EE 521 Analysis of Power Systems
ChenChing Liu
Washington State University
1. ShermanMorrisonWoodbury Formula:
Let M∈L(R
n
) be invertible and let U, V be n×m
(m≤n) matrices. Then (M+UV
T
) is invertible if
and only if (I+V
T
M
1
U) is invertible, and:
To prove (1), we simply premultiply and post
multiply (M+UV
T
) with its inverse matrix to check
whether the result is an identity matrix.
Let H=(I+V
T
M
1
U)
1
, then
) 1 ( M V U) M V (I U M  M ) UV (M
1 T
1
H
1 T 1 1 1 T
¸ ¸¸ ¸ ¸_ ¸
+ = +
a) Premultiplication
b) Postmultiplication
¸
 
( )  
I M V H H I U I
M V H U M V I I U I
M V UH M V H I U I
M UHV M UV M UHV MM M UV MM
T
I
T T
T T
T T T
I
T
=
(
¸
(
¸
− + =
+ − + =
− − + =
− − + =
(
(
¸
(
¸
+ +
− −
− −
− −
− − −
−
− −
1 1
1 1
1 1
1 1 1
1
1 1
1  T
1 
H
1  T 1  1  T
M V U) M V (I U M  M ) UV (M
¸ _ ¸
¸ ¸¸ ¸ ¸_ ¸
 
( )  
  I V HH I U M I
V U M V I H I U M I
V U M HV H I U M I
UV M UHV M M M UHV M UV M M M
T
T T
T T
T T T T
= − + =
+ − + =
− − + =
− − + =
+
(
(
¸
(
¸
+
− −
− −
− −
− − − − − −
1 1
1 1
1 1
1 1 1 1 1 1
T 1  T
1 
H
1  T 1  1 
) UV (M M V U) M V (I U M  M
¸ ¸¸ ¸ ¸_ ¸
2. ShermanMorrison Formula:
In eq. (1), if m=1, i.e., U and V are column
vectors, let u and v (u,v∈ R
n
) represent these two
vectors, then let equation (1) can be simplified:
3. Special Case:
Suppose M is symmetrical and M
1
is known; and
in (2) we have uv
T
=µaa
T
, where µ is a scalar and
a is an nvector, then (2) becomes:
where:
( )
( )
) 2 (
1
1
1 1
1
1
1
− −
−
−
−
(
¸
(
¸
+
− = + M uv M
u M v
M uv M
T
T
T
) 3 (
1
¸ _ ¸
T
rank
T
bb M aa M γ µ − =
(
(
¸
(
¸
+
−
( )
1
1
1
−
−
−
+ =
=
b a
a M b
T
µ γ
Example 1: Given
Let
Find M
1
1
Solution: Here µ=2 and a=[0 1 0]
T
, So
b=M
1
a=[1 2 –1]
T
γ=(µ
1
+a
T
b)
1
=0.4
Use(3)
The result can be verified by direct
multiplication.
(
(
(
¸
(
¸
−
− −
−
=
(
(
(
¸
(
¸
=
3 2 1
2 2 1
1 1 1
M
1 1 0
1 2 1
0 1 2
1 
M
¸
 
¸¸ ¸¸ ¸
T
a
M M 0 1 0
0
1
0
2
1 1 0
1 4 1
0 1 2
(
(
(
¸
(
¸
+ =
(
(
(
¸
(
¸
=
µ
 
(
(
(
¸
(
¸
−
− −
−
= − −
(
(
(
¸
(
¸
−
−
−
(
(
(
¸
(
¸
−
− −
−
=
−
4 . 1 4 . 0 2 . 0
4 . 0 4 . 0 2 . 0
2 . 0 2 . 0 6 . 0
2 2 1
2
2
1
4 . 0
3 2 1
2 2 1
1 1 1
1
1
M
4. Comments:
(M+UV
T
) is a modification of M. Since M
1
is
assumed known, the inverse of this modified
matrix can be performed without inverting an
n×n matrix again. Instead, the necessary
matrix inversion is only for an m×m matrix.
When m=1 (Eqs. (23)), the matrix inversion
requires only matrix multiplications and matrix
additions.
The admittance matrix is symmetrical, and its
change can be easily represented by finding
a vector and a scalar µ. The corresponding
computational efforts is lower.
a
5. Inverting Admittance Matrix When Line
Admittances Are Changed
Suppose that we have Y
bus
and its inverse Z
bus
.
Suppose that the admittance of the kth branch
connecting node i and node j is changed from y
k
To y
k
+∆ y
k
. Let to be the new admittance
matrix. We need to find its inverse, .
First, we try to find a vector and scalar µ
k
.
There are 2 cases to consider:
Case 1: Neither node i nor node j is the reference
node. Then we get the new admittance matrix:
i line k j
Ref.
new
bus
Z
new
bus
Y
k
a
Thus in we can put 1 in the ith row and –1 in the
jth row. All other entries are zero. Also µ
k
= ∆y
k
.
 
T
k k bus
bus
bus
new
bus
a a Y
Y
Y Y
k
k
k k
k k
y
1  1
1
1
y
y y 
y  y
∆ + =
(
(
(
(
(
(
¸
(
¸
−
∆ + =
(
(
(
(
(
(
¸
(
¸
∆ ∆
∆ ∆
+ =
. . .
k
a
Case 2: The kth branch connects the ith node to the
reference node. Similar to case 1, we can put a
1 in the ith row of , and put all other entries
zero. Also we have µ
k
= ∆y
k
.
i line k Ref
Now we consider calculation of . We already have
Z
bus
=[Y
bus
]1
. By eq. (3) we obtain:
where
k
a
new
bus
Z
(4) ) ( ) (
1 1
T
k k k bus
T
k k k bus
new
bus
new
bus
b b Z a a y Y Y Z γ − = ∆ + = =
− −
(5)
1
(
¸
(
¸
+
∆
=
=
k
T
k
k
k
k bus k
b a
y
a Z b
γ
Let the ith and the jth columns of Z
bus
be Z
i
and Z
j
,
respectively, and the iith, jjth, and ijth elements be
Z
ii
, Z
jj
, and Z
ij
, respectively. According to the two
cases above, Eq. (5) can be simplified:
Case 1: by Eq. (5):
(6)
Case 2: j is the reference node:
(7)
 
   
1
2
1
1 1
1
1
−


.

\

− + +
∆
=
− − =
− =
(
(
(
(
(
(
¸
(
¸
−
=
ij jj ii
k
k
j i k
T
k
j i j i k
z z z
y
Z Z T b a
Z Z Z Z b
γ
. . .
. . .
1
(
¸
(
¸
+
∆
=
=
ii
k
k
i k
z
y
Z b
γ
Example 2: Suppose that for a given network we already have
The network is shown in the figure.
(all values are line admittances p.u.)
( to are branch number)
Note that at the beginning there are no connections between node 1
and node 2. Now we add a line (branch number 5)
between node 1 and node 2. The line admittance is –j10.
Find:
(
(
(
¸
(
¸
⋅ =
(
(
(
¸
(
¸
−
−
−
⋅ =
2 . 0 2 . 0 1 . 0
2 . 0 5 . 0 1 . 0
1 . 0 1 . 0 2 . 0
3
and
30 10 10
10 10 0
10 0 20
j
Z j Y
bus bus
j10 j10
j10
j10
2
3 Ref.
1
1
2
5
4
3
1 5
new
bus
Z
Solution: The change of the admittance of branch 5
∆y
5
=j10j0=j10
Since neither node 1 nor node 2 is the reference,
we use Eq. (6).
75 . 3 )) 2 . 0 5 . 0 2 . 0 (
3 10
1
(
) 2
1
(
1 . 0
4 . 0
1 . 0
3
2 . 0
5 . 0
1 . 0
1 . 0
1 . 0
2 . 0
3
1
1
12 22 11
5
5
2 1 5
j
j
j
z z z
y
j j
z z b
− = − + +
−
=
− − +
∆
=
(
(
(
¸
(
¸
−
− =
¦
)
¦
`
¹
¦
¹
¦
´
¦
(
(
(
¸
(
¸
−
(
(
(
¸
(
¸
= − =
−
−
γ
Then by using equation (4):
We can also get directly from the figure:
To check the result, one can simply multiply with
 
(
(
(
¸
(
¸
⋅ =
(
(
(
¸
(
¸
−
−
− −
−
(
(
(
¸
(
¸
⋅ =
− − ⋅
(
(
(
¸
(
¸
−
− ⋅ + =
− =
1875 . 0 1500 . 0 1125 . 0
1500 . 0 3000 . 0 1500 . 0
1125 . 0 1500 . 0 1875 . 0
3
01 . 0 04 . 0 01 . 0
04 . 0 16 . 0 04 . 0
01 . 0 04 . 0 01 . 0
3
25 . 1
2 . 0 2 . 0 1 . 0
2 . 0 5 . 0 1 . 0
1 . 0 1 . 0 2 . 0
3
1 . 0 4 . 0 1 . 0
3
1 . 0
4 . 0
1 . 0
3
75 . 3
j
j
j
j j
j Z
b b Z Z
bus
T
k k k bus
new
bus
γ
new
bus
Z
new
bus
Y
new
bus
Y
(
(
(
¸
(
¸
−
−
−
⋅ =
30 10 10
10 20 10
10 10 30
j Y
new
bus