of the lines at t = 0 sec and is cleared at tcl = 0.4 sec
by isolating the faulted line.

Obtain the swing curve and check stability

70
NUMERICAL EXAMPLE (Cont)
Choose the modified Euler method and a time
step of 0.05 sec.
C 0.2 B
EB = 1.0 p.u.
E=1.03p.u. 0.3
- + + -
A
0.1 0.1
F
H=
H = 2.76

71
NUMERICAL EXAMPLE (Cont)

f
x1 = Pm -Pmax sinx 2 = f(1 x )

H ( 16.6)
x 2 = x1 -2f = f(2 x )

f *50
= = 56.913;
H 2.76
20
2f =100 = 314.16; Pm = = 0.8 p.u.
25
72
NUMERICAL EXAMPLE (Cont)

x1 = 56.913 [0.8 - Pmax sinx 2 ]

( 16.15)
x 2 = x1 -314.16

Pmax varies withtime


Prefaultconditions : ( t 0 -)
x pre = 0.3 +( 0.2/2) = 0.4p.u.
Pmax( pre)
=( 1.03 *1.0/0.4) = 2. 575p.u
73
NUMERICAL EXAMPLE (Cont)
During fault: (0+ t 0.4- sec)
XAB

0.3 0.2
F
A B

0.1
0.1

XAo
o
XBo
74
NUMERICAL EXAMPLE (Cont)

(0.3* 0.2) + (0.2* 0.1) + (0.1* 0.3)


x AB =
0.1
0.06 + 0.02 + 0.03
= = 1.1 p.u
0.1
x(f) = x AB = 1.1 p.u
Pmax(f) = (1.03*1.0)/(1.1) = 0.9364 p.u.
75
NUMERICAL EXAMPLE (Cont)

Post-fault condition: (0.4+ t Tmax)

Xpost = 0.3 + 0.2 = 0.5 p.u

Pmax(post) = (1.03*1.0)/(0.5) = 2.06 p.u

Swing Equations to be solved are different


during fault and after fault.

76
SWING EQUATIONS

During fault: (0+ t 0.4- sec)




x 1 = 56.913 [0.8 - 0.9364 sinx 2 ]
(16.16)
x 2 = x 1 - 314.16

Initial conditions : x 1 (0 + ) and x 2 (0 + )

77
INITIAL CONDITIONS
Rotor velocity or x1 and relative rotor angle
or x2 cannot jump suddenly on the occurrence of
fault at t = 0+ because of the inertia rotor
possesses.

x1 (0+ ) = (0+ ) = (0- ) and


x 2 (0+ ) = (0+ ) = (0- )
78
INITIAL CONDITIONS (Cont)

( 0- ) and ( 0- ) can be obtained by observing the

equilibrium condition that existed at t = 0-

d
At t = 0-, both the rotor acceleration or x1 and


dt
d
the relative rotor velocity, or x 2 are zero.
dt

79
INITIAL CONDITIONS (Cont)
S etting x =0 in eqn.( 16.15) , we get

0 = Pe( 0- ) - P sin
max( pre) ( 0-)

0 = 0.8 - 2.575 sin


( 0- )

( 0+ )=sin-1( 0.8/2.575)=0.3159rad.

x ( 0 + ) = ( 0 + ) = ( 0 - ) = 0.3159 rad
2
or 18.1o ( 16.17) 80
INITIAL CONDITIONS (Cont)

S etting x 2 = 0 in eqn.( 16.15) , we get


0 = x(1 0- )-314.16

x(1 0 + ) = x(1 0 - ) = 314.16 rad/sec ( 16.18)

E qn.( 16.16) can be solved using initial conditions


in eqn.( 16.17) and( 16.18)
81
MODIFIED EULER STEP
During fault: (0+ t 0.4- sec)


x1 = 56.913 [0.8 - 0.9364 sinx 2 ]
(16.16)
x 2 = x1 - 314.16
x1 (0+ ) = 314.16 rad / sec
x 2 (0+ ) = 0.3159 rad.
82
MODIFIED EULER STEP (Cont)

x( t+t )= x( t )+x(i t )* t;
p c
i=1, 2


i i

1
x( t+t )= x( t )+ x(i t )+x(i t+t ) * t; i=1, 2


c c p


2
i i

x(1 0 ) = 56.913[0.8 -0.9364 sin( 0.3159) ] = 28.974

x(2 0 ) = x(1 0)-314.16 =0

83
MODIFIED EULER STEP (Cont)

x(1p 0.05)= x(1c 0)+x(1 0)*0.05


= 314.16 + 28.974 * 0.05 = 315.6087 rad / sec

x(p2 0.05)= x(c2 0)+x(2 0)*0.05


= 0.3159 + 0 * 0.05 = 0.3159 rad

84
MODIFIED EULER STEP (Cont)

x( 0.05) = 56.913 [0.8 - 0.9364 sin x ]

p p
1 2

= 56.913 [0.8 - 0.9364 sin( 0.3159) ]


= 28.974 rad / sec

x(p2 0.05) = x(1p 0.05) - 314.16

= 315.6087 - 314.16
=1.4487 rad

85
MODIFIED EULER STEP (Cont)

c c p
x (0.05) = x (0) + 0.025 [x1 (0) + x (0.05)]
1 1 1

= 314.16 + 0.025 [28.974 + 28.974]


= 314.16 + 1.4487 = 315.6087 rad / sec

c c p
x (0.05) = x (0) + 0.025 [x 2 (0) + x (0.05)]
2 2 2

= 0.3159 + 0.025[0 + 1.4487]


o
= 0.35212 rad or 20.178
86
RESULTS
One time step is over, go for the next time step


x 1c (0.05) = ?


x c2 (0.05) = ?

Then x p1 (0.1) = ?

x p2 (0.1) = ? and so on until t becomes 0.4 sec


87
RESULTS (Cont)
At t = 0.4 sec fault is cleared and Pmax(post) =
2.06 p.u., simulation is carried out.
The rotor angle reaches 105o and then
oscillates around post-fault stable
operating point given by
s = sin-1 [0.8/2.08] = 22.85o.
Hence the system is stable.

88
PREDICTOR-CORRECTOR METHODS

The Modified Euler method is the simplest of


Predictor-Corrector (P-C) methods.
Adams-Bashforth method, Milne method, and
Hamming method are well known higher order
P-C methods.
Application of these methods to P.S. Stability
analysis has been found to suffer from a
number of limitations.
89
RUNGE KUTTA (R-K) METHODS

The R-K methods approximate the Taylor


series solution. However they do not require
explicit evaluation of derivatives higher than
the first.
The effects of higher derivatives are included
by several evaluations of the first derivative.

90
SECOND ORDER (R-K) METHOD
dx
= f( x, t) ( 16.19)
dt
The second order R -K formula for the value
of x at t = t0 + t is
k1 +k 2
x1 = x 0 + x = x 0 + ( 16.20)
2
where
k1 = f( x 0 , t0 ) * t
k 2 = f( x 0 +k1, t0 + t) * t
91
ALGORITHM

k1 + k 2
xn+1 = xn + ; n = 0,1, ...N max (16.21)
2

k 1 = f(xn , t n )* t

k 2 = f(xn + k 1 , t n + t)* t
92
FOURTH ORDER R-K METHOD

1
x n+1 = x n + k1 +2K 2 +2K 3 +K 4 ; ( 16.22)
6
n =0,1,....Nmax
where
k1 = f( x n , tn ) * t
k1 t
k2 = f( x n + , tn + ) * t
2 2

93
FOURTH ORDER R-K METHOD
(Cont)

k2 t
k 3 = f x n + , tn + * t
2 2
k 4 = f x n +k 3 , tn +t * t
Physical Interpretation of the above solution :
k1 = S lope at the beginning of time step * t
k 2 = first approx. to slope at midstep * t

94
FOURTH ORDER R-K METHOD
(Cont)
k3 = S econd approx. to slope at midstep * t
k 4 = S lope at the end of time step * t
1
x = k1 +2k 2 +2k3 +k 4
6
Thus x is the incremental value of x
given by the weighted average of estimates
based on slopes at the beginning, midpoint,
and end of the time step.
95

Anda mungkin juga menyukai