ECONOMIC OPERATION OF POWER SYSTEMS

INTRODUCTION
One of the earliest applications of on-line centralized control was to provide a central
facility, to operate economically, several generating plants supplying the loads of the
system. Modern integrated systems have different types of generating plants, such as coal
fired thermal plants, hydel plants, nuclear plants, oil and natural gas units etc. The capital
investment, operation and maintenance costs are different for different types of plants.

The operation economics can again be subdivided into two parts.

i) Problem of economic dispatch, which deals with determining the power
output of each plant to meet the specified load, such that the overall fuel cost
is minimized.
ii) Problem of optimal power flow, which deals with minimum – loss delivery,
where in the power flow, is optimized to minimize losses in the system. In
this chapter we consider the problem of economic dispatch.
During operation of the plant, a generator may be in one of the following states:
i) Base supply without regulation: the output is a constant.
ii) Base supply with regulation: output power is regulated based on system load.
iii) Automatic non-economic regulation: output level changes around a base
setting as area control error changes.
iv) Automatic economic regulation: output level is adjusted, with the area load
and area control error, while tracking an economic setting.

Regardless of the units operating state, it has a contribution to the economic operation,
even though its output is changed for different reasons. The factors influencing the cost
of generation are the generator efficiency, fuel cost and transmission losses. The most
efficient generator may not give minimum cost, since it may be located in a place where
fuel cost is high. Further, if the plant is located far from the load centers, transmission
losses may be high and running the plant may become uneconomical. The economic
dispatch problem basically determines the generation of different plants to minimize total
operating cost.

1
Modern generating plants like nuclear plants, geo-thermal plants etc, may require capital
investment of millions of rupees. The economic dispatch is however determined in terms
of fuel cost per unit power generated and does not include capital investment,
maintenance, depreciation, start-up and shut down costs etc.
PERFORMANCE CURVES
INPUT-OUTPUT CURVE

This is the fundamental curve for a thermal plant and is a plot of the input in British
thermal units (Btu) per hour versus the power output of the plant in MW as shown in Fig
1.
Btu / hr (Input)

(output) MW
Fig 1: Input – output curve

HEAT RATE CURVE

The heat rate is the ratio of fuel input in Btu to energy output in KWh. It is the slope of
the input – output curve at any point. The reciprocal of heat – rate is called fuel –
efficiency. The heat rate curve is a plot of heat rate versus output in MW. A typical plot
is shown in Fig .2
(Heat rate) Btu / kw-hr

(output) MW 2
 Fig .2 Heat rate curve.

INCREMENTAL FUEL RATE CURVE

The incremental fuel rate is equal to a small change in input divided by the
corresponding change in output.
Input
Incremental fuel rate =
Output

The unit is again Btu / KWh. A plot of incremental fuel rate versus the output is shown in
Fig 3
Incremental fuel rate

(output) MW

Fig 3: Incremental fuel rate curve

Incremental cost curve

The incremental cost is the product of incremental fuel rate and fuel cost (Rs / Btu or $ /
Btu). The curve in shown in Fig. 4. The unit of the incremental fuel cost is Rs / MWh or
$ /MWh.

3
 approximate linear cost

actual cost

Rs / Mwhr

(output) MW

Fig 4: Incremental cost curve

In general, the fuel cost Fi for a plant, is approximated as a quadratic function of the
generated output PGi.

Fi = ai + bi PGi + ci PGi2 Rs / h

The incremental fuel cost is given by

dFi
= bi + 2ci PGi Rs / MWh
dPGi
The incremental fuel cost is a measure of how costly it will be produce an increment of
power. The incremental production cost, is made up of incremental fuel cost plus the
incremental cost of labour, water, maintenance etc. which can be taken to be some
percentage of the incremental fuel cost, instead of resorting to a rigorous mathematical
model. The cost curve can be approximated by a linear curve. While there is negligible
operating cost for a hydel plant, there is a limitation on the power output possible. In any
plant, all units normally operate between PGmin, the minimum loading limit, below which
it is technically infeasible to operate a unit and PGmax, which is the maximum output
limit.

ECONOMIC GENERATION SCHEDULING NEGLECTING

LOSSES AND GENERATOR LIMITS
The simplest case of economic dispatch is the case when transmission losses are
neglected. The model does not consider the system configuration or line impedances.
Since losses are neglected, the total generation is equal to the total demand PD.

4
 Consider a system with ng number of generating plants supplying the total
demand PD. If Fi is the cost of plant i in Rs/h, the mathematical formulation of the
problem of economic scheduling can be stated as follows:
ng
Minimize FT = F
i 1
i

ng
Such that P
i 1
Gi  PD

where FT = total cost.

PGi = generation of plant i.
PD = total demand.

This is a constrained optimization problem, which can be solved by Lagrange’s method.

LAGRANGE METHOD FOR SOLUTION OF ECONOMIC SCHEDULE

The problem is restated below:

ng

Minimize FT   Fi
i 1

ng

Such that PD  P
i 1
Gi 0

The augmented cost function is given by

 ng

£  FT   PD   PGi 
 i 1 
The minimum is obtained when
£ £
 0 and 0
PGi 

£ F
 T   0
PGi PGi
£ g n

 PD   PGi  0
 i 1

5
The second equation is simply the original constraint of the problem. The cost of a plant
Fi depends only on its own output PGi, hence
FT Fi dFi
 
PGi PGi dPGi
Using the above,
Fi dFi
  ; i = 1……. ng
PGi dPGi

We can write

bi + 2ci PGi =  i = 1……. ng

The above equation is called the co-ordination equation. Simply stated, for economic
generation scheduling to meet a particular load demand, when transmission losses are
neglected and generation limits are not imposed, all plants must operate at equal
incremental production costs, subject to the constraint that the total generation be equal
to the demand. From we have
  bi
PGi 
2c i
We know in a loss less system
ng

P
i 1
Gi  PD

Substituting (8.16) we get

  bi
ng


i 1 2ci
 PD

An analytical solution of λ is obtained from (8.17) as

n g
bi
PD  
i1 2ci
  n g
1

i1 2ci
It can be seen that  is dependent on the demand and the coefficients of the cost function.
Example 1.
The fuel costs of two units are given by

6
F1 = 1.5 + 20 PG1 + 0.1 PG12 Rs/h
F2 = 1.9 + 30 PG2 + 0.1 PG22 Rs/h
PG1, PG2 are in MW. Find the optimal schedule neglecting losses, when the demand is
200 MW.

Solution:
dF1
 20  0.2 PG1 Rs / MWh
dPG1

dF2
 30  0.2 PG 2 Rs / MWh
dPG 2

PD  PG1  PG 2  200 MW
For economic schedule
dF1 dF2
 
dPG1 dPG 2
20 + 0.2 PG1 = 30 + 0.2 (200 - PG1)
Solving we get, PG1 = 125 MW
PG2 = 75 MW

 = 20 + 0.2 (125) = 45 Rs / MWh

Example 2
The fuel cost in $ / h for two 800 MW plants is given by
F1 = 400 + 6.0 PG1 + 0.004 PG12
F2 = 500 + b2 PG2 + c2 PG22
where PG1, PG2 are in MW
(a) The incremental cost of power,  is $8 / MWh when total demand is 550MW.
Determine optimal generation schedule neglecting losses.
(b) The incremental cost of power is $10/MWh when total demand is 1300 MW.
Determine optimal schedule neglecting losses.
(c) From (a) and (b) find the coefficients b2 and c2.

Solution:
  b1 8.0  6.0
a) PG1    250 MW
2c1 2  0.004

PG 2  PD  PG1  550  250  300 MW

7
   b1 10  6
b) PG1    500 MW
2C1 2  0.004

PG 2  PD  PG1  1300  500  800 MW

  b2
c) PG 2 
2c 2
8.0  b2
From (a) 300 
2c 2
10.0  b2
From (b) 800 
2c 2

Solving we get b2 = 6.8

c2 = 0.002

ECONOMIC SCHEDULE INCLUDING LIMITS ON GENERATOR

(NEGLECTING LOSSES)
The power output of any generator has a maximum value dependent on the rating of the
generator. It also has a minimum limit set by stable boiler operation. The economic
dispatch problem now is to schedule generation to minimize cost, subject to the equality
constraint.
ng

P
i 1
Gi  PD

and the inequality constraint

PGi (min) ≤ PGi ≤ PGi (max) ; i = 1, ……… ng

The procedure followed is same as before i.e. the plants are operated with equal
incremental fuel costs, till their limits are not violated. As soon as a plant reaches the
limit (maximum or minimum) its output is fixed at that point and is maintained a
constant. The other plants are operated at equal incremental costs.
Example 3
Incremental fuel costs in $ / MWh for two units are given below:
dF1
 0.01PG1  2.0 $ / MWh
dPG1

8
dF2
 0.012 PG 2  1.6 $ / MWh
dPG 2
The limits on the plants are Pmin = 20 MW, Pmax = 125 MW. Obtain the optimal schedule
if the load varies from 50 – 250 MW.
Solution:
The incremental fuel costs of the two plants are evaluated at their lower limits and upper
limits of generation.
At PG (min) = 20 MW.
dF1
 1(min)  = 0.01x 20+2.0 = 2.2$ / MWh
dPG1

dF2
 2 (min)  = 0.012 x 20 + 1.6 = 1.84 $ / MWh
dPG 2
At PG (Max) =125 Mw
1(max) = 0.01 x 125 + 2.0 = 3.25 $ / MWh
2(max) = 0.012 x 125 + 1.6 = 3.1 $ / MWh
Now at light loads unit 1 has a higher incremental cost and hence will operate at its lower
limit of 20 MW. Initially, additional load is taken up by unit 2, till such time its
incremental fuel cost becomes equal to 2.2$ / MWh at PG2 = 50 MW. Beyond this, the
two units are operated with equal incremental fuel costs. The contribution of each unit to
meet the demand is obtained by assuming different values of ; When  = 3.1 $ / MWh,
unit 2 operates at its upper limit. Further loads are taken up by unit 1. The computations
are show in Table
Table Plant output and output of the two units
dF1 dF2 Plant  PG1 PG2 Plant Output
dPG1 dPG 2
MW
$/MWh MW MW
$/MWh $/MWh
2.2 1.96 1.96 20+ 30 50
+
2.2 2.2 2.2 20 50 70
2.4 2.4 2.4 40 66.7 106.7
2.6 2.6 2.6 60 83.3 143.3
2.8 2.8 2.8 80 100 180
3.0 3.0 3.0 100 116.7 216.7
3.1 3.1 3.1 110 125* 235

9
 3.25 3.1 3.25 125* 125* 250

For a particular value of , PG1 and PG2 are calculated using (8.16). Fig 8.5 Shows plot of
each unit output versus the total plant output.

For any particular load, the schedule for each unit for economic dispatch can be obtained
.

Example 4.

In example 3, what is the saving in fuel cost for the economic schedule compared to the
case where the load is shared equally. The load is 180 MW.

Solution:

From Table it is seen that for a load of 180 MW, the economic schedule is P G1 = 80 MW
and PG2 = 100 MW. When load is shared equally PG1 = PG2 = 90 MW. Hence, the
generation of unit 1 increases from 80 MW to 90 MW and that of unit 2 decreases from
100 MW to 90 MW, when the load is shared equally. There is an increase in cost of unit
1 since PG1 increases and decrease in cost of unit 2 since PG2 decreases.

10
 90
 dF1 
Increase in cost of unit 1 = 80  dPG1 dPG1
90
=
80
 0.01P G1  2.0 dPG1  28.5 $ / h

90
 dF 2 
Decrease in cost of unit 2 =   dP
100
dPG 2
G2 

90
=  0.012 P
100
G2  1.6 dPG 2  27.4 $ / h

Total increase in cost if load is shared equally = 28.5 – 27.4 = 1.1 $ / h

Hence the saving in fuel cost is 1.1 $ / h if coordinated economic schedule is used.

ECONOMIC DISPATCH INCLUDING TRANSMISSION LOSSES

When transmission distances are large, the transmission losses are a significant part of
the generation and have to be considered in the generation schedule for economic
operation. The mathematical formulation is now stated as
ng
Minimize FT   Fi
i 1

ng
Such That P
i 1
Gi  PD  PL

where PL is the total loss.

The Lagrange function is now written as
 ng 
£ = FT     PGi  PD  PL   0
 i 1 
The minimum point is obtained when

£ F  P 
 T   1  L   0 ; i = 1……ng
PGi PGi  PGi 

£ g n

  PGi  PD  PL  0 (Same as the constraint)

 i 1
FT dFi
Since  , (8.27) can be written as
PGi dPGi

dFi P
 L 
dPGi PGi

11
  
 
dFi  1 

dPGi  1  PL 
 
 PGi 

1
The term is called the penalty factor of plant i, Li. The coordination
PL
1
P Gi

equations including losses are given by

dFi
 L i ; i = 1…….ng
dPGi

The minimum operation cost is obtained when the product of the incremental fuel cost
and the penalty factor of all units is the same, when losses are considered.

A rigorous general expression for the loss PL is given by

PL =  m n PGm Bmn PGn + n PGn Bno + Boo

where Bmn, Bno , Boo called loss – coefficients , depend on the load composition. The
assumption here is that the load varies linearly between maximum and minimum values.
A simpler expression is
PL =  m n PGm Bmn PGn

The expression assumes that all load currents vary together as a constant complex
fraction of the total load current. Experiences with large systems has shown that the loss
of accuracy is not significant if this approximation is used.

An average set of loss coefficients may be used over the complete daily cycle in the
coordination of incremental production costs and incremental transmission losses. In
general, Bmn = Bnm and can be expanded for a two plant system as
PL = B11 PG1 + 2 B12 PG1 PG2 + B22 PG22

12
Example 5
A generator is supplying a load. An incremental change in load of 4 MW requires
generation to be increased by 6 MW. The incremental cost at the plant bus is Rs 30 /
MWh. What is the incremental cost at the receiving end?
Solution:

dF1
= 30
dP G1

dF1
 30
dPG1
∆PL = 2MW
G Load

∆PG = 6MW ∆PD = 4MW

Fig ; One line diagram of example 5

 PL =  PG -  PD = 2 MW
 at receiving end is given by
dF1 PG 6
   30   45 Rs / MWh
dPG1 PD 4

dF1 1 1
or     30   45 Rs / MWh
dPG1 PL 2
1 1
PG 6
Example 6
In a system with two plants, the incremental fuel costs are given by
dF1
 0.01PG1  20 Rs / MWh
dPG1

dF2
 0.015 PG 2  22.5 Rs / MWh
dPG 2
The system is running under optimal schedule with PG1 = PG2 = 100 MW.
PL P
If = 0.2, find the plant penalty factors and L .
PG2 PG1

13
Solution:
For economic schedule,
dFi 1
Li   ; Li  ;
dPGi PL
1
PGi
For plant 2, PG2 = 100 MW

 0.015  100  22.5

1
 .
1  0 .2
Solving,   30 Rs / MWh
1
L2 =  1.25
1  0 .2
dF1
L1    (0.01x100+20) L1 = 30
dPG1
L1 = 1.428
1
L1 =
P
1 L
PG1

1 PL
1.428 = ; Solving = 0.3
PL PG1
1
PG1
Example 7

A two bus system is shown in Fig. 8.8 If 100 MW is transmitted from plant 1 to the load,
a loss of 10 MW is incurred. System incremental cost is Rs 30 / MWh. Find PG1, PG2 and
power received by load if
dF1
 0.02 PG1  16.0 Rs / MWh
dPG1

dF2
 0.04 PG 2  20.0 Rs / MWh
dPG 2
PG1
PG 2

G1 G2

Load

14
 Fig One line diagram of example 7

Solution;
Since the load is connected at bus 2 , no loss is incurred when plant two supplies the
load.
Therefore in (8.36) B12 = 0 and B22 = 0
PL PL
PL  B11 PG1 ;  2 B11 PG1 ;  0 .0
2

PG1 PG2
From data we have PL = 10 MW, if PG1 = 100 MW
10 = B11 (100)2
B11 = 0.001 MW-1
Coordination equation with loss is
dFi P
 L 
dPGi PGi

dF1 P
For plant 1  L 
dPG1 PG1
(0.02 PG1 +16.0) + 30 (2 x 0.001 x PG1) = 30

0.08 PG1 = 30 - 16.0. From which, PG1 = 175 MW

dF2 P
For Plant 2  L 
dPG1 PG2

0.04 PG2 + 20.0 = 30 or PG2 = 250 MW

Loss = B11 PG12 = 0.001 x (175)2 = 30.625 MW

PD = (PG1 + PG2) – PL = 394. 375 MW

DERIVATION OF TRANSMISSION LOSS FORMULA

An accurate method of obtaining general loss coefficients has been presented

by Kron. The method is elaborate and a simpler approach is possible by making the
following assumptions:
(i) All load currents have same phase angle with respect to a common reference
(ii) The ratio X / R is the same for all the network branches.

15
Consider the simple case of two generating plants connected to an arbitrary number of
loads through a transmission network as shown in Fig a

IG1
1

IG2
IK ID
2

(a)

IG1 = ID
1

IK1 ID
2
IG2 = 0
(b)

IG1 = 0
1

IG2 = ID
IK2 ID
2

(c)
Fig Two plants connected to a number of loads through a transmission network

Let’s assume that the total load is supplied by only generator 1 as shown in Fig 8.9b. Let
the current through a branch K in the network be IK1. We define
I K1
N K1 
ID
It is to be noted that IG1 = ID in this case. Similarly with only plant 2 supplying the load
current ID, as shown in Fig 8.9c, we define
IK2
NK2 
ID

16
NK1 and NK2 are called current distribution factors and their values depend on the
impedances of the lines and the network connection. They are independent of I D. When
both generators are supplying the load, then by principle of superposition
IK = NK1 IG1 + NK2 IG2

where IG1, IG2 are the currents supplied by plants 1 and 2 respectively, to meet the
demand ID. Because of the assumptions made, IK1 and ID have same phase angle, as do
IK2 and ID. Therefore, the current distribution factors are real rather than complex. Let
I G1  I G1  1 and I G 2  I G 2  2 .

where  1 and  2 are phase angles of IG1 and IG2 with respect to a common reference. We
can write

 N K 1 I G1 cos  1  N K 2 I G 2 cos  2   N K 1 I G1 sin  1  N K 2 I G 2 sin  2 

2 2 2
IK

=
2 2

N K 1 I G1 cos 2  1  sin 2  1  N K 2 I G 2  2 2
cos 2
 2  sin 2  2 
 2N K 1 I G1 cos  1 N K 2 I G 2 cos  2  N K 1 I G1 sin  1 N K 2 I G 2 sin  2 

= N K 1 I G1  N K 2 I G 2  2 N K 1 N K 2 I G1 I G 2 cos 1   2 
2 2 2 2

PG1 PG 2
Now I G1  and I G 2 
3 V1 cos 1 3 V2 cos  2
where PG1, PG2 are three phase real power outputs of plant1 and plant 2; V 1, V2 are the
line to line bus voltages of the plants and 1 , 2 are the power factor angles.
The total transmission loss in the system is given by

3 I
2
PL = K RK
K
where the summation is taken over all branches of the network and R K is the branch
resistance. Substituting we get

2 PG1 PG 2 cos 1   2 
2
PG1
PL  N RK   N K 1 N K 2 RK
2

cos   V1 V2 cos 1 cos  2 K

2 2 K1
V1 1 K

2
PG 2
 N
2
RK
cos  2 
2 2 K2
V2 K

PL  PG1 B11  2 PG1 PG 2 B12  PG 2 B22

2 2

1
B11  N
2
where RK
cos  
2 2 K1
V1 1 K

17
 cos 1   2 
B12 
V1 V2 cos 1 cos  2
N
K
K1 N K 2 RK

1
B22  N
2
RK
cos  2 
2 2 K2
V2 K

The loss – coefficients are called the B – coefficients and have unit MW-1.
For a general system with n plants the transmission loss is expressed as

2 2
PG1 PGn
PL   N K1  ........  N
2 2
RK
cos 1  cos  n 
2 2 2 2 Kn
V1 K Vn K

n PGP PGq cos p   q 

2 N KP N Kq RK
p , q 1 V p Vq cos  p cos  q k
pq

In a compact form
n n
PL   PGp B Pq PGq
p 1 q 1

cos p   q 
B Pq 
V p Vq cos  P cos  q
N
K
KP N Kq R K

B – Coefficients can be treated as constants over the load cycle by computing them at
average operating conditions, without significant loss of accuracy.

Example 8
Calculate the loss coefficients in pu and MW-1 on a base of 50MVA for the network of
Fig below. Corresponding data is given below.
Ia = 1.2 – j 0.4 pu Za = 0.02 + j 0.08 pu
Ib = 0.4 - j 0.2 pu Zb = 0.08 + j 0.32 pu
Ic = 0.8 - j 0.1 pu Zc = 0.02 + j 0.08 pu
Id = 0.8 - j 0.2 pu Zd = 0.03 + j 0.12 pu
Ie = 1.2 - j 0.3 pu Ze = 0.03 + j 0.12 pu

18
 Vref = 1.0  00
1 2
G1 a b c G2
Ia Ib Ic
I1 v I2

d Id e Ie

Load 1 Load 2
Fig : Example 8

Solution:
Total load current
IL = Id + Ie = 2.0 – j 0.5 = 2.061  -14.030A
IL1 = Id = 0.8 - j 0.2 = 0.8246  -14.030 A
I L1 I L2
 0.4;  1 .0  0 . 4  0 . 6
IL IL

If generator 1, supplies the load then I1 = IL. The current distribution is shown in Fig a.

a b c
G1 G2
IL 0.6 IL Ic = 0
IL I2 = 0

d 0.4 IL e 0.6 IL

Load 1 Load 2
Fig a : Generator 1 supplying the total load

Ia I
N a1   1.0; N b1  b  0.6; N C1  0; N d 1  0.4; N e1  0.6.
IL IL

Similarly the current distribution when only generator 2 supplies the load is shown in Fig
b.

19
 a b c
G1 G2
Ia = 0
0.4 IL IL
IL

d 0.4 IL e 0.6 IL

Load 1 Load 2
Fig b: Generator 2 supplying the total load

Na2 =0; Nb2 = -0.4; Nc2 = 1.0; Nd2 = 0.4; Ne2 = 0.6
From Fig 8.10, V1 = Vref + ZaIa
= 1  00 + (1.2 – j 0.4) (0.02 + j0.08)
= 1.06  4.780 = 1.056 + j 0.088 pu.
V2 = Vref – Ib Zb + Ic Zc

= 1.0  00 – (0.4 – j 0.2) (0.08 + j 0.32) + (0.8 – j 0.1) (0.02 + j 0.08)

= 0.928 – j 0.05 = 0.93  -3.100 pu.

Current Phase angles

  0 .4 
 1  angle of I1(=Ia) = tan-1    18.43
0

 1 . 2 
  
 2  angle of I 2  I c   tan 1 
0 . 1
  7.13
0

 0 . 8 
cos 1   2   0.98
Power factor angles
1  4.78 0  18.43  23.210 ; cos 1  0.92

 2  7.130  3.10 0  4.030 ; cos  2  0.998

N
2
RK
K1
1.0 2  0.02  0.6 2  0.08  0.4 2  0.03  0.6 2  0.03
B11  K

V1
2
cos 1 2 1.062 0.9202

= 0.0677 pu
1
= 0.0677 x  0.1354  10  2 MW-1
50
Cos  1   2 
B12   N K 1 N K 2 RK
V1 V2 cos 1 cos  2  K

20
=
0.98
 0.4  0.6  0.08  0.4  0.4  0.03  0.6  0.6  0.03
1.060.930.9980.92
= - 0.00389 pu
= - 0.0078 x 10-2 MW-1

N
2
K2 RK
B22 
cos 
K
2 2
V2 2


 0.42 0.08  1.0 2  0.02  0.4 2  0.03  0.6 2  0.03
0.932 0.9982
= 0.056pu  0.112  10 2 MW-1

21
 1

POWER SYSTEM STABILITY

INTRODUCTION:
Power system stability of modern large inter-connected systems is a major
problem for secure operation of the system. Recent major black-outs across the globe
caused by system instability, even in very sophisticated and secure systems, illustrate the
problems facing secure operation of power systems. Earlier, stability was defined as the
ability of a system to return to normal or stable operation after having been subjected to
some form of disturbance. This fundamentally refers to the ability of the system to
remain in synchronism. However, modern power systems operate under complex
interconnections, controls and extremely stressed conditions. Further, with increased
automation and use of electronic equipment, the quality of power has gained utmost
importance, shifting focus on to concepts of voltage stability, frequency stability,
inter-area oscillations etc.

The IEEE/CIGRE Joint Task Force on stability terms and conditions have
proposed the following definition in 2004: “Power System stability is the ability of an
electric power system, for a given initial operating condition, to regain a state of
operating equilibrium after being subjected to a physical disturbance, with most system
variables bounded, so that practically the entire system remains intact”.

The Power System is an extremely non-linear and dynamic system, with operating
parameters continuously varying. Stability is hence, a function of the initial operating
condition and the nature of the disturbance. Power systems are continually subjected to
small disturbances in the form of load changes. The system must be in a position to be
able to adjust to the changing conditions and operate satisfactorily. The system must also
withstand large disturbances, which may even cause structural changes due to isolation of
some faulted elements.

A power system may be stable for a particular (large) disturbance and unstable for
another disturbance. It is impossible to design a system which is stable under all
 2

disturbances. The power system is generally designed to be stable under those

disturbances which have a high degree of occurrence. The response to a disturbance is
extremely complex and involves practically all the equipment of the power system. For
example, a short circuit leading to a line isolation by circuit breakers will cause variations
in the power flows, network bus voltages and generators rotor speeds. The voltage
variations will actuate the voltage regulators in the system and generator speed variations
will actuate the prime mover governors; voltage and frequency variations will affect the
system loads. In stable systems, practically all generators and loads remain connected,
even though parts of the system may be isolated to preserve bulk operations. On the other
hand, an unstable system condition could lead to cascading outages and a shutdown of a
major portion of the power system.

ROTOR ANGLE STABILITY

Rotor angle stability refers to the ability of the synchronous machines of an
interconnected power system to remain in synchronism after being subjected to a
disturbance. Instability results in some generators accelerating (decelerating) and losing
synchronism with other generators. Rotor angle stability depends on the ability of each
synchronous machine to maintain equilibrium between electromagnetic torque and
mechanical torque. Under steady state, there is equilibrium between the input mechanical
torque and output electromagnetic torque of each generator, and its speed remains a
constant. Under a disturbance, this equilibrium is upset and the generators
accelerate/decelerate according to the mechanics of a rotating body. Rotor angle stability
is further categorized as follows:

Small single (or small disturbance) rotor angle stability

It is the ability of the power system to maintain synchronism under small
disturbances. In this case, the system equation can be linearized around the initial
operating point and the stability depends only on the operating point and not on the
disturbance. Instability may result in
(i) A non oscillatory or a periodic increase of rotor angle
(ii) Increasing amplitude of rotor oscillations due to insufficient damping.
 3

The first form of instability is largely eliminated by modern fast acting voltage regulators
and the second form of instability is more common. The time frame of small signal
stability is of the order of 10-20 seconds after a disturbance.

Large-signal rotor angle stability or transient stability

This refers to the ability of the power system to maintain synchronism under large
disturbances, such as short circuit, line outages etc. The system response involves large
excursions of the generator rotor angles. Transient stability depends on both the initial
operating point and the disturbance parameters like location, type, magnitude etc.
Instability is normally in the form of a periodic angular separation. The time frame of
interest is 3-5 seconds after disturbance.

The term dynamic stability was earlier used to denote the steady-state stability in
the presence of automatic controls (especially excitation controls) as opposed to manual
controls. Since all generators are equipped with automatic controllers today, dynamic
stability has lost relevance and the Task Force has recommended against its usage.

MECHANICS OF ROTATORY MOTION

Since a synchronous machine is a rotating body, the laws of mechanics of rotating
bodies are applicable to it. In rotation we first define the fundamental quantities. The
angle θm is defined, with respect to a circular arc with its center at the vertex of the angle,
as the ratio of the arc length s to radius r.
s
θm = (1)
r
The unit is radian. Angular velocity ωm is defined as
d m
ωm = (2)
dt
and angular acceleration as
d m d 2 m
  (3)
dt dt 2
 4

The torque on a body due to a tangential force F at a distance r from axis of rotation is
given by
T=rF (4)
The total torque is the summation of infinitesimal forces, given by
T = ∫ r dF (5)
The unit of torque is N-m. When torque is applied to a body, the body experiences
angular acceleration. Each particle experiences a tangential acceleration a  r , where r
is the distance of the particle from axis of rotation. The tangential force required to
accelerate a particle of mass dm is
dF = a dm = r α dm (6)

The torque required for the particle is

dT = r dF = r2 α dm (7)
and that required for the whole body is given by
T = α ∫ r2dm = I α (8)
Here
I = ∫ r2dm (9)
is called the moment of inertia of the body. The unit is Kg – m2. If the torque is assumed
to be the result of a number of tangential forces F, which act at different points of the
body
T=∑rF
Now each force acts through a distance
ds = r dθm
The work done is ∑ F . ds
dW = ∑ F r dθm = dθm T
W = ∫ T dθ m (10)
dW
and T= (11)
d m
Thus the unit of torque may also be Joule per radian.
The power is defined as rate of doing work. Using (11)
 5

d W T d m
P=   T m (12)
dt dt
The angular momentum M is defined as
M = I ωm (13)
and the kinetic energy is given by

1 1
I  m = M ωm
2
KE = (14)
2 2
From (14) we can see that the unit of M is seen to be J-sec/rad.
SWING EQUATION:
From (8)
I = T
I d 2 m
or T (15)
d t2
Here T is the net torque of all torques acting on the machine, which includes the shaft
torque (due to prime mover of a generator or load on a motor), torque due to rotational
losses (friction, windage and core loss) and electromagnetic torque.

Let Tm = shaft torque or mechanical torque corrected for rotational losses

Te = Electromagnetic or electrical torque

For a generator Tm tends to accelerate the rotor in positive direction of rotation and for a
motor retards the rotor.

The accelerating torque for a generator

Ta = Tm  Te (16)
Under steady-state operation of the generator, Tm is equal to Te and the accelerating
torque is zero. There is no acceleration or deceleration of the rotor masses and the
machines run at a constant synchronous speed. In the stability analysis in the following
sections, Tm is assumed to be a constant since the prime movers (steam turbines or hydro
turbines) do no act during the short time period in which rotor dynamics are of interest in
the stability studies.
 6

Now (15) has to be solved to determine  m as a function of time. Since  m is

measured with respect to a stationary reference axis on the stator, it is the measure of the
absolute rotor angle and increases continuously with time even at constant synchronous
speed. Since machine acceleration /deceleration is always measured relative to
synchronous speed, the rotor angle is measured with respect to a synchronously rotating
reference axis. Let
 m   m   sm t (17)

where  sm is the synchronous speed in mechanical rad/s and  m is the angular

displacement in mechanical radians.
Taking the derivative of (17) we get

d m d m
   sm
dt dt
d 2 m d 2 m
 (18)
dt 2 dt 2
Substituting (18) in (15) we get

d 2 m
I = Ta = Tm  Te N-m (19)
dt 2
Multiplying by  m on both sides we get

d 2 m
m I =  m ( Tm  Te ) N-m (20)
dt 2
From (12) and (13), we can write
d 2 m
M  Pm  Pa W (21)
dt 2
where M is the angular momentum, also called inertia constant
Pm = shaft power input less rotational losses
Pe = Electrical power output corrected for losses
Pa = acceleration power
 7

M depends on the angular velocity  m , and hence is strictly not a constant, because  m
deviates from the synchronous speed during and after a disturbance. However, under
stable conditions  m does not vary considerably and M can be treated as a constant. (21)
is called the “Swing equation”. The constant M depends on the rating of the machine and
varies widely with the size and type of the machine. Another constant called H constant
(also referred to as inertia constant) is defined as
stored kinetic energy in mega joules
at sychronous speed
H= MJ / MVA (22)
Machine rating in MVA
H falls within a narrow range and typical values are given in Table 9.1.
If the rating of the machine is G MVA, from (22) the stored kinetic energy is GH
Mega Joules. From (14)
1
GH = M s m MJ (23)
2
or
2 GH
M= MJ-s/mech rad (24)
s m

The swing equation (21) is written as

2 H d 2 m Pa Pm  Pe
  (25)
s m d t 2 G G

In (.25)  m is expressed in mechanical radians and  s m in mechanical radians per second

(the subscript m indicates mechanical units). If  and  have consistent units then mec

2 H d 2
 Pa  Pm Pe pu (26)
 s dt 2

 p
Here  s is the synchronous speed in electrical rad/s (  s     s m ) and Pa is
2
acceleration power in per unit on same base as H. For a system with an electrical
frequency f Hz, (26) becomes
H d 2
 Pa  Pm Pe pu (27)
 f dt 2
when  is in electrical radians and
 8

H d 2
 Pa  Pm Pe pu (28)
180 f dt 2
when  is in electrical degrees.
(27) and (28) also represent the swing equation. It can be seen that the swing equation is a
second order differential equation which can be written as two first order differential
equations:
2H d 
 Pm Pe pu (29)
 s dt

d
  s (30)
dt
in which  ,  s and  are in electrical units. In deriving the swing equation, damping
has been neglected.
Table 1 : H constants of synchronous machines
Type of machine H (MJ/MVA)
Turbine generator condensing 1800 rpm 9–6
3600 rpm 7–4
Non condensing 3600 rpm 4–3
Water wheel generator
Slow speed < 200 rpm 2–3
High speed > 200 rpm 2–4
Synchronous condenser
Large 1.25 25% less for hydrogen cooled
1.0 
Small
Synchronous motor with load varying
from 1.0 to 5.0 2.0

In defining the inertia constant H, the MVA base used is the rating of the machine. In a
multi machine system, swing equation has to be solved for each machine, in which case,
a common MVA base for the system has to chosen. The constant H of each machine must
be consistent with the system base.
Let
 9

Gmach = Machine MVA rating (base)

Gsystem = System MVA base
In (9.25), H is computed on the machine rating G  Gmach

Gmach
Multiplying (9.25) by on both sides we get
Gsystem

 Gmach  2 H d 2 m Pm  Pe  Gmach 
     (31)
G  2
Gmach G 
 system  s m dt  system 

2 H system d 2  m
 Pm  Pe pu (on system base)
s m dt 2

Gmach
where H system = H (32)
Gsystem

In the stability analysis of a multi machine system, computation is considerably

reduced if the number of swing equations to be solved is reduced. Machines within a
plant normally swing together after a disturbance. Such machines are called coherent
machines and can be replaced by a single equivalent machine, whose dynamics reflects
the dynamics of the plant.
Example 1:
A 50Hz, 4 pole turbo alternator rated 150 MVA, 11 kV has an inertia constant of
9 MJ / MVA. Find the (a) stored energy at synchronous speed (b) the rotor acceleration if
the input mechanical power is raised to 100 MW when the electrical load is 75 MW, (c)
the speed at the end of 10 cycles if acceleration is assumed constant at the initial value.
Solution:
(a) Stored energy = GH = 150 × 9 = 1350 MJ
(b) Pa = Pm – Pe = 100 – 75 = 25 MW
GH 1350
M=   0.15 MJ – s /ºe
180 f 180  50

d2
0.15  25
d t2
 10

d 2 25
Acceleration   2
 166.6 ºe/s2
dt 0.15
2
= 166.6 × ºm/s2
P
2 1
= 166.6 × × rps /s
P 360
2 1
= 166.6 × × × 60 rpm/s
P 360
= 13.88 rpm/s
* Note ºe = electrical degree; ºm = mechanical degree; P=number of poles.
10
(c) 10 cycles =  0 .2 s
50
120  50
NS = Synchronous speed =  1500 rpm
4
Rotor speed at end of 10 cycles = NS + α × 0.2
= 1500 + 13.88 × 0.2 = 1502.776 rpm
Example 2:
Two 50 Hz generating units operate in parallel within the same plant, with the
following ratings:
Unit 1: 500 MVA, 0.8 pf, 13.2 kV, 3600 rpm: H = 4 MJ/MVA
Unit 2: 1000 MVA, 0.9 pf, 13.8 kV, 1800 rpm: H = 5 MJ/MVA
Calculate the equivalent H constant on a base of 100 MVA.
Solution:
G1 mach
H 1system  H 1 mach 
Gsystem

500
= 4  20 MJ/MVA
100
G2 mach
H 2 system  H 2 mach 
Gsystem

1000
= 5  50 MJ/MVA
100
H eq  H 1  H 2 = 20 + 50 = 70 MJ/MVA
 11

This is the equivalent inertia constant on a base of 100 MVA and can be used
when the two machines swing coherently.

POWER–ANGLE EQUATION:
In solving the swing equation, certain assumptions are normally made
(i) Mechanical power input Pm is a constant during the period of interest,
immediately after the disturbance
(ii) Rotor speed changes are insignificant.
(iii) Effect of voltage regulating loop during the transient is neglected i.e the
excitation is assumed to be a constant.
As discussed in section 9.4, the power–angle relationship plays a vital role in the
solution of the swing equation.

POWER–ANGLE EQUATION FOR A NON–SALIENT POLE MACHINE:

The simplest model for the synchronous generator is that of a constant voltage
behind an impedance. This model is called the classical model and can be used for
cylindrical rotor (non–salient pole) machines. Practically all high–speed turbo alternators
are of cylindrical rotor construction, where the physical air gap around the periphery of
the rotor is uniform. This type of generator has approximately equal magnetic reluctance,
regardless of the angular position of the rotor, with respect to the armature mmf.
r
The power output of the generator is given by the real part of Eg Ia* .

E g   Vt 0
Ia  (38)
Ra  jx d

E g   Vt 0
Neglecting Ra, I a 
jxd

  E g 90   Vt 90  
*

P = R E g      
  xd xd  

 12

E g cos 90 E g Vt cos 90   

2

= 
xd xd

E g Vt sin 
= (39)
xd

(Note- R stands for real part of)

E g Vt
The maximum power that can be transferred for a particular excitation is given by
xd
at δ = 90o.
POWER ANGLE EQUATION FOR A SALIENT POLE MACHINE:
Here because of the salient poles, the reluctance of the magnetic circuit in which flows
the flux produced by an armature mmf in line with the quadrature axis is higher than that
of the magnetic circuit in which flows the flux produced by the armature mmf in line with
the direct axis. These two components of armature mmf are proportional to the
corresponding components of armature current. The component of armature current
producing an mmf acting in line with direct axis is called the direct component, Id. The
component of armature current producing an mmf acting in line with the quadrature axis
is called the quadrature axis component, Iq.
Power output P  Vt I a cos

 Ed I d  Eq I q (40)

E d  Vt sin  (41a)

E q  Vt cos  (41b)

E g  Eq
Id   I a sin     (41c)
xd
Ed
Iq   I a cos    (41d)
xq

Substituting (9.41c) and (9.41d) in (9.40), we obtain

E g Vt sin  Vt x d  x q sin 2
2

P  (42)
xd 2 xd xq
 13

(9.42) gives the steady state power angle relationship for a salient pole machine. The
second term does not depend on the excitation and is called the reluctance power
component. This component makes the maximum power greater than in the classical
model. However, the angle at which the maximum power occurs is less than 90 o.
TRANSIENT STABILITY:
As defined earlier, transient stability is the ability of the system to remain stable under
large disturbances like short circuits, line outages, generation or load loss etc. The
evaluation of the transient stability is required offline for planning, design etc. and online
for load management, emergency control and security assessment. Transient stability
analysis deals with actual solution of the nonlinear differential equations describing the
dynamics of the machines and their controls and interfacing it with the algebraic
equations describing the interconnections through the transmission network.
Since the disturbance is large, linearized analysis of the swing equation (which
describes the rotor dynamics) is not possible. Further, the fault may cause structural
changes in the network, because of which the power angle curve prior to fault, during the
fault and post fault may be different. Due to these reasons, a general stability criteria for
transient stability cannot be established, as was done in the case of steady state stability
(namely PS > 0). Stability can be established, for a given fault, by actual solution of the
swing equation. The time taken for the fault to be cleared (by the circuit breakers) is
called the clearing time. If the fault is cleared fast enough, the probability of the system
remaining stable after the clearance is more. If the fault persists for a longer time,
likelihood of instability is increased. Critical clearing time is the maximum time
available for clearing the fault, before the system loses stability. Modern circuit breakers
are equipped with auto reclosure facility, wherein the breaker automatically recloses after
two sequential openings. If the fault still persists, the breakers open permanently. Since
most faults are transient, the first reclosure is in general successful. Hence, transient
stability has been greatly enhanced by auto closure breakers.
Some common assumptions made during transient stability studies are as follows:
1. Transmission line and synchronous machine resistances are neglected. Since
resistance introduces a damping term in the swing equation, this gives
pessimistic results.
 14

2. Effect of damper windings is neglected which again gives pessimistic results.

3. Variations in rotor speed are neglected.
4. Mechanical input to the generator is assumed constant. The governor control
loop is neglected. This also leads to pessimistic results.
5. The generator is modeled as a constant voltage source behind a transient
reactance, neglecting the voltage regulator action.
6. Loads are modeled as constant admittances and absorbed into the bus
admittance matrix.
The above assumptions, vastly simplify the equations. A digital computer program for
transient stability analysis can easily include more detailed generator models and effect of
controls, the discussion of which is beyond the scope of present treatment. Studies on the
transient stability of an SMIB system, can shed light on some important aspects of
stability of larger systems. The figure below shows an example of how the clearing
time has an effect on the swing curve of the machine.
 15

Modified Euler’s method:

Euler’s method is one of the easiest methods to program for solution of differential
equations using a digital computer . It uses the Taylor’s series expansion, discarding all
second–order and higher–order terms. Modified Euler’s algorithm uses the derivatives at
the beginning of a time step, to predict the values of the dependent variables at the end of
the step (t1 = t0 + Δt). Using the predicted values, the derivatives at the end of the interval
are computed. The average of the two derivatives is used in updating the variables.
Consider two simultaneous differential equations:
dx
 f x ( x, y , t )
dt
dy
 f y ( x, y , t )
dt
 16

Starting from initial values x0, y0, t0 at the beginning of a time step and a step size h we
solve as follows:
Let
dx
Dx = fx(x0,y0,t0) =
dt 0

dy
Dy = fy(x0,y0,t0) =
dt 0

x P  x 0  D x h 
 Predicted values
y P  y 0  D y h 

dx
DxP = = fx(xP,yP,t1)
dt P

dy
DyP = = fy(xP,yP,t1)
dt P

 D  D xP 
x1 = x o +  x h
 2 
 D y  D yP 
y1 = yo +   h
 2 
x 1 and y1 are used in the next iteration. To solve the swing equation by Modified Euler’s
method, it is written as two first order differential equations:
d

dt
d Pa P  Pmax sin 
  m
dt M M
Starting from an initial value δo, ωo at the beginning of any time step, and choosing a step
size Δt s, the equations to be solved in modified Euler’s are as follows:
d
= D1 = ωo
dt 0

d P  Pmax sin  0
= D2 = m
dt 0 M
δP = δ0 + D1 Δt
ωP = ω0 + D2 Δt
 17

d
= D1P = ωP
dt P

d Pm  Pmax sin  P
= D2P =
dt P M

 D  D1P 
δ1 = δ0 +  1  Δt
 2 
 D  D2 P 
ω1 = ω0 +  2  Δt
 2 
δ1 and ω1 are used as initial values for the successive time step. Numerical errors are
introduced because of discarding higher–order terms in Taylor’s expansion. Errors can be
decreased by choosing smaller values of step size. Too small a step size, will increase
computation, which can lead to large errors due to rounding off. The Runge- Kutta
method which uses higher–order terms is more popular.

Example :A 50 Hz, synchronous generator having inertia constant H = 5.2 MJ/MVA and

x d = 0.3 pu is connected to an infinite bus through a double circuit line as shown in
Fig. 9.21. The reactance of the connecting HT transformer is 0.2 pu and reactance of each
line is 0.4 pu. E g = 1.2 pu and V = 1.0 pu and Pe = 0.8 pu. Obtain the swing curve

using modified Eulers method for a three phase fault occurs at the middle of one of the
transmission lines and is cleared by isolating the faulted line.
 18

Solution:
Before fault transfer reactance between generator and infinite bus
0 .4
XI = 0.3 + 0.2 + = 0.7 pu
2
1 .2  1 .0
Pmax I = = 1.714 pu.
0 .7
Initial Pe = 0.8 pu = Pm
0 .8
Initial operating angle δ o = sin-1 = 27.82o = 0.485 rad.
1.714
When fault occurs at middle of one of the transmission lines, the network and its
reduction is as shown in Fig a to Fig c.
 19

The transfer reactance is 1.9 pu.

1 .2  1 .0
Pmax II = = 0.63 pu
1 .9
Since there is no outage, Pmax III = Pmax I = 1.714
 P 
δmax =   sin 1  m     sin 1  0.8  = 2.656 rad
 Pmax III   1.714 
 
Pm  max   o   Pmax II cos  o  Pmax III cos  max
cos δcr =
Pmax III  Pmax II

0.8 2.656  0.485  0.63 cos 0.485  1.714 cos 2.656 

=
1.714  0.63

1.7368  0.5573  1.5158

= = – 0.3102
1.084
δcr = cos-1 (– 0.3102) = 1.886 rad = 108.07o
with line outage
XIII = 0.3 + 0.2 + 0.4 = 0.9 pu
1 .2  1 .0
Pmax III = = 1.333 pu
0 .9
0 .8
δmax =   sin 1 = 2.498 rad
1.333
 20

Modified Eulers method

δo = 27.82o = 0.485 rad
ωo = 0.0 rad / sec ( at t = 00)

Choosing a step size of 0.05 s, the swing is computed. Table a gives the values of the
derivatives and predicted values. Table b gives the initial values δo, ωo and the values at
the end of the interval δ1, ω1. Calculations are illustrated for the time step t = 0.2 s.
δo = 0.761
ωo = 2.072
Pm = 0.8
 5.2 
M=   = 0.0331 s2 / rad
   50 
Pmax (after fault clearance) = 1.333 pu
D1 = 2.072
0.8  1.333 sin ( 0.761)
D2 = = – 3.604
0.0331
δP = 0.761 + ( 2.072  0.05) = 0.865
ωP = 2.072 + (– 3.604  0.05) = 1.892
D1P = 1.892
0.8  1.333 sin (0.865)
D2P = = – 6.482
0.0331
 2.072  1.892 
δ1 = 0.761 +   0.05 = 0.860
 2 
  3.604  6.482 
ω1 = 2.072 +   0.05 = 1.82
 2 
δ1, ω1 are used as initial values in next time step.

Table a : Calculation of derivatives in modified Euler’s method

 21

t D1 D2 δP ωP D1P D2P
0+ 0.0 15.296 0.485 0.765 0.765 15.296
0.05 0.765 14.977 0.542 1.514 1.514 14.350
0.10 1.498 14.043 0.636 2.200 2.200 12.860
0.15 2.17 - 0.299 0.761 2.155 2.155 - 3.600
0.20 2.072 - 3.604 0.865 1.892 1.892 - 6.482
0.25 1.820 - 6.350 0.951 1.502 1.502 - 8.612
0.30 1.446 - 8.424 1.015 1.025 1.025 - 10.041
0.35 0.984 - 9.827 1.054 0.493 0.493 - 10.843
0.40 0.467 - 10.602 1.065 - 0.063 - 0.063 - 11.060
0.45 - 0.074 - 10.803 1.048 - 0.614 - 0.614 - 10.720
0.50 - 0.612 - 10.46 1.004 - 1.135 - 1.135 - 9.800

Table b : calculations of δo, ωo and δ1, ω1 in modified Euler’s method

δo ωo δ1 ω1 δ1
T Pmax(pu)
rad rad / sec rad rad / sec deg
0- 1.714 0.485 0.0 – – -
0+ 0.630 0.485 0.0 0.504 0.765 28.87
0.05 0.630 0.504 0.765 0.561 1.498 32.14
0.10 0.630 0.561 1.498 0.653 2.170 37.41
0.15 1.333 0.653 2.170 0.761 2.072 43.60
0.20 1.333 0.761 2.072 0.860 1.820 49.27
0.25 1.333 0.860 1.820 0.943 1.446 54.03
0.30 1.333 0.943 1.446 1.005 0.984 57.58
0.35 1.333 1.005 0.984 1.042 0.467 59.70
0.40 1.333 1.042 0.467 1.052 - 0.074 60.27
0.45 1.333 1.052 - 0.074 1.035 - 0.612 59.30
0.50 1.333 1.035 - 0.612 0.991 - 1.118 56.78
 22

Runge - Kutta method

In Range - Kutta method, the changes in dependent variables are calculated from
a given set of formulae, derived by using an approximation, to replace a truncated
Taylor’s series expansion. The formulae for the Runge - Kutta fourth order
approximation, for solution of two simultaneous differential equations are given below;
dx
Given = fx (x, y, t)
dt
dy
= fy (x, y, t)
dt
Starting from initial values x0, y0, t0 and step size h, the updated values are
1
x 1 = x0 + (k1 + 2k2 + 2k3 + k4)
6
1
y1 = y0 + (l1 + 2l2 + 2l3 + l4)
6
where k1 = fx (x0, y0 ,t0) h
 k l h
k2 = fx  x0  1 , y 0  1 , t 0   h
 2 2 2

 k l h
k3 = fx  x0  2 , y 0  2 , t 0   h
 2 2 2
k4 = fx (x0 + k3, y0 + l3, t0 + h) h
l1 = fy (x0, y0, t0) h
 k l h
l2 = fy  x0  1 , y 0  1 , t 0   h
 2 2 2

 k l h
l3 = fy  x0  2 , y 0  2 , t 0   h
 2 2 2
l4 = fy (x0 + k3, y0 + l3, t0 + h) h
The two first order differential equations to be solved to obtain solution for the swing
equation are:
d
=ω
dt
 23

d P P  Pmax sin 
 a  m
dt M M
Starting from initial value δ0, ω0, t0 and a step size of Δt the formulae are as follows
k1 = ω0 Δt
 P  Pmax sin  0 
l1 =  m  Δt
 M 

 l 
k2 =   0  1  Δt
 2

  k1  
 Pm  Pmax sin   0  2  
l2 =     Δt
 M 
 
 
 l 
k3 =   0  2  Δt
 2
  k2  
 Pm  Pmax sin   0  2  
l3 =     Δt
 M 
 
 
k4 = (ω0 + l3) Δt
 P  Pmax sin  0  k 3  
l4 =  m  Δt
 M 
1
δ1 = δ0 + [k1 + 2k2 + 2k3 + k4]
6
1
ω1 = ω0 + [l1 + 2l2 + 2l3 + l4]
6

Example
Obtain the swing curve for previous example using Runge - Kutta method.

Solution:
δ0 = 27.820 = 0.485 rad.
ω0 = 0.0 rad / sec. ( at t = 0-)
 24

Choosing a step size of 0.05 s, the coefficient k1, k2, k3, k4 and l1, l2, l3, and l4 are
calculated for each time step. The values of δ and ω are then updated. Table a gives the
coefficient for different time steps. Table b gives the starting values δ0, ω0 for a time step
and the updated values δ1, ω1 obtained by Runge - Kutta method. The updated values are
used as initial values for the next time step and process continued. Calculations are
illustrated for the time step t = 0.2 s.
δ0 = 0.756
M = 0.0331 s2 / rad
ω0 = 2.067
Pm = 0.8
Pmax = 1.333 (after fault is cleared)
k1 = 2.067 × 0.05 = 0.103
 0.8  1.333 sin (0.756) 
l1 =    0.05 = – 0.173
 0.0331 
 0.173 
k2 = 2.067  0.05 = 0.099
 2 

  0.103  
 0.8  1.333 sin  0.756  2  
l2 =      0.05 = – 0. 246
 0.0331 
 
 
 0.246 
k3 = 2.067  0.05 = 0.097
 2 

  0.099  
 0.8  1.333 sin  0.756  2  
l3 =      0.05 = – 0. 244
 0 . 0331 
 
 
k4 = (2.067 – 0.244) 0.05 = 0.091
 0.8  1.333 sin 0.756  0.097  
l4 =    0.05 = – 0. 308
 0.0331 
1
δ1 = 0.756 + [0.103 + 2 × 0.099 + 2 × 0.097 + 0.091] = 0.854
6
 25

1
ω1 = 2.067 + [– 0.173 + 2 × – 0. 246 + 2 × – 0. 244 – 0. 308] = 1.823
6
Now δ = 0.854 and ω = 1.823 are used as initial values for the next time step. The
computations have been rounded off to three digits. Greater accuracy is obtained by
reducing the step size.

Table a : Coefficients in Runge - Kutta method

T k1 l1 k2 l2 k3 l3 K4 l4
0.0 0.0 0.764 0.019 0.764 0.019 0.757 0.038 0.749
0.05 0.031 0.749 0.056 0.736 0.056 0.736 0.075 0.703
0.10 0.075 0.704 0.092 0.674 0.091 0.667 0.108 0.632
0.15 0.108 – 0.010 0.108 – 0.094 0.106 – 0.095 0.103 – 0.173
0.20 0.103 – 0.173 0.099 – 0.246 0.097 – 0.244 0.091 – 0.308
8.25 0.091 – 0.309 0.083 – 0.368 0.082 – 0.363 0.073 – 0.413
0.30 0.073 – 0.413 0.063 – 0.455 0.061 – 0.450 0.050 – 0.480
0.35 0.050 – 0.483 0.038 – 0.510 0.037 – 0.504 0.025 – 0.523
0.40 0.025 – 0.523 0.012 – 0.536 0.011 – 0.529 – 0.001 – 0.534
0.45 – 0. 001 – 0.534 – 0.015 – 0.533 – 0.015 – 0.526 – 0.027 – 0.519
0.50 – 0.028 – 0.519 – 0.040 – 0.504 – 0.040 – 0.498 – 0.053 – 0.476

Table b: δ, ω computations by Runge - Kutta method

t Pmax δ0 ω0 δ1 ω1 δ1
(sec) (pu) (rad) rad/sec rad rad/sec deg
0- 1.714 0.485 0.0
0+ 0.630 0.485 0.0 0.504 0.759 28.87
0.05 0.630 0.504 0.756 0.559 1.492 32.03
0.10 0.630 0.559 1.492 0.650 2.161 37.24
0.15 1.333 0.650 2.161 0.756 2.067 43.32
0.20 1.333 0.756 2.067 0.854 1.823 48.93
 26

0.25 1.333 0.854 1.823 0.936 1.459 53.63

0.30 1.333 0.936 1.459 0.998 1.008 57.18
0.35 1.333 0.998 1.008 1.035 0.502 59.30
0.40 1.333 1.035 0.502 1.046 – 0.029 59.93
0.45 1.333 1.046 – 0.029 1.031 – 0.557 59.07
0.50 1.333 1.031 – 0.557 0.990 – 1.057 56.72

Note: δ0, ω0 indicate values at beginning of interval and δ 1, ω1 at end of interval. The
fault is cleared at 0.125 seconds.  Pmax = 0.63 at t = 0.1 sec and Pmax = 1.333 at t = 0.15
sec, since fault is already cleared at that time. The swing curves obtained from modified
Euler’s method and Runge - Kutta method are shown in Fig. It can be seen that the two
methods yield very close results.

Fig: : Swing curves with Modified Euler’ and Runge-Kutta methods

 27

Milne’s Predictor Corrector method:

The Milne’s formulae for solving two simultaneous differential equations are
given below.
dx
Consider = fx (x, y, t)
dt
dy
= fy (x, y, t)
dt
With values of x and y known for four consecutive previous times, the predicted value for
n + 1th time step is given by

x nP 1  x n  3 
4h
3

2 x n  2  x n 1  2 x n 
y nP 1  y n  3 
4h
3

2 y n  2  y n 1  2 y n 
Where x  and y  are derivatives at the corresponding time step. The corrected values are

xn+1 = x n 1 
h
3

x n 1  4 x n  x n  1 
yn+1 = y n 1 
h
3

y n 1  4 y n  y n  1 

where xn 1  f x xnP 1 , y nP 1 , t n  1 
y n 1  f y x P
n 1 , y nP 1 , t n  1 
To start the computations we need four initial values which may be obtained by
modified Euler’s method, Runge - Kutta method or any other numerical method which is
self starting, before applying Milne’s method. The method is applied to the solution of
swing equation as follows:
d
Define  n   n
dt n

d Pm  Pmax sin  n
 n  
dt n M

 nP 1   n  3 
4 t
3

2  n  2   n 1  2  n 
 28

 nP 1   n  3 
4 t
3

2  n  2   n 1  2  n 
δn+1 = δn-1 +
t
3

 n 1  4  n   n  1 
ωn+1 = ωn-1 +
t
3

 n 1  4  n   n  1 
where  n 1   nP1

Pm  Pmax sin  nP 1

 n  1 
M

Example
Solve example using Milne’s method.

Solution:
To start the process, we take the first four computations from Range Kutta method
t = 0.0 s δ1 = 0.504 ω1 = 0.759
t = 0.05 s δ2 = 0.559 ω2 = 1.492
t = 0.10 s δ3 = 0.650 ω3 = 2.161
t = 0.15 s δ4 = 0.756 ω4 = 2.067
The corresponding derivatives are calculated using the formulae for  n and  n . We get

 1 = 0.759 1 = 14.97

 2 = 1.492  2 = 14.075

 3 = 2.161 3 = 12.65

 4 =2.067  4 = – 3.46
We now compute δ5 and ω5, at the next time step i.e t = 0.2 s.

 5P   1 
4t
3

2  2   3  2  4 
4  0.05
= 0.504 + 2  1.492  2.161  2  2.067 = 0.834
3

 5P  1 
4t
3

2  2  3  2  4 
 29

4  0.05
= 0.759 + 2  14.075  12.65  2  (3.46) = 1.331
3
 5 = 1.331

0.8  1.333 sin (0.834)

5  = – 5.657
0.0331

δ5 = δ3 +
t
3

 3  4  4   5 
= 0.65 +
0.05
2.161  4  2.067  1.331 = 0.846
3

ω5 = ω3 +
t
3

3  4  4  5 
= 2.161 +
0.05
12.65  4  3.46  5.657  = 2.047
3
 5 = ω5 = 2.047

0.8  1.333 sin (0.846)

 5  = – 5.98
0.0331
The computations are continued for the next time step in a similar manner.

MULTI MACHINE TRANSIENT STABILITY ANALYSIS

A typical modern power system consists of a few thousands of nodes with heavy
interconnections. Computation simplification and memory reduction have been two
major issues in the development of mathematical models and algorithms for digital
computation of transient stability. In its simplest form, the problem of a multi machine
power system under going a disturbance can be mathematically stated as follows:

x t   f I ( x(t ))   t  0
x t   f II ( x(t )) 0  t  t ce

x t   f III ( x(t )) t ce  t  

xt  is the vector of state variables to describe the differential equations

governing the generator rotor dynamics, dynamics of flux decay and associated generator
 30

controller dynamics (like excitation control, PSS, governor control etc). The function f I
describes the dynamics prior to the fault. Since the system is assumed to be in steady
state, all the state variable are constant. If the fault occurs at t = 0, f II describes the
dynamics during fault, till the fault is cleared at time t cl. The post–fault dynamics is
governed by fIII. The state of the system xcl at the end of the fault-on period (at t = tcl)
provides the initial condition for the post–fault network described which determines
whether a system is stable or not after the fault is cleared. Some methods are presented in
the following sections to evaluate multi machine transient stability. However, a detailed
exposition is beyond the scope of the present book.

REDUCED ORDER MODEL

This is the simplest model used in stability analysis and requires minimum data.
The following assumptions are made:
 Mechanical power input to each synchronous machine is assumed to be
constant.
 Damping is neglected.
 Synchronous machines are modeled as constant voltage sources behind
transient reactance.
 Loads are represented as constant impedances.

With these assumptions, the multi machine system is represented as in Fig. 9.26.
 31

Fig 9.26 Multi machine system

Nodes 1, 2 …… n are introduced in the model and are called internal nodes (the
terminal node is the external node connected to the transmission network). The swing
equations are formed for the various generators using the following steps:
Step 1: All system data is converted to a common base.
Step 2: A prefault load flow is performed, to determine the prefault steady state voltages,
at all the external buses. Using the prefault voltages, the loads are converted into
equivalent shunt admittance, connected between the respective bus and the reference
node. If the complex load at bus i is given by
S i  PLi  QLi
the equivalent admittance is given by
S L* i PLi  jQLi
YLi = 2
 2
VLi VLi

Step 3: The internal voltages are calculated from the terminal voltages, using
 32

Ei  i  Vi  j x d i I i

S G* i
= Vi  jxdi
Vi

= Vi  j xd' i
P Gi  j QG i 
Vi

 i is the angle of Ei with respect to Vi. If the angle of Vi is βi, then the angle of

Ei, with respect to common reference is given by  i   i   i . PGi and QGi are obtained
from load flow solution.
Step4: The bus admittance matrix Ybus formed to run the load flow is modified to include
the following.
(i) The equivalent shunt load admittance given by, connected between the
respective load bus and the reference node.
(ii) Additional nodes are introduced to represent the generator internal nodes.
Appropriate values of admittances corresponding to x d , connected between
the internal nodes and terminal nodes are used to update the Ybus.
(iii) Ybus corresponding to the faulted network is formed. Generally transient
stability analysis is performed, considering three phase faults, since they are
the most severe. The Ybus during the fault is obtained by setting the elements
of the row and column corresponding to the faulted bus to zero.
(iv) Ybus corresponding to the post–fault network is obtained, taking into account
line outages if any. If the structure of the network does not change, the Ybus
of the post-fault network is same as the prefault network.
Step 5: The admittance form of the network equations is
I = Ybus V
Since loads are all converted into passive admittances, current injections are present only
at the n generator internal nodes. The injections at all other nodes are zero. Therefore, the
current vector I can be partitioned as
I 
I=  n
0
 33

where In is the vector of current injections corresponding to the n generator internal

nodes. Ybus and V are also partitioned appropriately, so that

 I n   Y1 Y2   En 
 0   Y   
   3 Y4   Vt 
where En is the vector of internal emfs of the generators and Vt is the vector of external
bus voltages. From (9.91) we can write
In = Y1 En + Y2 Vt
0 = Y3 En + Y4 Vt
we get
Vt =  Y41Y3 E n

In = Y1  Y2 Y4 1 Y3  En = Y En



where Y  Y1  Y2 Y41 Y3 is called the reduced admittance matrix and has dimension

n n . Y gives the relationship between the injected currents and the internal generator
voltages. It is to be noted we have eliminated all nodes except the n internal nodes.
Step 6: The electric power output of the generators are given by
PGi = R [Ei I i* .]

Substituting for Ii from (9.94) we get

 
n
Gˆ i i   Ei E j Bˆ i j sin ( i   j )  Gˆ i j cos ( i   j )
2
PGi = Ei
j 1  i

(This equation is derived in chapter on load flows)

Step 7: The rotor dynamics representing the swing is now given by
d 2 i
Mi = PMi - PGi i = 1………..n
dt 2
The mechanical power PMi is equal to the pre-fault electrical power output, obtained from
pre-fault load flow solution.
Step 8: The n second order differential equations can be decomposed into 2n first order
differential equations which can be solved by any numerical method .
 34

Though reduced order models, also called classical models, require less
computation and memory, their results are not reliable. Further, the interconnections of
the physical network of the system is lost.
FACTORS AFFECTING TRANSIENT STABILITY:
The relative swing of a machine and the critical clearing time are a measure of the
stability of a generating unit. From the swing equation, it is obvious that the generating
units with smaller H, have larger angular swings at any time interval. The maximum
Eg V
power transfer Pmax = , where V is the terminal voltage of the generators. Therefore
xd

an increase in x d , would reduce Pmax. Hence, to transfer a given power Pe, the angle δ

would increase since Pe = Pmax sin δ, for a machine with larger x d . This would reduce the
critical clearing time, thus, increasing the probability of losing stability.
Generating units of present day have lower values of H, due to advanced cooling
techniques, which have made it possible to increase the rating of the machines without
significant increase in the size. Modern control schemes like generator excitation control,
Turbine valve control, single-pole operation of circuit breakers and fast-acting circuit
breakers with auto re-closure facility have helped in enhancing overall system stability.
Factors which can improve transient stability are
(i) Reduction of transfer reactance by using parallel lines.
(ii) Reducing transmission line reactance by reducing conductor spacing and
increasing conductor diameter, by using hollow cores.
(iii) Use of bundled conductors.
(iv) Series compensation of the transmission lines with series capacitors. This
also increases the steady state stability limit. However it can lead to problem
of sub-synchronous resonance.
(v) Since most faults are transient, fast acting circuit breakers with rapid
re-closure facility can aid stability.
(vi) The most common type of fault being the single-line-to-ground fault,
selective single pole opening and re-closing can improve stability.
(vii) Use of braking resistors at generator buses. During a fault, there is a sudden
decrease in electric power output of generator. A large resistor, connected at
 35

the generator bus, would partially compensate for the load loss and help in
decreasing the acceleration of the generator. The braking resistors are
switched during a fault through circuit breakers and remain for a few cycles
after fault is cleared till system voltage is restored.
(viii) Short circuit current limiters, which can be used to increase transfer
impedance during fault, there by reducing short circuit currents.
(ix) A recent method is fast valving of the turbine where in the mechanical
power is lowered quickly during the fault, and restored once fault is cleared.