The Newton-Raphson Method of Solution

The Newton-Raphson method of solution enables us to replace the

nonlinear set of power equations with a linear set, simplifying the
solution to some extent. We first review the iterative procedure of NR
and then apply to power systems.

The Process of Linearization:

Consider the Taylor series expansion of a nonlinear function f(x) around

a point ‘a’

f 2 f (x)2
f ( x)  f ( x0 )  x  2  .......  Rn
x x x0 x x  x0
2!

The derivatives are calculated at point ‘x0’ and Δx=x-x0. The series
converges if the higher order terms Rn→0. If Δx<<1, the higher order
terms can be neglected. We then write,

f
f ( x) f (a)  x
x x x0
Now, consider the nonlinear equation

f ( x)  c (1)
Linearized form of this equation around x0, would be

f
f ( x)  f ( x0 )  x
x x x0
f
c  f ( x0 )  x
x x x0
If we have n sets of equations similar to (1) given below,
 f1 ( x1 , x2 ,........xn )  c1
f 2 ( x1 , x2 ,........xn )  c2



f n ( x1 , x2 ,........xn )  cn
X 0 [ x1o , x2o ,.....xno ],

The linearized set of equations can be written in the matrix form as,

 f1 f1 f1 

 x   
x2 xn 
 c1  f1 ( x1 , x2 ,....xn )     x1 
0 0 0 1

 0   f 2 f 2 f 2   
   x
 c2  f 2 ( x1 , x2 ,....xn )   x1
0 0

x2 xn   2 
      
         
        
     
      
 0 
        
cn  f n ( x1 , x2 ,....xn )   f   xn 
0 0

 n f n f n 
  
 x1 x2 xn 
This can be written in the form,

K  J X
Solve for X as,
X  J 1 K

J is called the Jacobian matrix. The procedure for iterative calculation

involves the following steps:

1) For a given value of X [ x1 , x2 ,.....xn ], calculate Jacobian J and the

0 o o o

‘mismatch’ K

2) Solve for ΔX from the equation, ΔX=J-1K

3) Update the value of X by,

X1 = X0 + ΔX

4) Replace this value to calculate new J and new K

5)The process is repeated until,

X k 1  X k  xk   , for all k , where  is the pre-selected tolerance

No acceleration factor is needed in the NR algorithm. Calculations are

somewhat complicated, but the convergence is rapid.
 Example

Solve the following set of equations by NR method. Assume starting

values to be 4 and 9.

x1  x2  15
x1 x2  50
Here,

f1  x1  x2
f 2  x1 x2

The Jacobian is,

 f1 f1 
 x x2   1 1
J  1 
 f 2 f 2   x2 x1 
 x x2 
 1

The algorithm is,

X i 1  X i  x
 X i  J 1 K
1
 x1i 1   x1i   1 1  15  ( x1i  x2i ) 
Or ,  i 1    i    i
 x2   x2   x2 x1i   50  x1i x2i 
Iteration 1

For x10  4 and x20  9

1
 x   4  1 1   2 
1

 1   9   9 4  14 
1

 x2       
 x11  5.2, x21  9.8

Iteration 2

Starting with x11  5.2, x21  9.8

1
 x12  5.2   1 1   0 
 2   9.8  9.8 5.2   0.96 
 x2       
 x12  4.9913, x22  10.0087

Complete the table:

i 0 1 2 3 4

x1 i 4 5.2 4.9913 4.99918 5

x2 i 9 9.8 10.0087 10.00002 10

 Newton-Raphson Method for Power Flow Problems

To apply the Newton-Raphson method to the solution of AC load flow

equations, we may choose to express the bus voltages and line
admittances in polar form or rectangular form. Here, in the following, we
use the polar form.

Let,

Vi  Vi e ji , or , Vi  i
j j
Vj  Vj e
jij
Yij  Yij e

Then the complex power at bus i is

Pi  jQi  Vi * I i
n
 Vi  YijV j
*

j 1
n j (   i  j ij )
  VV Y e
i j ij
j 1

Or,
n
Pi  jQi   VV Y (ij   j   i )
i j ij
j 1

So, we get,

n
Pi   VV Y cos (ij   j   i )
i j ij
j 1 (1)
n
Qi    VV Y sin(ij   j   i )
i j ij
j 1
The Newton-Raphson formulation requires the solution of the above 2
nonlinear equations for each bus.

PQ Buses

Consider that real and reactive power Pi and Qi are known and magnitude
and angle of the voltages are unknown. Exclude bus 1 because normally
this is selected as the slack bus. Calculations at the slack bus are
performed after the power flow has converged.

As in the nonlinear system considered earlier, we make a Taylor series

expansion for real and reactive power expressions at each bus. Then
equation

K=J ΔX

takes the following shape,

 P2 P2 P2 P2 

       
 3  V2  V3
 2 
 P2   P3 P3 P3 P3    2 
 P           
 3  V2  V3  
 3  2  3

              
    
              
              
   
 Q2   Q2 Q2
  
Q2 Q2


   2 
V
 Q3    2  3  V2  V3   V 
    3 
    Q3 Q3
  
Q3 Q3
      
    2  3  V2  V3   
     
         
   
 
          
Here,

Pi  Pi ,specified  Pi ,calculated

Qi  Qi ,specified  Qi ,calculated
Pi, calculated and Qi, calculated terms in the above are obtained from (1)

n
Pi   VV Y cos (ij   j   i )
i j ij
j 1
n
Qi    VV Y sin(ij   j   i )
i j ij
j 1

by substituting the estimated values of voltage magnitude and angle. The

partial derivatives in the Jacobian matrix are also calculated at the
estimated values of voltages. The linear equation is written in the
simplified form as,

 P   J1 J 2    
 Q    J J 4    V 
   3
The expressions for sub-Jacobians are,

P P Q Q
J1  , J2  , J3  , J4 
 V  V
PV Buses

Consider bus 2 to be a PV (generator) bus. Since magnitude of V2 is

constant,  V2 is zero. The corresponding column in the Jacobian is to be
deleted.

Also, since Q2 is not pre-specified, the row corresponding to ΔQ2 also

should not appear in the formulation.