(Reference: Power System Analysis by Grainger & Stevenson)

Formation of Z Bus

Nodal Admittance equation I BUS = YBUS VBUS

Nodal Impedance equation VBUS = Z BUS I BUS

For 3 Bus System

éV1 ù é Z11 Z12 Z13 ù é I1 ù
êV ú = ê Z Z Z ú ê I ú
ê 2 ú ê 21 22 23 ú ê 2 ú
ëV3 û ë Z 31 Z 32 Z 33 û ë I 3 û

Z ii for i = 1,2,3 are the driving point impedances

Z ij for i, j = 1,2,3 and i ¹ j are the Transfer impedances
 Formation of Z Bus contd.

éV1 ù é Z11 Z12 Z13 ù é I1 ù

êV ú = ê Z Z Z ú ê I ú
ê 2 ú ê 21 22 23 ú ê 2 ú
ëV3 û ë Z 31 Z 32 Z 33 û ë I 3 û

Z ii for i = 1,2,3 are the driving point impedances

Z ij for i, j = 1,2,3 and i ¹ j are the Transfer impedances

Vi
Z ii = " i, j - Thevenin eqt impedance at node i
Ii I j =0, I i ¹ 0

Vi
Z ij = " i, j
Ij I i =0, j ¹i
I j ¹0
 Z Bus Building Algorithm
(Modification of an existing Zbus)

- A straight forward method on digital Computer

- Start by adding a new bus to a reference bus

Original network bus voltages and currents remain same

 VP - VK = Z b I P
where,
VK = Z K 1 I1 + Z K 2 I 2 + .......Z KK ( I K + I P ) + .......Z Kn I n

and VP = Z K 1 I1 + Z K 2 I 2 + .......Z KK ( I K + I P ) + .......Z Kn I n + Z b I P

Þ VP = Z K 1 I1 + Z K 2 I 2 + .......Z KK I K + .......Z Kn I n + ( Z KK + Z b ) I P
By Applying Kron’s reduction,
Z Bus can be easily modified for network changes
 (1)

(2)

From (1)

Substitute in (2)
Substitute for VP , Vm and Vn
Vq-Vk
Zbus old

Derive
Form Zbus Using Building Algorithm
Branches 5 and 3 are mutually coupled through
mutual impedance j0.1
Home work
 Home work

1. Form ZBUS after removing node 5 by converting voltage source to current source.
2. Determine node voltages for all 4 nodes w.r.t . ref. node
3. Draw Thevenin Equivalent circuit at bus 4 and use it to determine the current drawn
by a capacitor of reactance 5.4p.u connected between bus 4 and reference
4. Modify ZBUS by removing the reactance connected between 1 and 3
 Calculation of ZBUS elements from YBUS
Calculation of mth column elements of ZBUS
Since the product of YBUS and ZBUS equals the unity matrix,

é0 ù é0 ù
ê0 ú ê0 ú
ê ú ê ú
ê. ú ê. ú
ê ú ê ú
Ybus Z bus ê. ú = ê. ú
ê1m ú ê1m ú
ê ú ê ú
ê. ú ê. ú
ê0 ú ê0 ú
ë û ë û é0 ù
ê0 ú
ê ú
ê. ú
(m) ê ú
Ybus Z bus = ê. ú
ê1m ú
ê ú
ê. ú
ê0 ú
ë û
Solve using LU decomposition Technique
é0 ù
ê0 ú
ê ú
ê. ú
(m) ê ú
LU Z bus = ê. ú
ê1m ú
ê ú
ê. ú
ê0 ú
ë û
Consider a 4 bus system, and It is required to compute the elements of
3rd column of Z bus
Solve for elements of column m of ZBUS in two steps
In order to find the difference between 2 columns of ZBUS

Where,

ZBUS (m-n) = ZBUS (m) - ZBUS (n)

Computational Techniques for Solving Large Network equations

Network equation:

Ybus Vbus = Ibus

Vbus = Ybus -1 Ibus

Direct Matrix Inversion is a tedious task for large systems
Computational Techniques for Solving Large Network equations

Solution of Linear Algebraic equations - Summary

• Direct Method Vbus = Ybus -1 Ibus

• Method of Successive Elimination

– Gaussian Elimination Method
(Forward Elimination and backward substitution Method)- Traingularised
eqns
– Gauss Jordan Elimination Method (Diagonal set of equations)- No
backward substitution
– LU Decomposition method

• Iterative Techniques
• Gauss Iterative Technique
• Gauss- Seidel Iterative Technique
 Solution of Non-Linear Algebraic equations

• Iterative Techniques
• Gauss Iterative Technique
• Gauss- Seidel Iterative Technique
• Newton Raphson Method
Gaussian Elimination (Method of successive elimination)

Forward Elimination and Backward Substitution

Solve the network equation using Gaussian elimination process. At each
step, draw the network equivalent
Step1: Eliminate Variable V1 from eqns 2,3, and 4

Eqt. Ckt after step 1

Step 2: Eliminate Variable V2 from eqns 3, and 4

Eqt. Ckt after step 2

 Step 3:

Eqt. Ckt after step 3

Step4:

Backward substution
 (Kron’s Reduction (Node Elimination)

• The buses to which there is no external load or generating sources connected

• Not necessary to calculate the voltage explicitly
• These nodes are eliminated from the nodal admittance equations to obtain a
reduced order model
 Formation of L and U matrices

Ybus=

Step 1
1. Eliminate the coeffs. Y11, Y21, Y31, Y41 using Gaussian elimination from the first
column.
Enter these coeffs in the first column of L matrix
2. New coeffs 1, Y12/Y11, Y13/Y11, Y14/Y11 are generated to replace those in the first row
Enter these coeffs in the first row of U matrix

L= U=

Similarly step 2, step3 and step4

Step 2

L= U=

Step 3

L= U=
Step4

L= U=
 Sparsity Oriented programming
Sparsity S=Z/n2

• Compact storage scheme, optimize the computer memory, and to

decrease the computation time
• Only non zero elements are stored
• Full position identification takes place

Consider a Ybus

15 -10 -5 0 0 0
-10 10 0 0 0 0 Total 36 elements(n)
20 are zero elements (Z)
-5 0 25 -12 -8 0
0 0 -12 12 0 0 Sparsity S=Z/n2 = 0.556
0 0 -8 0 23 -15 = 55.6%
0 0 0 0 -15 15
 15 -10 -5 0 0 0
-10 10 0 0 0 0
-5 0 25 -12 -8 0
0 0 -12 12 0 0
0 0 -8 0 23 -15
0 0 0 0 -15 15

4 vectors are defined

1. YDIAD=[15, 10, 25, 12, 23, 15]

2. YOFFD=[-10, -5, -10, -5, -12, -8, -12, -8, -15, -15]

To position the elements of YOFFD in their proper place in Ybus, 2 more vectors
are defined

1. IROW=[1, 3, 4, 7, 8, 10] defines where each row starts in the vector YOFFD

2. ICOL= [2, 3, 1, 1, 4, 5, 3, 3, 6, 5] identifies the column position of each

element in YOFFD
1. YDIAD=[15, 10, 25, 12, 23, 15]

15
10
25
12
23
15

2. YOFFD=[-10, -5, -10, -5, -12, -8, -12, -8, -15, -15]
YOFFD=[-10, -5, -10, -5, -12, -8, -12, -8, -15, -15]

• Position vectors
1. IROW=[1, 3, 4, 7, 8, 10] identifies where each row starts in the vector YOFFD
2. ICOL= [2, 3, 1, 1, 4, 5, 3, 3, 6, 5] identifies the column position of each element in
YOFFD
Element 1of YOFFD is in the first
row
15 -10 -5 Element 3 of YOFFD is in row 2
-10 10 Element 4 of YOFFD is in row 3
Element 7 of YOFFD is in row 4
25
Element 8 of YOFFD is in row 5
12 Element 10 of YOFFD is in row 6
23
15
 Gaussian Elimination & Optimal Ordering
When a Sparse matrix is triangulated - sparsity get effected

- Non-zero terms are added to the upper Triangle

- Called “ Fill –in Process”

The total computation time increases with more “ Fill –in” terms

How to conserve sparsity??

 Sparsity & Optimal Ordering

To Minimize the accumulation of non-zero terms ‘Optimal order’ of

row elimination is used

Near Optimal Ordering

- Optimal ordering to conserve sparsity is practically impossible and complex

- So .. Uses Near or semi optimal ordering
- Applicable to sparse matrix that are symmetric in ‘pattern’
 Scheme 1
v Preorder the rows before triangulation
v Number the rows according to the ascending order of number of non-
zero off diagonal elements

Dynamic ordering
Scheme 2:
At each step of elimination, the next row to be processed is that with fewest non-
zero off-diagonal terms
Scheme 3:
At each step of elimination, the next row to be processed is the one that will
introduce the fewest non-zero off-diagonal terms after elimination
 Tutorial

1. The following vectors were defined to represent the Bus

admittance matrix (YBUS) of a 5-bus power system, in sparsity
oriented programming. Form the Bus admittance matrix.
YDIAG= [3, 13, 5, 10, 13]
YOFFD=[-1, -2, -1, -5, -4, -3, -5, -4, -6, -2, -3, -6]
Position vector ROW=[ 1, 3, 7, 8, 10]
Position vector COL=[2, 5, 1, 3, 4, 5, 2, 2, 5, 1, 2, 4]

2. Optimally order the buses of 5-bus power system given in

previous question, using scheme 1 and write the optimally
ordered YBUS.