Introduction to Power System Analysis

Power system analysis includes

Load Flow Calculations

Power System Economic Analysis
Short Circuit Calculations
Stability Analysis, etc

For steady state analysis, balanced circuit conditions can be assumed. It

is required to represent the power system by a single phase network.
The power network contains very many nodes, lines, etc. They can be
solved through

Nodal Method (involves bus admittance matrix)

Loop method (involves bus impedance matrix)

Node Equations: The bus admittance matrix

Junctions formed by 2 or more elements are called nodes. Major

nodes are those where more than 2 elements are connected. For
power networks, ground is the reference node. The remaining major
nodes are called buses.

Consider the one-line diagram of the following power network.

j0.1
j0.1

1 j0.2 2

j0.25 j0.4

A 3-bus, 2-generator power system

The reactance diagram of the power system is given in the following.

I1 y10   j10 I2 y20   j10

1 2

y12   j 5
y13   j 4 y23   j 2.5

Apply KCL at the 3 nodes (buses) to get,

I 1  y 10V 1  y 12 (V 1 V 2 )  y 13 (V 1 V 3 )
I 2  y 20V 2  y 12 (V 2 V 1 )  y 23 (V 2 V 3 )
0  y 13 (V 3 V 1 )  y 23 (V 3 V 2 )
Rewrite these equations in the form,

 I1  Y11 Y12 Y13  V1 

 I   Y Y Y  V 
 2   21 22 23   2 
 0  Y31 Y32 Y33  V3 
This is expressed as,

I BUS  YBUSVBUS
where,
 Y11  y10  y12  y13
Y12  Y21   y12
Y13  Y31   y13
Y22  y20  y12  y23
Y23  Y32   y23
Y33  y13  y23
Substitute the values and find the bus admittance matrix as,

  j19 j5 j4 
YBUS   j 5  j17.5 j 2.5 
 
 j 4 j 2.5  j 6.5

Observe that the YBUS is symmetric.

Extending the above relation to an n bus system, the node

voltage equation in matrix form is

 I1  Y11 Y12 Y1i Y1n  V1 

 I  Y Y22 Y2 i Y2 n  V2 
 2   21  
    
    
    
    
   
 I i   Yi1 Yi 2 Yii Yin  Vi 
    
    
    
    
    

 In  
Yn1 Yn 2 Yni Ynn  
Vn 
 The general form for the elements of the admittance matrix is
n
Yii   yij j i
j 0

Yij  Yji   yij

When the bus currents are known, the bus voltages can be solved
through,

VBUS  YBUS
1
I BUS
 Z BUS I BUS
ZBUS is the bus impedance matrix, and is inverse of YBUS.

NETWORK REDUCTION (Node Elimination)

In circuit analysis we have seen that some nodes can be

eliminated by Y-Δ (star mesh) conversions. The procedure
would be quite complicated and laborious for large systems.
Following gives a procedure for elimination of nodes. Only
those nodes can be eliminated where there is no source
(voltage or current source).
Suppose we have a 5-bus power system with 3 generator buses
numbered as 1, 2, and 3. The node equations are:

 I1  Y11 Y12 Y13 Y14 Y15  V1 

 I  Y Y22 Y23 Y24 Y25  V2 
 2   21
 I 3   Y31 Y32 Y33 Y34 Y35  V3 
    
 0  Y41 Y42 Y43 Y44 Y45  V4 
 0  Y51 Y52 Y53 Y54 Y55  V5 
In order to eliminate the non-source nodes, we partition the
matrix equation as,

 I A  YA YB  VA 
 0   Y Y  V 
   C D B

where, IA = [I1 I2 I3]’ and VA = [V1 V2 V3]’; VB = [V4 V5]’

From the above, we can write,

I A  YAVA  YBVB
0  YCVA  YDVB
Solving the second equation for VB we get,

VB  YD1YCVA

Substituting in the first equation,

I A  YAVA  YBYD!YCVA
 [YA  YBYD!YC ]VA
 YEQVA
For YBUS , YB  YC'
The above equation can also be written as,
1
VA  YEQ IA
 Z EQ I A

We will show that ZEQ happens to be a partitioned part of

ZBUS. Thus, when ZBUS is available, YEQ can be obtained
directly by inverting ZEQ.
Example

The one line diagram of a power system is given below.

Reactances are in pu on a 100 MVA base.

j0.25
j0.25

1 j0.5 2

j0.2 j0.25

a) Form YBUS
b) Obtain ZBUS from YBUS
c) Obtain YEQ if bus 3 is eliminated
d) Obtain ZEQ from YEQ and compare with ZBUS
 a)

 1 1 1 1 1 
 
 j 0.25 j 0.5 j 0.2  
 j 0.5 j 0.2 
 1 1 1 1 1 
YBUS       
 j 0.5 j 0.25 j 0.5 j 0.2 j 0.25 
 1 1 1 1 
    
 j 0.2 j 0.25 j 0.25 j 0.2 
  j11 j2 j5   11 2 5
  j 2  j10 j 4   ( j )  2 10 4 
 j 5 j 4  j 9   5 4 9 

b)

74 38 58  0.1652 0.0848 0.1295

j  
1
Z BUS  YBUS  38 74 54 j 0.0848 0.1652 0.1205
448  
58 54 106 0.1295 0.1205 0.2366
c)

 11 2 1  5   8.222 4.222

YEQ   j 
       5 4    j  4.222 8.222 
 2 10  9  4   
d)

1 j 8.222 4.222  0.1652 0.0848

Z EQ  Y EQ     j 
49.778  4.222 8.222   0.0848 0.1652 

Check with ZA term in the expression for ZBUS in (b).

 General Relationships for Bus Elimination:
Kron Reduction

Regardless of which node has zero current injection, a

system can be Kron reduced without having to
rearrange the equations.

For example, if Ip=0 in the nodal equations of the N bus

system, we may directly calculate the elements of the
new, reduced bus admittance matrix by choosing Ypp as
the pivot and by eliminating bus p using the formula,

Y jpYpk
Y jk (new)  Y jk (original ) 
Ypp

where j and k take on all the integer values from 1 to

N except p since row p and column p are to be
eliminated. The new YBUS is of dimension (N-1)x(N-
1).
 Solution of Nonlinear Equations

We will employ the following 2 methods for solving the

nonlinear power flow equations

 The Gauss-Seidel Method

 The Newton-Raphson Method

The Gauss-Seidel Method of Solution

Consider the following 3 simultaneous equations. They may be

linear or nonlinear.

a11 x1  a12 x2  a13 x3  y1

a21 x1  a22 x2  a23 x3  y2 (1)

a31 x1  a32 x2  a33 x3  y3

To represent a general set of N equations, we may write,

N
(2)
 aik xk  yi i  1,2,......N
k 1
Solving for x1 from the first equation of (1), x2 from the second
equation, and x3 from the third equation gives,

1
x1  ( y1  a12 x2  a13 x3 )
a11
1 (3)
x2  ( y2  a21 x1  a23 x3 )
a22
1
x3  ( y3  a31 x1  a32 x2 )
a33
Or, from (2),

1 N (4)
xk  ( yk   akj x j ) j  k ; k=1,2,.....N
akk j 1

To solve the equations by Gaussian iteration, we proceed as follows:

yk
1. Choose initial values x10 , x20 ,........xn0 ( generally, xk0  )
akk
2. Substitute the initial values xk0 into (4) to obtain the values of xk1
at the end of first iteration

1
xk1  ( yk  ak 1 x10  ak 2 x20  .........akN xN0 ) k=1,2,.......N (5)
akk
3. Repeat step 2 with new values of xki until,

xk  xki  xki1  
where i is the iteration number and ε is the precision index
(tolerance) of the solution. The value of xk at the (i+1)th iteration
is,

1
x 
i 1
k
( ykk  ak 1 x1i  ..........  akN xNi ) (6)
akk
Example

Solve the following simultaneous linear equations within a tolerance of ε

= 0.01 by Gaussian iteration

8 x1  4 x2  24
4 x1  7 x2  2 x3  0
2 x2  8 x3  12
By equation (6),
1
x  (24  4 x2i )
i 1
1
8
1
x2i 1  (4 x1i  2 x3i )
7
1
x3i 1  (12  2 x2i )
8
Let, x10 = 24/8 =3, x20 =0, x30 =12/8=1.5 and continue the iterative
process. The results are tabulated below.
Iteration No. x1 x2 x3 Δxmax
0 3 0 1.5

1 3 1.2857 1.5 1.2857

2 3.6429 1.2857 1.1786 0.6429
3 3.6429 1.7449 1.1786 0.4592
. . . . .
. . . . .
7 3.9545 1.9675 1.0228 0.0586
. . . . .
. . . . .
10 3.9942 1.9884 1.0029 0.0105
11 3.9942 1.9959 1.0029 0.0075
 The Gauss-Seidel Method
To solve the equations by Gauss-Seidel iteration, we follow the Gaussian
iteration procedure except that we substitute the newly calculated value
of a variable immediately in the next calculation rather than waiting for
the beginning of the next iteration. In this case, (6) becomes

1
xki1  ( ykk  ak 1 x1i1  ak 2 x2i1.....  ak ,k 1 xki 1.....  akN xNi )
akk
This method results in a faster convergence of the solution.

Example

Re-work the previous problem with Gauss-Seidel method.

1
x1i 1  (24  4 x2i )
8
1
x2i 1  (4 x1i 1  2 x3i )
7
1
x3i 1  (12  2 x2i 1 )
8
The iterative results are tabulated below.
Iteration No. x1 x2 x3 Δxmax
0 3 0 1.5

1 3 1.2857 1.1786 1.2857

2 3.6429 1.7449 1.0638 0.6429
.

6 3.9942 1.9959 1.0010 0.0105

7 3.9980 1.9985 1.0004 0.0038
 Power Flow Solutions
A power flow study or a load study is the determination of voltage,
current, real and reactive power (or power factor) at different points of
an electric network under existing or contemplated (planned) conditions
of normal operation. Load studies are essential in planning the future
development of the system, because satisfactory operation of the
system depends on knowing the effects of interconnection with other
power networks, of new loads, new generating stations and new
transmission lines before they are installed.

In solving power flow problems, balanced operation is considered. Single

phase model is used.

Four quantities are associated with each node (bus). These are:

Voltage magnitude, V

Phase angle of the voltage, δ

Real power, P

Reactive power, Q

Buses are classified in the following categories:

Generator bus: these are regulated bus; also known as voltage

controlled bus or PV bus. Voltage magnitude and real power are
specified at this bus

Slack bus or swing bus: it is the reference bus (usually the bus of
the largest generator). Voltage magnitude and angle ( V , δ) are
specified. This makes up the difference between scheduled plus
losses with total generation

Load bus: active and reactive powers are specified. These are also
called PQ bus
A 6-bus power system
 Power Flow Solution: Gauss-Seidel Method
Consider bus-i in the n-bus power system

Vi V1
V2

Ii

.
Vn .
.

The injected current at bus Ii can be obtained by KCL as,

I i  yi 0Vi  yi1 (Vi  V1 )  yi 2 (Vi  V2 )  .......  yin (Vi  Vn )

 ( yi 0  yi1  ....  yin )Vi  yi1V1  yi 2V2  ......  yinVn
(1)
 YiiVi  Yi1V1  Yi 2V2  ......  YinVn
n n
 YiiVi   YijV j [or ,  YijV j ]
j 1 j 1
i

Note, yi0 is the admittance of the lines represented by a π-model. The

generator admittance is not included in the above.

The real and reactive power at bus i is,

Vi I i*  Pi  jQi
(2)
P  jQ
Or , I i  i * i
Vi
Equate (1) and (2),

n Pi  jQi
YiiVi   YijV j 
j 1
i
Vi * (3)

1 Pi  jQi n
Vi  [ *
  YijV j ]
Yii Vi j 1
i

Notice that the real and reactive power at bus i are expressed as:

n
Pi  Re al{Vi [YiiVi   YijV j }
*
(4)
j 1
i

n
(5)
Qi   Im ag{Vi [YiiVi   YijV j }
*

j 1
i

In the Gauss-Seidel method, the voltages at the different buses are

solved from (3). If Q is not known for a particular bus, it is obtained
from (5) first and then (3) is solved.

The procedure is continued iteratively until the difference between

the voltage magnitudes of all the buses are less than a pre-
determined value ε. We then say that the solution has converged.
Once the solution converges line flows are calculated and slack
generation is determined.

Convergence can be accelerated by using an acceleration factor as

Vik 1  Vi k   [Vi k 1  Vi k ] (6)

where, Viα is the accelerated value and will be used instead of Vik+1
in the next calculation. α is in the range of 1.3-1.6.
Recollect that:

Known: V1  V100 Swing bus

Vi , Pi i  2,3,.....G; G is the number of generators
Pi , Qi i  G  1, G  2,......n
Unknown: P1 , Q1 swing bus
Qi ,  i i=1,3,.......G
Vi ,  i i=G+1, G+2,.......n

For the 6-bus power system is shown above, reference bus is assigned
as bus #1, followed by generator bus (2,3,….G), then the load buses
(G+1,G+2,……,n)