Chapter 1

Power System Analysis / Power

Flow

1.1 Lecture

11
 1. Power flow calculation

1.1 WHAT IS POWER FLOW CALCULATION?

1.2 WHY DO WE NEED POWER FLOW CALCULATION?

G1 G3
1 Bus (node)
11
Line
Area 1 Area 2
10 Generator
3 101 13 120 110
Transformer
Load
20

2 4 14 12
G2 G4

1.3 THE YBUS MATRIX

Example 1

In the system shown below, determine the relationship between currents and voltages

Yd

Yb Yc
I1 Ya I3
1.3 THE YBUS MATRIX
Example 1 Yd
Currents at node 1:
I1 U 1Y a U 1
U 2
Yb U 1
U 3
Yd Yb Yc
I1 Ya I3
Currents at node 2:
0 U 2
U 1 Yb U 2
U 3
Yc

Currents at node 3:
I3 U 3
U 1
Yd U 3
U 2
Yc

I1 Ya Yb Yc Yb Yd U 1

In matrix form: 0 Yb Yb Yc Yc U 2

I3 Yd Yc Yc Yd U 3

Yd
1.3 THE YBUS MATRIX
Example 1
Yb Yc
I1 Ya I3

I1 Ya Yb Y cd Yb Yd U 1

0 Yb Yb Yc Yc U 2

I3 Yd Yc Yc Yd U 3

Generalizing for any system, the network admittance matrix, Ybus, relates the
nodal voltages vector, U, to the nodal currents vector, I, such that:

I N x1
Y bus U N x1
N : to ta l n u m b e r o f n e tw o r k b u s e s ( n o d e s )
N xN

There are two ways to build Ybus:

a) Manual Inspection.
b) Through the incidence Matrix.
 1.3 THE YBUS MATRIX
Manual Inspection
(Easy to apply in small networks)
Rule:
Diagonal Elements: Sum of all the admittances connected to node i.
Off-diagonal Elements: Negative of the admittance between node i and j.

Example 2 G1 YL1 G2

I1 Y L1 Y L1 0 0 U1
Y L1 Y L1 YL2 YL4 YL4 YL2 YL4
I2 U 2 YL2
I3 0 YL4 Y L3 YL4 YL3 U 3 P3 jQ3
I4 0 YL2 YL3 YL2 YL3 U 4 G4
1 4 4 4 4 4 4 4 4 4 2 4 4 4 4 4 4 4 4 4 3 YL3
Y bus

1.3 THE YBUS MATRIX

Incidence Matrix: AM
(More convenient for automated generation of Ybus in large networks)

T
Y bus A Yp A Y p : p r im itiv e a d m itta n c e m a tr ix
M M

Y L1 0 L 0
N B : n u m b er of b ra n c h e s
0 Y L1 0 0
Yp =
M M O M O rd e r of Yp: N B
N B

0 0 L Y LN
B
 1.3 THE YBUS MATRIX
Incidence Matrix: AM
(More convenient for automated generation of Ybus in large networks)

O rd e r o f A M
: N B
N
T
Y bus A M
Yp A M N B : n u m b er of b ra n c h e s

N : to ta l n u m b e r o f n e tw o r k b u s e s ( n o d e s )

Elements of AM
Convention for current direction is important, e.g.
m I n 0 If l-th branch is not connected to node m
amn 1 If current in l-th branch is directed away from node m

l-th branch 1 If current in l-th branch is directed towards node m

1.3 THE YBUS MATRIX

Incidence Matrix: AM
(More convenient for automated generation of Ybus in large networks)
Example 3 G1 YL1 G2
T
Y bus A M
Yp A M

YL4
Y L1 0 0 0 1 1 0 0 YL2
P3 jQ3
T 0 YL2 0 0 0 1 0 1
Y bus AM G4
{ 1 2 3 0 0 YL3 0 0 0 1 1 YL3
N N N N B

0 0 0 YL4 0 1 1 0
1 4 4 4 2 4 4 4 3 1 4 44 2 4 4 43
Yp [ N B N B ] A M [N B N ]

N B: num ber of branches

N : t o ta l n u m b e r o f n e t w o r k b u s e s ( n o d e s )
1.4 THE POWER FLOW EQUATIONS
Recalling

The elements (e.g. between nodes i and j) of Ybus:

Y ij Y ij ij
Y ij c o s ij
j s in ij
G ij j B ij

Ii Ui
The voltage of the network nodes (e.g. at node i) :

U i
U i i
U i
cos i
j s in i
i
Si Pi jQi
Therefore, the current injected at node i can be obtained as:
N

Ii Y i1 U 1
Yi2 U 2
. .. Y iN U N
Y in U n
n 1

1.4 THE POWER FLOW FLOW EQUATIONS

Considering
N

Ii Y i1 U 1
Yi2 U 2
. .. Y iN U N
Y in U n
n 1

Ii Ui
The active power injected into node i, is given by
i
*
Si Pi jQ i U i
Ii Si Pi jQi
N N

Pi U iU n Y i n c o s i n in
Qi U iU n Y i j s i n i n in
n 1 n 1

N : to ta l n u m b e r o f n e tw o r k b u s e s ( n o d e s )
 1.5 NETWORK NODE TYPES
N

Pi U iU n Y i n c o s i n in
n 1

For every node i, four quantities are defined:

1. Active Power: Pi Ii Ui
2. Reactive Power: Qi
3. Voltage phasor amplitude: Ui i
4. Voltage phasor angle: i Si Pi jQi
Since
Pi f p U 1 ,K ,U i K ,U N
, 1
,K , i
,K , N
Qi f q U 1 ,K ,U i K ,U N
, 1
,K , i
,K , N

Ui and i are state variables: x = [ U]T

Thus, for every node there are, in principle, two equations:

One for the active power, Pi, and one for the reactive power Qi.

1.5 NETWORK NODE TYPES

However, without a reference node, there would be infinite solutions possible.

Reference = slack node A node in which the voltage phasor amplitude

and angle are known (usually the node with the largest generator.)

Node Number of Known Unkown Number of equations per

Type nodes Variables Variables bus type
Slack bus (reference)
1 U1 and 1= 0 P1 and Q1 0

PV bus (generation)
Pi and Ui i and Qi
(associated to Pi)
PQ bus (load)
Pi and Qi i and Ui
(associated to Pi and Qi)
T
N : to ta l n u m b e r o f n e tw o r k n o d e s T
x 2
,K , N
,U N 2
K ,U N
g
 G1 YL1 G2
1.5 NETWORK NODE TYPES
Example 4
YL4
In the 4 node network: YL2
1. Slack node node 1 P3 jQ3
2. PV node 1 active power equation (P2)
3. PQ node 1 active and 1 reactive power equation (P3,Q3) G4
YL3
4. PV node 1 active power equation (P4).
Node Number of Known Unkown Number of equations
Type nodes Variables Variables per bus type
Slack bus 1 U1 and 1 P1 and Q1 0
P2 and U2 2 and Q2
PV buses 2 2 (related to P2 and P4)
P4 and U4 4 and Q4
PQ bus 4-2-1 P3 and Q3 3 and U3 2 (related to P3 and Q3)
Total of 4 equations, and 4 state variables, x = [ T
2 3 4 U 3]

1.6 NEWTON RAPHSON METHOD

The power flow equations cannot be solved analytically due to the
complexity of the non linear equations.

Newton Raphson is an iterative method:

k k k
1 : J (x ) x h(x )
k 1 k k
2 : x x x

h is a vector which holds the mismatch power flow equations:

h = [ P(x) Q(x)]T

J is called the power flow Jacobian matrix. J = -H, where H is the

Jacobian of the power mismatches, i.e. H x=-h(x)
1.6 NEWTON RAPHSON METHOD
Example 5: 2nd order polynomial

h(x)

1.6 NEWTON RAPHSON METHOD

Example 5: 2nd order polynomial

dh ( x0 )

dx
 1.6 NEWTON RAPHSON METHOD

The mismatch equations are nothing more than the power flow equations,
cleverly re-written, so that they become zero when the method converges:

Pi x Pi Pi x Pi U iU n Y in c o s i n in
n 1
Specified for node i
Iteratively computed for node i
N

Qi x Qi Qi x Qi U iU n Y in s in i n in
n 1

Specified for node i

Iteratively computed for node i

T
T
x 2
,K , N
,U N 2
K ,U N
Updated throughout the iterations
g

1.6 NEWTON RAPHSON METHOD

The Jacobian Matrix J contains the partial derivatives of the equations
of the injected P and Q with respect to the state variables, that is
P2 . .. P2 P2 . .. P2

2 N
U N +2
U N
J 11 g

J 12
M . .. M M . .. M

fp fp PN . .. PN PN . .. PN

J 11 J 12 2 N
U N g
+2
U N
J J
J 21 J 22
fq fq QN
g
+2
. .. QN
g
+2
QN
g
+2
. .. QN
g
+2

2 N
U N +2
U N
g

J 21 M . .. M . .. . .. M J 22
QN . .. QN QN . .. QN

2 N
U N +2
U N
g

J contains four submatrices, each one with a myriad of partial derivatives

1.6 NEWTON RAPHSON METHOD
So the following equation system is solved at every iteration:
k
P2 . .. P2 P2 . .. P2

2 N
U N +2
U N
g

k k
M . .. M M . .. M 2
P2

PN . .. PN PN . .. PN M M
k k
2 N
U N +2
U N PN
g N

k k
QN +2
. .. QN +2
QN +2
. .. QN +2 U QN
g g g g N g +2 g +2

2 N
U N +2
U N M M
g

M . .. M . .. . .. M U
k
QN
k
N

QN . .. QN QN . .. QN
U U
2 N N g
+2 N Corrections Mismatches

Jacobian

1.6 NEWTON RAPHSON METHOD

Example 6: Partial derivatives

Suppose a function with multiple variables:
2
f x, y 1 2x x y 3y

Partial derivative w.r.t. x: regard y terms as constants!

f x, y
2 2 xy
x

Partial derivative w.r.t. y: regard x terms as constants!

f x, y 2
x 3
y
1.6 NEWTON RAPHSON METHOD

Calculation of the terms inside the Jacobian Matrix

J11 block:

N
Pi
Diagonal elements: U iU n Y i n s in in n i
i n 1
n i

Pi
Off-diagonal elements: U iU j Y i j s i n ij j i
j

1.6 NEWTON RAPHSON METHOD

J12 block: N
Pi
Diagonal elements: 2 U iY i i c o s ii
U n Y in c o s in n i
U i n 1
n i

Pi
Off-diagonal elements: U iY i j c o s ij j i
U j

J21 block:
N
Qi
Diagonal elements: U iU n Y i n c o s in n i
i n 1
n i

Qi
Off-diagonal elements: U iU j Y i j c o s ij j i
j
 1.6 NEWTON RAPHSON METHOD

J22 block:
N
Qi
Diagonal elements: 2 U i Y i i s in ii
U n Y in s i n in n i
U i n 1
n i

Qi
Off-diagonal elements: U iY i j s i n ij j i
U j

1.7 ITERATIVE POWER FLOW CALCULATION

Ybus
T
k k k
k 0, x
T
k k k
h (x ) Pi x Q i
x

k
m ax( h(x ) ) ? s to p .
k
m ax( h(x ) ) ? g o to 5 .

k k
J J (x )
T
k k k k 1 k
x ( ) ( )

k 1 k k
k 1 k k
N e w i t e r a t io n k 1: x x x
k 1 k k
U U U

i s a p r e - s p e c if i e d t o l e r a n c e e .g . 0 .0 0 0 1
1.8 DECOUPLED POWER FLOW (DPF)

For real electrical systems, inverting the Jacobian matrix at every iteration of the
power flow calculation process increases the computational burden.

Solution: use the Jacobian matrix calculated at step k = 0, in all steps of the power flow
calculation. Approach usually termed as dishonest Newton method
k 0 0
J J J (x )

Advantage: Lower computational burden.

Disadvantage: More iterations are needed to reach the final results (longer convergence).

1.8 DECOUPLED POWER FLOW (DPF)

Simplifications were introduced by Stott and Alsac in 1974:

1. Resistance of the overhead lines much smaller than the reactance (X >> R)
1 1 j
2
Y ij j Y ij e
jX ij
X ij

2. Small differences between the voltage angles i.e. ( i j) 0

s in ( i j
) i j

cos( i j
) 1
1.8 DECOUPLED POWER FLOW (DPF)
Consequences:
J12 and J21 = 0

Decoupling of the active power and the reactive power channels:

k k k
J (x ) x h (x )
k
As long as the mismatch
0
J 11 0
k Pi x equations are evaluated in the
0 k k
exact manner, i.e. without the
0 J U Q x
22 i simplifications, the DPF method
1
will provide accurate solutions,
k 0
11 i
k
but it will just need more
iterations to converge.
1
k 0 k
U J 22
Q i
x

1.8 DECOUPLED POWER FLOW (DPF)

So the following equation system is solved at every iteration:
P2
B 22 L B2N 2
U 2
J 11 B
M L M M M
B and B are taken
from the imaginary BN2 L BNN N
PN
part of Ybus. J 22
B U N

Row and column

Q
corresponding to the N g 2

slack bus is omitted BN 2 ,N 2

L BN 2 ,N
U N 2
U N 2
g g g g g

M L M M M

B N ,N 2
L BNN U N
Q N
g

U N
 1.8 DECOUPLED POWER FLOW (DPF)
Ybus
T
k k k
x

k k k
Pi U i
, i
U i

1
k 0 k k
11 i
,

k 1 k k

k k 1 k
Qi U i
, i
U i

1
k 0 k k 1
U J 22 Q i
U ,
k 1 k k
U U U
k
m ax( h (x ) ) ? s to p .
k
m ax( h(x ) ) ? g o to 3 .

i s a p r e - s p e c if i e d t o l e r a n c e e .g . 0 .0 0 0 1

1.9

Based on the same hypothesis of the DPF, the non linear equations are
linearized to solve only the active power channel.

Further speeds up the computation: only 1 matrix inversion needed.

Not an accurate solution.

N
Original power
Pi x Pi Pi x Pi U iU n Y in c o s i n in
0 flow equations
n 1

j
U i
U j
1 p u , Y ij Y ij e 2
, i j
0 DC power flow hypotheses
1.9
Resulting DCPF equation:
N N

Pi U iU n Y i n c o s i n in
Y in s i n n i
n 1 n 1
n 1
N

Pi Y ij i n
n 1
n 1 Row and column corresponding
to the slack bus deleted
In Matrix form, it is possible to demonstrate that:

P2 B 22 . .. BN 2
P B
1
M M . .. M M

PN B N2 . .. B NN N

1.10 SUITABILITY OF STUDIED POWER FLOW CALCULATION METHODS

Application Gauss-Seidel Newton-Raphson DPF DCPF

Small networks X X
Large networks X X
N-1 analysis X X X
Market studies X