Lecture 3:

Power Flow Analysis

Acknowledgement: The slides are developed based on parts in

Power System Analysis, Hadi Saadat, McGraw-Hill, 2002

GEE/PSAOpt/L3-Power Flow Analysis 1

 The Bus Admittance Matrix - Ybus Matrix
l The matrix equation for relating the nodal voltages to the
currents that flow into and out of a network using the
admittance values of circuit branches
Ik Vk
Iinj = Ybus ⋅ Vnode
Network

l Used to form the network model of an interconnected

power system
u Nodes represent substation bus bars
u Branches represent transmission lines and transformers
u Injected currents are the flows from generator and loads

GEE/PSAOpt/L3-Power Flow Analysis 2

 The Bus Admittance Matrix - Ybus Matrix
l Constructing the Bus Admittance Matrix (or the Y bus
matrix)
u form the nodal solution based upon Kirchhoff’s current law

I k −inj = yk 0 Vk + yk 1 (Vk − V1 ) + yk 2 (Vk − V2 ) + K + ykn (Vk − Vn )

u impedances are converted to admittances
1 1
yij = =
zij rij + j xij

GEE/PSAOpt/L3-Power Flow Analysis 3

 Example 1: Matrix Formation
generator 1 generator 2
z = j1.0 z = j0.8 V1 V2

line 12 j1.0 j0.8

j0.4
1 z = j0.4 2 1 2
line 13 line 23 j0.2 j0.2
z = j0.2 z = j0.2
3
3
line 34 j0.08
z = j0.08
4 4
Network Diagram Impedance Diagram

GEE/PSAOpt/L3-Power Flow Analysis 4

 Example 1: Matrix Formation

I1 y10= -j1.0 I2
y20= -j1.25

1 2
y12 = -j2.5
y13= -j5 y23= -j5
3 KCL Equations

y34 = -j12.5 I1 = y10V1 + y12 (V1 − V2 ) + y13 (V1 − V3 )

4 I 2 = y20V2 + y21 (V2 − V1 ) + y23 (V2 − V3 )
0 = y31 (V3 − V1 ) + y32 (V3 − V2 ) + y34 (V3 − V4 )
Admittance Diagram
0 = y43 (V4 − V3 )

GEE/PSAOpt/L3-Power Flow Analysis 5

 Example 1: Matrix Formation
Rearranging the KCL Equations
I1 = ( y10 + y12 + y13 )V1 − y12V2 − y13V3
I 2 = − y21V1 + ( y20 + y21 + y23 )V2 − y23V3
0 = − y31V1 − y32V2 + ( y31 + y32 + y34 )V3 − y34V4
0 = − y43V3 + y43V4
Matrix Formation of the Equations
 I1  ( y10 + y12 + y13 )=Y11 − y12 =Y12 − y13=Y13 0=Y14 V1 
I   − y =Y ( y + y + y )=Y − y =Y 0 =Y  V 
24
21 21 22 23 23
 2 =  20 21 23  ⋅  2
0  − y31=Y31 − y32=Y32 ( y31 + y32 + y34)=Y33 −y34=Y34 V3 
     
  
0 0 =Y 41 0 =Y42 − y =Y
43 43 y =Y
43  V4 
44

GEE/PSAOpt/L3-Power Flow Analysis 6

 Example 1: Matrix Formation
Completed Matrix Equation
Y11 = ( y10 + y12 + y13 ) = − j8.50 Y23 = Y32 = − y23 = j5.00
Y12 = Y21 = − y12 = j 2.50 Y33 = ( y31 + y32 + y34 ) = − j 22.50
Y13 = Y31 = − y13 = j5.00 Y34 = Y43 = − y34 = j12.50
Y22 = ( y20 + y21 + y23 ) = − j8.75 Y44 = y34 = − j12.50

 I1  − j8.50 j 2.50 j5.00 0  V1 

 I   j 2.50 − j8.75 j 5 . 00 0  V 
 2 =   ⋅  2
 0   j5.00 j5.00 − j 22.50 j12.50  V3 
     
  
0 0 0 j12 . 50 − j12 . 50  V4 

GEE/PSAOpt/L3-Power Flow Analysis 7

 Y-Bus Matrix Building Rules
l Square matrix with dimensions equal to the number of
buses
l Convert all network impedances into admittances
l Diagonal elements: n
Yii = ∑ yij j≠i
j =0

l Off-diagonal elements:
Yij = Y ji = − yij
l Matrix is symmetrical along the leading diagonal

GEE/PSAOpt/L3-Power Flow Analysis 8

 Example 2: Ybus Matrix

System Data Questions:

Line Start End X value 1) Draw the network diagram
2) Draw the impedance diagram
g1 1 0 1.00 3) Draw the admittance diagram
g2 5 0 1.25 4) Determine Ybus matrix
L1 1 2 0.40
L2 1 3 0.50
L3 2 3 0.25
L4 2 5 0.20
L5 3 4 0.125
L6 4 5 0.50

GEE/PSAOpt/L3-Power Flow Analysis 9

 Tap-Changing Transformers
l The tap-changing transform gives some control of the
power network by changing the voltages and current
magnitudes and angles by small amounts
u The flow of real power along a network branch is controlled by
the angular difference of the terminal voltages
u The flow of reactive power along a network branch is controlled
by the magnitude difference of the terminal voltages
u Real and reactive powers can be adjusted by voltage-regulating
transformers and by phase-shifting transformers

bus i 1:a bus j

a can be a
complex number

GEE/PSAOpt/L3-Power Flow Analysis 10

 Modeling of Tap-Changers
u the off-nominal tap ratio is given as 1:a
u the nominal turns-ratio (N1/N2) was addressed with the
conversion of the network to per unit
u the transformer is modeled as two elements joined together at a
fictitious bus x
Ij Vj
Vi yt Vx
Ii
1:a

u basic circuit equations:

Vx = 1a V j I i = −a* ⋅ I j I i = yt (Vi − Vx )

GEE/PSAOpt/L3-Power Flow Analysis 11

 Modeling of Tap-Changers
l Making substitutions

Vx = 1a V j I i = yt (Vi − Vx )
I i = yt (Vi − 1a V j )

I i = −a* ⋅ I j
I j = − a1* I i

I j = − * (Vi − a V j ) = − * Vi + 2 V j
yt 1 yt yt
a a a

GEE/PSAOpt/L3-Power Flow Analysis 12

 YBus Formation of Tap-Changers
l Matrix formation

 yt 
I i = {yt }Vi + − V j
 a
 yt   y 
I j = − * Vi +  t2 V j
 a   a 

 I i   yt − yt a  Vi 
 I  = − y a * 2⋅
yt a  V j 
 j  t

GEE/PSAOpt/L3-Power Flow Analysis 13

 Pi-Circuit Model of Tap-Changers
l Valid for real values of a
l Taking the y-bus formation, break the diagonal elements
into two components
u the off-diagonal element represent the impedance across the two
buses
u the remainder form the shunt element

yt / a tap side
non-tap side

i j
(a - 1) yt / a (1 - a) yt / a2

GEE/PSAOpt/L3-Power Flow Analysis 14

 The Power Flow Solution
l Most common and important tool in power system
analysis
u also known as the “Load Flow” solution
u used for planning and controlling a system
u assumptions: balanced condition and single phase analysis
l Problem:
u determine the voltage magnitude and phase angle at each bus
u determine the active and reactive power flow in each line
u each bus has four state variables:
n voltage magnitude
n voltage phase angle
n real power injection
n reactive power injection

GEE/PSAOpt/L3-Power Flow Analysis 15

 The Power Flow Solution
u Each bus has two of the four state variables defined or given
l Types of buses:
u Slack bus (swing bus)
n voltage magnitude and angle are specified, reference bus
n solution: active and reactive power injections
u Regulated bus (generator bus, P-V bus)
n models generation-station buses
n real power and voltage magnitude are specified
n solution: reactive power injection and voltage angle
u Load bus (P-Q bus)
n models load-center buses
n active and reactive powers are specified (negative values for loads)
n solution: voltage magnitude and angle

GEE/PSAOpt/L3-Power Flow Analysis 16

 Power Flow Equations
l KCL
I i = yi 0Vi + yi1 (Vi − V1 ) + yi 2 (Vi − V2 ) + L + yin (Vi − Vn )
= ( yi 0 + yi1 + yi 2 + L + yin )Vi − yi1V1 − yi 2V2 − L − yinVn
n n
= Vi ∑ yij − ∑ yijV j j≠i
j =0 j =1

l Power Law
Pi − jQi
Pi + jQi = V I *
i i → Ii =
Vi*
Pi − jQi n n

*
= Vi ∑ yij − ∑ yijV j j≠i
Vi j =0 j =1

GEE/PSAOpt/L3-Power Flow Analysis 17

 Line Flows and Losses
l After solving for bus voltages and angles, power flows
and losses on the network branches are calculated
u Transmission lines and transformers are network branches
u The direction of positive current flow are defined as follows for a
branch element (demonstrated on a medium length line)
u Power flow is defined for each end of the branch
n Example: the power leaving bus i and flowing to bus j
Bus i Bus j
Iij Vi IL yij Vj Iji
Ii0 Ij0
yi0 yj0

GEE/PSAOpt/L3-Power Flow Analysis 18

 Line Flows and Losses
lcurrent and power flows:
i→ j j→ i
I ij = I L + I i 0 = yij (Vi − V j )+ yi 0 Vi I ji = − I L + I j 0 = yij (V j − Vi )+ y j 0 V j
Sij = Vi I ij* = Vi 2 ( yij + yi 0 ) − Vi yij* V j* S ji = V j I *ji = V j2 ( yij + y j 0 ) − V j yij* Vi *
* *

l power loss: S Loss ij = Sij + S ji

Bus i Bus j
Iij Vi IL yij Vj Iji
Ii0 Ij0
yi0 yj0

GEE/PSAOpt/L3-Power Flow Analysis 19

 Gauss-Seidel Method
l A non-linear algebraic equation solver
u method of successive displacements
u iterative steps:
n take a function and rearrange it into the form x = g(x)
{there are several possible arrangements}
n make an an initial estimate of the variable x: x[0] = initial value
n find an iterative improvement of x[k], that is: x[k+1] = g( x[k] )
n a solution is reached when the difference between two iterations is
less than a specified accuracy: x[k+1] - x[k]  ≤ ε
u acceleration factors
n can improve the rate of convergence: α > 1
n modified step: the improvement is found as
x[k+1] = x[k] + α( g( x[k] ) - x[k] )

GEE/PSAOpt/L3-Power Flow Analysis 20

 Example 3: Gauss-Seidel
l Find the root of the equation: f(x) = x3 - 6x2 + 9x - 4 = 0
u Step 1. Cast the equation into the g(x) form.
9 x = − x3 + 6x2 + 4
x = − 19 x 3 + 69 x 2 + 94 = g ( x )

GEE/PSAOpt/L3-Power Flow Analysis 21

 Example 3: Gauss-Seidel
u Step 2. Starting with an initial guess of x[0] = 2, several iterations
are performed.
x[1] = g ( x[ 0 ] = 2) = − 19 (2)3 + 96 (2) 2 + 94 = 2.2222
x[ 2 ] = g ( x[1] = 2.2222) = − 19 (2.2222)3 + 96 (2.2222) 2 + 94 = 2.5173
x[3] = g ( x[ 2 ] = 2.5173) = − 19 (2.5173)3 + 96 (2.5173) 2 + 94 = 2.8966
x[ 4 ] = 3.3376
x[5] = 3.7398
x[ 6 ] = 3.9568
x[ 7 ] = 3.9988
x[8] = 4.0000

GEE/PSAOpt/L3-Power Flow Analysis 22

 Example 3: Gauss-Seidel
Matlab Results
4.5
3 2
4
g(x) =-1/9x +6/9x +4/9

3.5

3 x
Iterations
2.5

2 Initial Value x = g(x)

1.5

0.5 Solution Points

0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
x

GEE/PSAOpt/L3-Power Flow Analysis 23

 Example 4: Gauss-Seidel
l Find the root of the equation: f(x) = x3 - 6x2 + 9x - 4 = 0 with
an acceleration factor of 1.25
u Starting with an initial guess of x[0] = 2.
x[ 0 ] = 2
g (2) = − 19 (2) 3 + 69 (2) 2 + 94 = 2.2222
x[1] = 2 + 1.25 [2.2222 − 2] = 2.2778

g (2.2778) = − 19 (2.2778) 3 + 96 (2.2778) 2 + 94 = 2.5902

x[ 2 ] = 2.2778 + 1.25 [2.5902 − 2.2778] = 2.6683

GEE/PSAOpt/L3-Power Flow Analysis 24

 Example 4: Gauss-Seidel
u Additional iterations
x[3] = 3.0801
x[ 4 ] = 3.1831
x[5] = 3.7238
x[ 6 ] = 4.0084
x[ 7 ] = 3.9978
x[8] = 4.0005

GEE/PSAOpt/L3-Power Flow Analysis 25

 Example 4: Gauss-Seidel
Matlab Results
4.5
3 2
4
g(x) = -1/9x +6/9x +4/9

3.5

3 x
Iterations
2.5

2 Initial Value x = g(x)

1.5

0.5 Solution Points

0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
x
with acceleration factor: 1.25
GEE/PSAOpt/L3-Power Flow Analysis 26
 Gauss-Seidel for a System of Equations
l Consider a system of n equations
f1 ( x1 , x2 , L , xn ) = c1
f 2 ( x1 , x2 , L , xn ) = c2
M M
f n ( x1 , x2 , L , xn ) = cn
l Rearrange each equation for one of the variables
x1 = c1 + g1 ( x1 , x2 , L , xn )
x2 = c2 + g 2 ( x1 , x2 , L , xn )
M M
xn = cn + g n ( x1 , x2 , L , xn )

GEE/PSAOpt/L3-Power Flow Analysis 27

 Gauss-Seidel for a System of Equations
l steps
u assume an approximate solution for the independent variables

(x[ ], x[ ],L, x[ ] )
1
0
2
0
n
0

u find the results in a new approximate solution

(x[ 1
k +1]
, x2[k +1] , L , xn[k +1] )
u in the Gauss-Seidel method, the updated values of the variables
calculated in the preceding equations are used immediately in
the solution of the subsequent equations

GEE/PSAOpt/L3-Power Flow Analysis 28

 The Power Flow Equation
l The equation
Pi − jQi
Pi + jQi = V I *
i i → Ii =
Vi*
Pi − jQi n n

*
= Vi ∑ yij − ∑ yijV j j≠i
Vi j =0 j =1

l The Gauss-Siedel form

Pi − jQi n Pi − jQi n
*
+ ∑ yijV j + ∑ y V
ij j
[k ]

Vi j =1 Vi*[ k ] j =1
Vi = n
j ≠ i ⇒ Vi[ k +1] = n
j≠i
∑y
j =0
ij ∑y
j =0
ij

GEE/PSAOpt/L3-Power Flow Analysis 29

 Power Injections
l Rewriting the power equation to find P and Q

  n n
[k ]  
Pi [k +1]
= ℜVi Vi ∑ yij − ∑ yij V j  
* [k ] [k ]
j≠i
  j =0 j =1  

  n n
[k ]  
Qi = −ℑVi Vi ∑ yij − ∑ yij V j  
[k +1] * [k ] [k ]
j≠i
  j =0 j =1  
u the real and reactive powers are scheduled for the load buses
that is, they remain fixed
u the currents and powers are expressed as going into the bus
n for generation the powers are positive
n for loads the powers are negative
n the scheduled power is the sum of the generation and load powers

GEE/PSAOpt/L3-Power Flow Analysis 30

 Solution by Gauss-Seidel
l The complete set of equations become:

Pi [sch ] − jQi[sch ] n
Vi*[k ]
+ ∑ y ij V j
[k ]

Vi [k +1] =
j =1
n
j≠i
∑y
j =0
ij


  n n
[k ]  
Pi [k +1]
= ℜVi Vi ∑ yij − ∑ yij V j   j ≠ i
* [k ] [k ]
  j =0 j =1  

  n n
[k ]   
Qi = −ℑVi Vi ∑ yij − ∑ yij V j   j ≠ i
[k +1] * [k ] [k ]
  j =0 j =1  

GEE/PSAOpt/L3-Power Flow Analysis 31

 Solution by Gauss-Seidel
l Rewriting the equations in terms of the Y-Bus
Pi [sch ] − jQi[sch ] n

Vi *[k ]
− ∑ Y ij V [k ]
j
[k +1] j =1, j ≠ i
Vi =
Yii
  
 * [k ]  [k ] n
[k ]  
Pi [k +1]
= ℜVi Vi Yii + ∑ Yij V j 
 
 
j = 1
 
  j ≠ i  
  
 n

Qi[k +1] = −ℑVi*[k ] Vi [k ]Yii + ∑ Yij V j[k ]  
 
 
j =1
 
  j ≠ i  

GEE/PSAOpt/L3-Power Flow Analysis 32

 Solution by Gauss-Seidel
l System characteristics
u Since both components (V & δ) are specified for the slack bus,
there are 2(n - 1) equations which must be solved iteratively
u For the load buses, the real and reactive powers are known:
scheduled
n the voltage magnitude and angle must be estimated
n in per unit, the nominal voltage magnitude is 1 pu
n the angles are generally close together, so an initial value of 0
degrees is appropriate

GEE/PSAOpt/L3-Power Flow Analysis 33

 Solution by Gauss-Seidel
u For the generator buses, the real power and voltage magnitude
are known
n the real power is scheduled
n the reactive power is computed based on the estimated voltage
values
n the voltage is computed by Gauss-Seidel, only the imaginary part is
kept
n the complex voltage is found from the magnitude and the iterative
imaginary part

e [ k +1]
i = Vi [ sch ] 2
(
− fi )
[ k +1] 2
Vi = ei + j f i

GEE/PSAOpt/L3-Power Flow Analysis 34

 Example 5: Gauss-Seidel Method
l Using the Gauss-Seidel method, determine the phasor
values of the voltage at the load buses 2 and 3, accurate
to 2 decimal places

V1 = 1.05∠0°
0.02 + j0.04
2.566 pu

1.102 pu
0.0125 + j0.025
1.386 pu
0.01 + j0.03
0.452 pu

GEE/PSAOpt/L3-Power Flow Analysis 35

 Newton-Raphson PF Solution
l Quadratic convergence
u mathematically superior to Guass-Seidel method
l More efficient for large networks
u number of iterations required for solution is independent of
system size
l The Newton-Raphson equations are cast in natural power
system form
u solving for voltage magnitude and angle, given real and reactive
power injections

GEE/PSAOpt/L3-Power Flow Analysis 36

 Newton-Raphson Method
l A method of successive approximation using Taylor’s
expansion
u Consider the function: f(x) = c, where x is unknown

u Let x[0] be an initial estimate, then ∆x[0] is a small deviation from

the correct solution

(
f x[0 ] + ∆x[ 0] = c)
u Expand the left-hand side into a Taylor’s series about x[0] yeilds
 f  [ 0] 2
( ) df  [0 ]
( )
2
d
f x[0] +  ∆x + 1
2  2 
∆x + L =c
dx   dx 

GEE/PSAOpt/L3-Power Flow Analysis 37

 Newton-Raphson Method
u Assuming the error, ∆x[0], is small, the higher-order terms are
neglected, resulting in

( )
f x [0] df  [0 ]
+  ∆x ≈c ⇒ ∆c [0] df  [ 0]
≈ ∆x
dx  dx 
u where
∆c[ 0 ] = c − f x[0 ] ( )
u rearranging the equations
∆ c [0]
∆x[ 0] =
df 

dx  
x[1] = x[0 ] + ∆x[ 0]

GEE/PSAOpt/L3-Power Flow Analysis 38

 Example 6: Newton-Raphson Method
l Find the root of the equation: f(x) = x3 - 6x2 + 9x - 4 = 0
50

40

30

20

10
3 2
f(x) = x -6x +9x-4
0

-10
0 1 2 3 4 5 6
x

GEE/PSAOpt/L3-Power Flow Analysis 39

 Power Flow Equations
l KCL for current injection
n n
I i = ∑ Yij V j = ∑ Yij V j ∠ θij + δj
j =1 j =1
l Real and reactive power injection

Pi − j Qi = Vi* I i
l Substituting for Ii yields:

Pi − j Qi = (Vi ∠ − δ)∑ Yij V j ∠ θij + δj

n

j =1

GEE/PSAOpt/L3-Power Flow Analysis 40

 Power Flow Equations
l Divide into real and reactive parts

Pi = ∑ Vi V j Yij cos(θij − δi + δj )
n

j =1

V j Yij sin (θij − δi + δj )

n
Qi = − ∑V
j =1
i

GEE/PSAOpt/L3-Power Flow Analysis 41

 Newton-Raphson Formation
l Cast power equations into iterative form

( )
n
Pi[ k ] = ∑ Vi[ k ] V j[ k ] Yij cos θij − δi[ k ] + δ[j k ]
j =1

( )
n
Qi[ k ] = − ∑ j =1
Vi[ k ] V j[ k ] Yij sin θij − δi[ k ] + δ[j k ]

l Matrix function formation of the system of equations

Pinjsch 
c =  sch 
δ[ k ] 
=  [k ]  ( ) =
( )
Pinj x[ k ]
( )

[k ] [k ]
x f x [k ]

Qinj   
V   Q
 inj x

GEE/PSAOpt/L3-Power Flow Analysis 42

 Newton-Raphson Formation
l General formation of the equation to find a solution
c = f (xsolution ) x[ 0] = initial estimate of xsolution
l The iterative equation

[ k + 1]
= x[ k ] +
c − f( )x [k ]

( )
x
df x[ k ] 
 dx 
 
l The Jacobian - the first derivative of a set of functions

( )
df x[ k ] a matrix of all combinatorial pairs
dx

GEE/PSAOpt/L3-Power Flow Analysis 43

 The Jacobian Matrix
∂P =J1 ∂P =J2
df (x ) ∆P   ∂δ ∂V ∆δ
⇒ ∆Q  = ∂Q =J3 ∂Q =J4∆ V 
dx    ∂δ
∂V  
 
 ∆P1   ∂P1
∂δ1 L ∂P1
∂δn− 1
∂P1
∂V1
L ∂P1
∂Vn− m
  ∆δ1 
 M    
 M O M M O M M
   
 ∆Pn − 1   ∂δn1− 1
∂P ∂Pn− 1 ∂Pn− 1 ∂Pn− 1 
L ∂δn− 1 L  ∆δn − 1 
 =  ∂Q1 
∂V1 ∂Vn− m
 ∂Q1 ∂Q1 ∂Q1 
 ∆Q1   ∂δ1 L ∂δn− 1 ∂V1
L ∂Vn− m  
∆ V1 
 M   M O M M O M  M 
  ∂Qn− m ∂Qn− m ∂Qn − m ∂Qn − m
 
∆Qn − m 
    ∂δ1
L ∂δn− 1 ∂V1
L  ∆ Vn − m 
∂Vn− m   
GEE/PSAOpt/L3-Power Flow Analysis 44
 Jacobian Terms
l Real power w.r.t. the voltage angle
∂Pi
= ∑ Vi V j Yij sin (θij − δi + δj )
∂δi j ≠ i
∂Pi
= − Vi V j Yij sin (θij − δi + δj ) i ≠ j
∂δj
l Real power w.r.t. the voltage magnitude
∂Pi
= 2 Vi Yii cosθii + ∑V Yij cos(θij − δi + δj )
∂Vi
j
j≠i

∂Pi
= Vi Yij cos(θij − δi + δj ) i ≠ j
∂V j

GEE/PSAOpt/L3-Power Flow Analysis 45

 Jacobian Terms
l Reactive power w.r.t. the voltage angle
∂Qi
= ∑ Vi V j Yij cos(θij − δi + δj )
∂δi j≠i

∂Qi
= − Vi V j Yij cos(θij − δi + δj ) i ≠ j
∂δj
l Reactive power w.r.t. the voltage magnitude
∂Qi
= − 2 Vi Yii sin θii + ∑V Yij sin (θij − δi + δj )
∂Vi
j
j≠i

∂Qi
= − Vi Yij sin (θij − δi + δj ) i ≠ j
∂V j

GEE/PSAOpt/L3-Power Flow Analysis 46

 Iteration process
l Power mismatch or power residuals
u difference in schedule to calculated power

∆Pi [k ]
= Pi sch
− Pi [k ]

∆Qi[ k ] = Qisch − Qi[ k ]

l New estimates for the voltages
[ k + 1]
δ i = δ + ∆δi
[k ]
i
[k ]

Vi[ k + 1] = Vi[ k ] + ∆ Vi[ k ]

GEE/PSAOpt/L3-Power Flow Analysis 47

 Bus Type and the Jacobian Formation
l Slack Bus / Swing Bus
u one generator bus must be selected and defined as the voltage
and angular reference
n The voltage and angle are known for this bus
n The angle is arbitrarily selected as zero degrees

n bus is not included in the Jacobian matrix formation

l Generator Bus
n have known terminal voltage and real (actual) power injection
n the bus voltage angle and reactive power injection are computed

n bus is included in the real power parts of the Jacobian matrix

l Load Bus
n have known real and reactive power injections
n bus is fully included in the Jacobian matrix

GEE/PSAOpt/L3-Power Flow Analysis 48

 Newton-Raphson Steps
1. Set flat start
u For load buses, set voltages equal to the slack bus or 1.0∠0°
u For generator buses, set the angles equal the slack bus or 0°
2. Calculate power mismatch
u For load buses, calculate P and Q injections using the known and
estimated system voltages
u For generator buses, calculate P injections
u Obtain the power mismatches, ∆P and ∆Q
3. Form the Jacobian matrix
u Use the various equations for the partial derivatives w.r.t. the
voltage angles and magnitudes

GEE/PSAOpt/L3-Power Flow Analysis 49

 Newton-Raphson Steps
4. Find the matrix solution (choose a or b)
u a. inverse the Jacobian matrix and multiply by the mismatch
power
u b. perform gaussian elimination on the Jacobian matrix with the b
vector equal to the mismatch power
compute ∆δand ∆V
5. Find new estimates for the voltage magnitude and angle
6. Repeat the process until the mismatch (residuals) are
less than the specified accuracy
∆Pi[ k ] ≤ε
∆Qi[ k ] ≤ε

GEE/PSAOpt/L3-Power Flow Analysis 50

 Example 7: Newton-Raphson Method
l Using the Newton-Raphson PF,
find the power flow solution Slack Bus
l Find the voltages at buses 2 and 3 1 V1 = 1.05∠0°
l Find the slack bus real and reactive power
u 100 MVA base 0.01 + j0.03 0.02 + j0.04
0.0125 + j0.025
y12 = 10 − j 20 pu 2
3
y13 = 10 − j 30 pu
|V3| = 1.04
y23 = 16 − j 32 pu 200 MW 400 MW
400 + j 250 250 MVAR
S =− sch
2 = − 4.0 − j 2.5 pu
100
200
P3sch = = 2.0 pu
100

GEE/PSAOpt/L3-Power Flow Analysis 51

 Example 7: Newton-Raphson Method
 20 − j 50 − 10 + j 20 − 10 + j 30
Ybus = − 10 + j 20 26 − j52 − 16 + j 32
 

− 10 + j 30 − 16 + j 32 26 − j 62 

53.9∠ − 1.90 22.4∠ 2.03 31.6∠1.89 
=  22.4∠ 2.03 58.1∠ − 1.11 35.8∠ 2.03  Note: the angles are in radians
 

 31.6∠1.89 35.8∠ 2.03 67.2∠ − 1.17

P2 = V2 V1 Y21 cos(θ21 − δ2 + δ1 )+ V2 Y22 cos(θ22 )+ V2 V3 Y23 cos(θ23 − δ2 + δ3 )

2

P3 = V3 V1 Y31 cos(θ31 − δ3 + δ1 )+ V3 V2 Y32 cos(θ32 − δ3 + δ2 )+ V3 Y33 cos(θ33 )

2

Q2 = − V2 V1 Y21 sin (θ21 − δ2 + δ1 )− V2 Y22 sin (θ22 )− V2 V3 Y23 sin (θ23 − δ2 + δ3 )

2

GEE/PSAOpt/L3-Power Flow Analysis 52

 Example 7: Newton-Raphson Method
δ2  P2 (δ2 , δ3 ,V2 ) 
   
x = δ3  f (x )= P3 (δ2 , δ3 ,V2 ) 
V2 
  Q2 (δ2 , δ3 ,V2 )
 
 V 1.05 22.3 cos(2.03 − δ )+ V 2 58.1 cos(− 1.11)+ V 1.04 35.8 cos(2.03 − δ + δ) 
 2 21 2 2 2 3

= V3 1.05 31.6 cos(1.89 − δ3 )+ 1.04 V2 35.8 cos(2.03 − δ3 + δ2 )+ 1.04 67.2 cos(− 1.17 )
2

 


− V2 1.05 22 .3 sin (2. 03 − δ 2 )− V 2
2
58 .1 sin (− 1. 11)− V 2 1.04 35 .8 sin (2.03 − δ2 + δ )
3 

∆P2  P2sch  P2 (δ2 , δ3 ,V2 )  − 4.0 P2 (δ2 , δ3 ,V2 ) 
      
∆c = ∆P3  = c − f (x )= P3sch − P (δ , δ , V ) =
  2.0 − P (δ , δ
 3 2 3 2  , V )
  3 2 3 2 
Q2sch  
∆Q2 
    Q2 (δ2 , δ3 ,V2 )  − 2.5
 
Q2 (δ2 , δ3 ,V2 )

GEE/PSAOpt/L3-Power Flow Analysis 53

 Example 7: Newton-Raphson Method
∂P2
V j Y2 j sin (θ2 j − δ2 + δj )
3

∂δ2
= ∑V
j =1, j ≠ 2
2

= V2 V1 Y21 sin(θ21 − δ2 )+ V2 V3 Y23 sin (θ23 − δ2 + δ3 )

= V2 1.05 22.4 sin(2.03 − δ2 )+ V2 1.04 35.8 sin (2.03 − δ2 + δ3 )
∂P2
= − V2 V3 Y23 sin(θ23 − δ2 + δ3 )= − V2 1.04 35.8 sin(2.03 − δ2 + δ3 )
∂δ3
∂P2
Y2 j cos(θ2 j − δ2 + δj )
3
= 2 V2 Y22 cos(θ22 )+ ∑V
∂V2
j
j =1, j ≠ 2

= 2 V2 Y22 cos(θ22 )+ V2 Y21 cos(θ21 − δ2 + δ1 )+ V2 Y23 cos(θ23 − δ2 + δ3 )

= 2 V2 58.1 cos(2.03)+ 1.05 22.4 cos(2.03 − δ2 )
+ 1.04 35.8 cos(2.03 − δ2 + δ3 )

GEE/PSAOpt/L3-Power Flow Analysis 54

 Example 7: Newton-Raphson Method

∂P3
= − V3 V2 Y32 sin (θ32 − δ3 + δ2 )= − 1.04 V2 35.8 sin (2.03 − δ2 + δ3 )
∂δ2
∂P3
V j Y3 j sin (θ3 j − δ3 + δj )
3

∂δ3
= ∑V
j =1, j ≠ 3
3

= V3 V1 Y31 sin (θ31 − δ3 + δ1 )+ V3 V2 Y32 sin (θ32 − δ3 + δ2 )

= 1.04 1.05 31.6 sin (1.89 − δ3 )+ 1.04 V2 35.8 sin (2.03 − δ3 + δ2 )
∂P3
= V3 Y32 cos(θ32 − δ3 + δ2 )= 1.04 35.8 cos(2.03 − δ2 + δ3 )
∂V2

GEE/PSAOpt/L3-Power Flow Analysis 55

 Example 7: Newton-Raphson Method
∂Q2
V j Y2 j cos(θ2 j − δ2 + δj )
3

∂δ2
= ∑V
j =1, j ≠ 2
2

= V2 V1 Y21 cos(θ21 − δ2 + δ1 )+ V2 V3 Y23 cos(θ23 − δ2 + δ3 )

= V2 1.05 22.4 cos(2.03 − δ2 )+ V2 1.04 35.8 cos(2.03 − δ2 + δ3 )
∂Q2
= − V2 V3 Y23 cos(θ23 − δ2 + δ3 )= − V2 1.04 35.8 cos(2.03 − δ2 + δ3 )
∂δ3
∂Q2
Y2 j sin (θ2 j − δ2 + δj )
3
= − 2 V2 Y22 sin (θ22 )− ∑V
∂V2
j
j =1, j ≠ 2

= − 2 V2 Y22 sin (θ22 )− V1 Y21 sin (θ21 − δ2 + δ1 )− V3 Y23 sin (θ23 − δ2 + δ3 )

= − 2 V2 58.1 sin (− 1.11)− 1.05 22.4 sin (2.03 − δ2 )
− 1.04 35.8 sin (2.03 − δ2 + δ3 )

GEE/PSAOpt/L3-Power Flow Analysis 56

 Example 7: Newton-Raphson Method

x [ k + 1] = x [ k ] + J − 1 ⋅∆c[ k ]
[ k + 1] [k ] −1
δ2  δ2  ∂P2 ∂δ2 ∂P2 ∂δ3 ∂P2 ∂V2  ∆P2 
[k ]

   
= δ3  = δ3  + ∂P3 ∂δ2 ∂P3 ∂δ3 ∂P3 ∂V2  ⋅∆P3 
   
V2  V2  ∂Q2 ∂δ2 ∂Q2 ∂δ3 ∂Q2 ∂V2 
  ∆Q2 

   

GEE/PSAOpt/L3-Power Flow Analysis 57

 Example 7: Newton-Raphson Method
0.0 P2sch  P2[ 0 ]  − 4.0 − 1.14  − 2.86
   [0]  
x [ 0 ] = 0.0 ∆c[ 0 ] = P3sch − P
3   = 2.0 − 0.562  = 1.438 
      
  Q2  
    − 2.5 − 2.28
  − 0.22
sch [0]
 
1.0 Q
 2   
∆x[ 0 ] = J − 1∆c[ 0 ]
∆δ2[ 0 ]   54.28 − 33.28 24.86 − 1 − 2.86 − 0.04526
 [ 0]  
∆x[ 0 ] = ∆δ3  = − 33.28 66.04 − 16.64 1.438  = − 0.00772
     
∆ V2  
 − 27.14 16.64 49.72 
  − 0.22 − 0.02655
 
[ 0 ]
 
δ2[1]  0.0 + (− 0.04526) − 0.04526
 
x [1] = δ3[1]  = 0.0 + (− 0.00772 ) = − 0.00772
   
 1.0 + (− 0.02655)
V2  
[1]
   0.9734 


GEE/PSAOpt/L3-Power Flow Analysis 58

 Example 7: Newton-Raphson Method

− 0.04526 P2sch  P2[1]  − 4.0 − 3.901 − 0.099

   [1]  
x [1] = − 0.00772 ∆c[1] = P3sch − P
3   = 2. 0 −  1.978  = 0.0217 
      

 0.9734  Q2sch  Q2  
  − 2.5 
− 2.449
  − 0.051
[1]
    
−1
 51.72 − 31.77 21.30  − 0.099 − 0.001795
∆x [1] = − 32.98 65.66 − 15.38 0.0217  = − 0.000985
     
− 28.54 17.40
 48.10 
 − 0.051
   − 0.001767

δ2[ 2 ]  − 0.04526 + ( − 0.001795) − 0.04706
 [2]  
x = δ3  = − 0.00772 + ( − 0.000985) = − 0.00870
[2]
   
V2  
  0.9734 + ( − 0.001767)  
 0.9717 
[ 2 ]
 

GEE/PSAOpt/L3-Power Flow Analysis 59

 Example 7: Newton-Raphson Method

− 0.04706 P2sch  P2[1]  − 4.0 − 3.999 − 0.0002

   [1]  
x [2] = − 0.00870 ∆c[ 2 ] = P3sch − P
3   = 2. 0 −  1.999  = 0.00004 
      
  Q2  
    − 2.5 − 2.499
   − 0.0001
sch [1]
 0.9717  Q
 2   
−1
 51.60 − 31.69 21.14  − 0.000216 − 0.000038
∆x [ 2 ] = − 32.93 65.60 − 15.35  0.000038  = − 0.000002
     

− 28.55 17.40 47.95 
 
− 0.000143
  − 0.000004

δ2[ 3]  − 0.04706 + ( − 0.000038)  − 0.04706 
 
x [ 3] = δ3[ 3]  = − 0.00870 + ( − 0.000002) = − 0.008705
   
V2  
  0.9717 + ( − 0.000004)     0.97168 
[ 3 ]
 

GEE/PSAOpt/L3-Power Flow Analysis 60

 Example 7: Newton-Raphson Method
 − 0.04706  P2sch  P2[1]  − 4.0 − 4.0 0.0000
   [1]  
x [ 3] = − 0.008705 ∆c[ 2 ] = P3sch − P
   = 2.0 −  2.0  = 0.0000
  3     
Q2sch  Q2  

 0.97168    − 2.5 − 2.5
   0.0000
[1]
    
εmax = 2.5 ×10− 4
P1 = V1 Y11 cos(θ11 )+ V1 V2 Y12 cos(θ12 − δ1 + δ2 )+ V1 V3 Y13 cos(θ13 − δ1 + δ3 )
2

Q1 = − V1 Y11 sin (θ11 )− V1 V2 Y12 sin (θ12 − δ1 + δ2 )− V1 V3 Y13 sin (θ13 − δ1 + δ3 )

2

Q3 = − V3 V1 Y31 sin (θ31 − δ3 + δ1 )− V3 V2 Y32 sin (θ32 − δ3 + δ2 )− V3 Y33 sin (θ33 )

2

P1 = 2.1842 pu
Q1 = 1.4085 pu
Q3 = 1.4617 pu

GEE/PSAOpt/L3-Power Flow Analysis 61

 Fast Decoupled Power Flow
l Transmission lines and transformers have high X/R ratios
u Real power change, ∆P
n is less sensitive to changes in the voltage magnitude, ∆|V|
n is more sensitive to changes in the phase angle, ∆δ
u Reactive power changes, ∆Q
n is less sensitive to changes in the phase angle , ∆δ
n is more sensitive to changes in the voltage magnitude, ∆|V|
u Jacobian submatrices JQd and JPV tend to be much smaller in
magnitude compared to JPd and JQV
l Jacobian submatrices JQd and JPV can be set to zero
∆ 
  = Pδ
P J 0  ∆ δ  ∆P = J P δ ⋅∆δ = ∂P
∂δ∆δ
∆
  
Q   0 J  
QV  
∆ V 
 ∆Q = J QV ⋅ ∆ V = ∂Q
∂V ∆ V

GEE/PSAOpt/L3-Power Flow Analysis 62

 Fast Decoupled Power Flow
∂Pi
l JPV elements = Vi Yij cos(θij − δi + δj )
∂V j
θij ≈90° δi ≈δj
∂Pi
≈Vi Yij cos(90°)= 0.0
∂V j
∂Qi
= − Vi V j Yij cos(θij − δi + δj )
l JQd elements
∂δj
θij ≈90° δi ≈δj
∂Qi
≈− Vi V j Yij cos(90°)= 0.0
∂δj

GEE/PSAOpt/L3-Power Flow Analysis 63

 Fast Decoupled Power Flow
l The matrix equation is separated into two decoupled
equations
u requires considerably less time to solve compared to the full
Newton-Raphson method
u JPd and JQV submatrices can be further simplified to eliminate the
need for recomputing of the submatrices during each iteration
n some terms in each element are relatively small and can be
eliminated
n the remaining equations consist of constant terms and one variable
term
n the one variable term can be moved and coupled with the change in
power variable
n the result is a Jacobian matrix with constant term elements

GEE/PSAOpt/L3-Power Flow Analysis 64

 Jacobian JPd Diagonal Terms
∂Pi
= ∑ Vi V j Yij sin (θij − δi + δj )
n

∂δi j =1
j ≠i

Yij sin (θij − δi + δj )

n
= − Vi Yii sin (θii )+ ∑VV
2
i j
j =1

∂Pi
Yij sin (θij − δi + δj )
n
= − Vi Yii sin (θii )− Qi Qi = − ∑VV
2

∂δi
i j
j =1

∂Pi
Yii sin (θii )= Bii Bii > > Qi = − Vi Bii
2

∂δi
∂Pi
Vi ≈Vi ⇒ = − Vi Bii
2

∂δi

GEE/PSAOpt/L3-Power Flow Analysis 65

 Jacobian JPd Off-diagonal Terms
∂Pi
= − Vi V j Yij sin (θij − δi + δj )
∂δj
δj − δi ≈0
∂Pi
= − Vi V j Yij sin (θij )
∂δj
Yij sin (θij )= Bij V j ≈1
∂Pi
= − Vi Bij
∂δj

GEE/PSAOpt/L3-Power Flow Analysis 66

 Jacobian JQV Diagonal Terms
∂Qi
Yij sin (θij − δi + δj )
n
= − 2 Vi Yii sin (θii )− ∑V
∂Vi
j
j =1
j ≠i

∂Qi
= − Vi Yii sin (θii )− Vi ∑ Vi V j Yij sin (θij − δi + δj )
n
−1

∂Vi j =1

∂Qi
Yij sin (θij − δi + δj )
n
= − Vi Yii sin (θii )+ Vi Qi ∑VV
−1
Qi = −
∂Vi
i j
j =1

∂Qi
Yii sin (θii )= Bii Bii > > Qi ⇒ = − Vi Bii
∂Vi

GEE/PSAOpt/L3-Power Flow Analysis 67

 Jacobian JQV Off-diagonal Terms
∂Qi
= − Vi Yij sin (θij − δi + δj )
∂V j
δj − δi ≈0
∂Qi
= − Vi Yij sin (θij )
∂V j
Yij sin (θij )= Bij
∂Qi
= − Vi Bij
∂V j

GEE/PSAOpt/L3-Power Flow Analysis 68

 Fast Decoupled Power Flow
l Individual power change equations in JPd and JQV
n
∆Pi n
∆Pi = ∑ − Vi Bij ∆δj ⇒ = ∑ − Bij ∆δj
j =1 Vi j =1
n
∆Qi n
∆Qi = ∑ − Vi Bij ∆ V j ⇒ = ∑ − Bij ∆ V j
j =1 Vi j =1

l Matrix equation for JPd and JQV

∆P ∆P
= − B′
∆δ ⇒ ∆δ= − [
B′]− 1
Vi V
∆Q ∆Q
∆V = − [ ] −1
= − B′
′
∆V ⇒ B′′
Vi V

GEE/PSAOpt/L3-Power Flow Analysis 69

 Example 8: Fast Decoupled Method
l Using the fast decoupled PF,
find the power flow solution Slack Bus
1 V1 = 1.05∠0°

0.01 + j0.03 0.02 + j0.04

0.0125 + j0.025
y12 = 10 − j 20 pu 2
3
y13 = 10 − j 30 pu
|V3| = 1.04
y23 = 16 − j 32 pu 200 MW 400 MW
400 + j 250 250 MVAR
S =−
sch
2 = − 4.0 − j 2.5 pu
100
200
P3sch = = 2.0 pu
100

GEE/PSAOpt/L3-Power Flow Analysis 70

 Example 8: Fast Decoupled Method

− 52 32 
B′= 
 32 − 62 
− 0.028182 − 0.014545
[B′
] =
−1

− 0.014545 − 0.023636

B′
′=[
− 52]
[B′
]− 1 = [− 0.019231]
′

GEE/PSAOpt/L3-Power Flow Analysis 71

 Example 8: Fast Decoupled Method
Initial values: 1.05∠0°
V [ 0] =
1. 00∠ 0°


1.00∠ 0°
First iteration:

P2sch  − 4.0 δ2[ k ]  0.0

   x [ k ] = δ[ k ]  x [ 0 ] = 0.0
y = P3sch  = 
 2.0  3   
Q2sch   −   [k ] 

1.0 
   2 .5 V
2  

Pinj 2 (x ) Pinj i = ∑ Vi V j Yij cos(θij − δi + δj )

n

 
f (x )= Pinj 3 (x ) j =1

Qinj 2 (x ) Qinj i = − ∑ Vi V j Yij sin (θij − δi + δj )

n

  j =1

GEE/PSAOpt/L3-Power Flow Analysis 72

 Example 8: Fast Decoupled Method
V2 2 Y22 cos(θ22 )+ V2 V1 Y21 cos(θ21 − δ2 + δ1 )+ V2 V3 Y23 cos(θ23 − δ2 + δ3 )
 2 
V Y cos(θ33 )+ V3 V1 Y31 cos(θ31 − δ3 + δ1 )+ V3 V2 Y32 cos(θ32 − δ3 + δ2 )
f (x )=  3 2 33
− V Y sin (θ )− V V Y sin (θ − δ + δ)− V V Y sin (θ − δ + δ)
 2 22 22 2 1 21 21 2 1 2 3 23 23 2 3


 

 V2 2 58.1 cos(− 1.11)+ V2 1.05 22.4 cos(2.03 − δ2 )+ V2 1.04 35.8 cos(2.03 − δ2 + δ3 ) 
 
=
1.04
2
67.2 cos (− 1 .17 )+ 1.04 1.05 31 .6 cos (1.89 − δ 3 )+ 1 .04 V2 35.8 cos (2.03 − δ 3 + δ )
2 
 − V 2 58.1 sin (− 1.11)− V 1.05 22.4 sin (2.03 − δ)− V 1.04 35.8 sin (2.03 − δ + δ) 
 2 2 2 2 2 3


 


GEE/PSAOpt/L3-Power Flow Analysis 73

 Example 8: Fast Decoupled Method
P2sch  P2[ 0]  − 4.0 − 1.14  − 2.86
   [ 0]   − 0.562  = 1.438 
∆y [ 0 ] = P3sch − P
3   = 2.0
    
Q2sch  Q2  
  − 2.5 
− 2.28 − 0.22
[ 0]
   

∆δ2[ 0 ]  0.028182 0.014545 − 2.86 1.0  − 0.06048

 [0]  =     = 
∆δ
 3   0.014545 0.023636 1.438 1.04   − 0.00891
[∆ V ]= [0.019231][− 0.22 1.0]= [− 0.004231]
2
[0]

δ2[1] = 0.0 + (− 0.06048)= − 0.06048

δ3[1] = 0.0 + (− 0.00891)= − 0.00891
V2[1] = 1.0 + (− 0.004231)= 0.995769

GEE/PSAOpt/L3-Power Flow Analysis 74

 Example 8: Fast Decoupled Method
Remaining iterations:
Iter δ2 δ3 |V2| ∆ P2 ∆ P3 ∆ Q2
1 -0.060482 -0.008909 0.995769 -2.860000 1.438400 -0.220000
2 -0.056496 -0.007952 0.965274 0.175895 -0.070951 -1.579042
3 -0.044194 -0.008690 0.965711 0.640309 -0.457039 0.021948
4 -0.044802 -0.008986 0.972985 -0.021395 0.001195 0.365249
5 -0.047665 -0.008713 0.973116 -0.153368 0.112899 0.006657
6 -0.047614 -0.008645 0.971414 0.000520 0.002610 -0.086136
7 -0.046936 -0.008702 0.971333 0.035980 -0.026190 -0.004067
8 -0.046928 -0.008720 0.971732 0.000948 -0.001411 0.020119
9 -0.047087 -0.008707 0.971762 -0.008442 0.006133 0.001558
10 -0.047094 -0.008702 0.971669 -0.000470 0.000510 -0.004688

GEE/PSAOpt/L3-Power Flow Analysis 75

The DC Power Flow

• The power flow problem can be simplified by

extending the fast decoupled power flow
 Completely neglect the 𝑄𝑄 − 𝑉𝑉 equation
 Assuming that the voltage magnitudes are constant at 1pu

𝛿𝛿𝑗𝑗 − 𝛿𝛿𝑘𝑘
𝑃𝑃𝑗𝑗𝑗𝑗 =
𝑋𝑋𝑗𝑗𝑗𝑗
• The real power balance equations:
Δ𝑃𝑃 = −𝑩𝑩′ Δ𝛿𝛿 ⇒ Δ𝛿𝛿 = −[𝑩𝑩′ ]−𝟏𝟏 Δ𝑃𝑃

GEE/PSAOpt/L3-Power Flow Analysis 76

 End of Lecture

GEE/PSAOpt/L3-Power Flow Analysis 77