01 LoadFlow R1 PDF

LOAD FLOW
A. J. Conejo
Univ. Castilla – La Mancha

2011
22 July 2011 1
The load flow problem
0. References
1. Introduction
2. Problem formulation
Two-bus case
Matrix YBUS
General equations
Bus classification
Variable types and
limits
22 July 2011 2
3. The Gauss-Seidel solution technique

Introduction
Algorithm initialization
PQ Buses
PV Buses
Stopping criterion
22 July 2011 3
4. The Newton-Raphson solution technique

Introduction
General fomulation
Load flow case
Jacobian matrix
Solution outline
22 July 2011 4
5. Fast decoupled AC load flow

6. Adjustment of bounds
7. DC load flow
8. Comparison of load flow solution methods
22 July 2011 5
References
1. A. Gómez Expósito, A. J. Conejo, C.

Cañizares. “Electric Energy Systems.
Analysis and Operation”. CRC Press, Boca
Raton, Florida, 2008.
2. A. R. Bergen, V. Vittal. “Power Systems

Analysis”. Second Edition. Prentice Hall,
Upper Saddle River, New Jersey, 1999.
22 July 2011 6
1. Introduction
22 July 2011 7
Introduction
• A snapshot of the system
• Most used tool in steady state power system

analysis
• Knowning the demand and/or generation of

power in each bus, find out:
– buses voltages
– load flow in lines and transformers
22 July 2011 8
Introduction
• The problem is described throught a non-

lineal system of equations
• Need of iterative solution techniques
• Solution technique: accuracy vs. computing

time
22 July 2011 9
Introduction
• Applications:
1. On-line analyses
• State estimation
• Security
• Economic analyses
22 July 2011 10
Introduction
2. Off-line analyses
• Operation analyses
• Plannig analyses
Network expansion planning
Power exchange planning
Security and adecuacy analyses
- Faults
- Stability
22 July 2011 11
2. Problem formulation
22 July 2011 12
Problem formulation
Two-bus case
We want to find out the relationship between

j
Si  Pi  jQi and Vi  Vie i in all buses of the power
system
I1 YS I2
V1 YG V2
YG
22 July 2011 13
Transmission line π model
Problem formulation
Two-bus case
Using Kirchhoff laws:

I1  V1 YG  ( V1  V2 ) Y S
I2  V2 YG  ( V2  V1) Y S
Matrix notation:
 I1   Y G  Y S  Y S   V1   Y11 Y12   V1 
I      V      V 
 2    YS Y G  Y S   2   Y 21 Y 22   2 
IBUS   YBUS  VBUS 
22 July 2011 14
Problem formulation
Two-bus case
• Complex power injected in each bus:

S1  SG1  SD1  P1  jQ1  V1  I1*
S2  SG2  SD2  P2  jQ2  V2  I2*
S1  V1  I1*  V1  ( Y11
*
V1*  Y12
*
V2* )
S2  V2  I2*  V2  ( Y21
*
V1*  Y22
*
V2* )
22 July 2011 15
Problem formulation
Two-bus case continuation
Notation:
Yik  Yik e j ik
Vi  Vi e j i
Replacing:
2
P1  jQ1  V1  Y1k Vk e j( 1 k 1k )
k 1
2
P2  jQ2  V2  Y2k Vk e j( 2 k 2k )
k 1
22 July 2011 16
Problem formulation
Two-bus case continuation
Therefore, the no lineal equations for the 2

buses network are:
2
Pi  Vi   Yik Vk cos (i  k  ik )
k 1
2
Qi  Vi   Yik Vk sin (i  k  ik )
k 1
i  1, 2
22 July 2011 17
Problem formulation
Matrix Ybus
• Two bus case
 Y11 Y12   YG  YS  YS 
YBUS  

 Y21 Y22    YS YG  YS 
22 July 2011 18
General building rules
Matrix Ybus
1. Self admittance of node i, Yii ,equals the
algebraic sum of all the admittances
connected to node i
2. Mutual admittance between nodes i and

k, Yik ,equals the negative of the sum of
all admittances connecting nodes i and k
3. Yik  Yki
22 July 2011 19
Problem formulation
Matrix Ybus
• Caracteristics of YBUS
1. YBUS is symmetric
is very sparse
2. YBUS
(>90% for more than 100 buses)
22 July 2011 20
Ybus example
Shunt element YG  j 0.01 (two per line)

Series element ZS  j 0.1
 j19.98 j10 j10 

YBUS   j10  j19.98 j10 
 
 j10 j10  j19.98 
22 July 2011 21
Problem formulation
General equations
• 2n equations (static load flow equations)

n
Pi  PGi  PDi  Vi   Yik Vk cos (i  k  ik ) 1
k 1
n
Qi  QGi  QDi  Vi   Yik Vk sin (i  k  ik ) 2
k 1
i  1,..., n
n  bus system
22 July 2011 22
Problem formulation
General equations
Polar representation for voltages and rectangular
for admittances
n
Pi  Vi   Vk (Gik cos ik  Bik senik ) [1]
k 1
n
Qi  Vi   Vk (Gik senik  Bik cos ik ) [2]
k 1
where
Y ik  Gik  j Bik
ik  i  k
22 July 2011 23
Problem formulation
General equations
• 4n variables
Vi, i,Pi, Qi ; i  1,...,n
• If 2n variables are specified, the other 2n are

determined by equations [1] and [2]
22 July 2011 24
Problem formulation
BUS Classification
1. PQ buses
Pi known ( PDi known, PGi zero)

Qi known (QDi known, QGi zero)
Vi unknown
i unknown
22 July 2011 25
Problem formulation
BUS Classification
2. PV buses
Pi known
Qi unknown
Pi known ( PGi specified, PDi known)

Vi known (specified)
Qi unknown ( QGi unknown, QDi known)

i unknown
22 July 2011 26
Problem formulation
BUS Classification
3. Slack bus, generator with large capacity.

V1 known (specified)
1 known (specified, typically 1  0 
reference)
P1 unknown ( PD1 known, PG1 unknown)
Q1 unknown ( QD1 known, QG1 unknown)
22 July 2011 27
Problem formulation
Variable types and limits
• Power balance
n n
 PGi   PDi  PLOSS
i1 i1
n n
 QGi   QDi  QLOSS
i1 i1
22 July 2011 28
Problem formulation
• Variable types
 Control variables
PGi (excepting slack bus)
QGi or V i
 Non-control variables
PDi QDi
 State variables
Vi i
22 July 2011 29
Problem formulation
• Variable limits
 Voltage magnitude Vi min  Vi  Vi max
 Power angle i  k  i  k max (every existing line)
 Power limits
PGi,min  PGi  PGi,max

QGi,min  QGi  QGi,max
22 July 2011 30
3.The Gauss-Seidel solution
technique
22 July 2011 31
Gauss-Seidel solution technique
No lineal system:
f (x)  0  x  F( x )
Iteration
x (r 1)  F( x (r ) )
Stoping rule
x (r 1)  x (r )  
22 July 2011 32
 Example f ( x)  x 2  5x  4  0
x  0 . 2 x 2  0 .8
x (r 1)  0.2( x (r ) )2  0.8 ; x ( 0 )  2
r  0, x (1)  0.2  22  0.8  1.6
r  1, x ( 2 )  0.2  1.6 2  0.8  1.312
r  2, x ( 3 )  0.2  1.312 2  0.8  1.1442
...
r  5, x ( 6 )  1.0103
r  6, x ( 7 )  1.0042
...
r  10, x (11)  1.0001
r  11, x (12 )
 1.0000
Many iterations!
r  12, x (13 )  1.0000
22 July 2011 33
Algorithm beginning
1) Known
Pi i  2,...,n (PV & PQ Buses)
Qi i  m  1,...,n (PQ Buses)
Vi i  2,...,m (PV Buses)
V1 slack bus
V i min , V i max i  m  1,...,n (PQ Buses)
QGi,min , QGi,max i  2,...,m (PV Buses)
22 July 2011 34
Algorithm beginning
2) Build YBUS
3) Initialize voltages
0
Vi  Vi i  m  1,...,n
0
i  i i  2,...,n
22 July 2011 35
PQ buses
4) PQ buses
* * * * n
Si  Pi  jQi  V i Ii  Vi Y ii V i  
V i Y ik Vk
k 1
k i
 
1 Pi  jQi n

Vi 
Y ii  V i*
  Y ik V k

k 1
 k i 
P  jQi
Ai  i i  m  1,...,n
Y ii
Y ik
Bik  i  m  1,...,n ; k  1,...,n ; k  i
Y ii
22 July 2011 36
PQ buses
At iteration (r+1) and bus i, the available values of

voltages at previous buses are used:
Ai i1 n
(r 1) (r 1)
  Bik V k  Bik V k
(r )
Vi  (r ) *

(Vi ) k 1 k i1
22 July 2011 37
PV buses
5) PV buses
* * * n
Si  Pi  jQi  V i Ii  
V i Y ik V k
k 1
 * n 
Qi   V i  Y ik V k 
 k 1 
22 July 2011 38
PV buses
At iteration (r+1):
 (r ) * i1 (r 1) (r ) * n (r ) 
Q(ir 1)  ( V i )  Y ik V k  ( V i )  Y ik V k 
 k 1 k i 
Pi  jQ(ir 1)
i  angleV i 
(r 1)
Ai 
Y ii
 A (r 1) i1 (r 1) n (r ) 


(ir 1)  angle (r )   Bik V k   Bik V k 
i
( V i )* k 1 k i1 
22 July 2011 39
PV buses
Beware of limits!
Q(ir 1)  Qi,min  Q(ir 1)  Qi,min & i becomes PQ
Q(ir 1)  Qi,max  Q(ir 1)  Qi,max & i becomes PQ
(more on this below)
22 July 2011 40
PV buses
6) Stop criterion
Vi(r 1)  Vi(r )   ; i  2,...,n
6.1) Slack bus power (after convergence)

* n
P1  jQ1  
V1 Y1k V k
k 1
22 July 2011 41
Algorith final calculations
6.2) Compute line currents (after convergence)

ILik  ( V i  V k ) YLik ; Iik 0  V i Y ik 0
ILki  ( V k  V i ) YLik ; Iki0  V k Y ki 0
Vi Vk
i ILik YLik ILki k
Sik Iik 0 Iki0 Sik
Y ik 0 Y ki0
22 July 2011 42
6.2) Compute line complex power (after convergence)
* * * * *
Ski  V k ( V k  V i )YLik  V k V k Y ki 0 ;
* * * * *
Sik  V i ( V i  V k )YLik  V i V i Y ik 0 ;
Vi Vk
i ILik YLik ILki k
Sik Iik 0 Iki0 Sik
Y ik 0 Y ki0
22 July 2011 43
6.3) Compute losses (after convergence)

Sloss , ik  Sik  Ski ;  k, i
Sloss   Sloss , ik
 k,i
7) If no convergence, go to step 4.
22 July 2011 44
Algorithm improvement
• Acceleration factor (in order to decrease

the number of iterations):
~
(r 1)
Vi(r 1)  V i   V i
(r ) (r )
 Vi 
  1.6 (generally recommended)
22 July 2011 45
Gauss-Seidel
Matlab code
function [Vfinal,angfinal,nite,P,Q,errorplot,tiempo]=Gaussgen(m,n,Ybus,Vmodini,Angini,P,Q,tol,Vmax,Vmin,Qmax,Qmin)
%--------------------------------------------------------------------------------------------------------------------------------------------------
%-function [Vmod,ang,nite,P,Qerrorplot,tiempo]=Gaussgen(m,n,Ybus,Vmodini,Angini,P,Q,tol,Vmax,Vmin,Qmax,Qmin)
%-Resuelve de forma general problema de carga por el método de Gauss-Seidel
%-donde:
%-2...m nudos PV; (m=1 cuando no hay nudos PV)
%-n nudos totales
%-Ybus matriz de admitancias
%-Vmodini tensiones iniciales modulo
%-Angini angulos iniciales RADIANES
%-P potencia activa inicial
%-Q potencia reactiva inicial
%-tol tolerancia para error en tension y potencia reactiva
%-Vmax Vmin valores limites aceptables para las tensiones
%-Qmax Qmin valores limites aceptables para las potencias reactivas
%-nite número de iteraciones
%-Vfinal vector con todas las potencias para cada iteración
%-angfinal igual pero con los ángulos
%-tiempo, tiempo invertido en hacer las operaciones
%---------------------------------------------------------------------------------------------------------------------------------------------------
%calculo de la matriz B
B=zeros(n,n);
for t=1:n
for k=1:n
B(t,k)=Ybus(t,k)/Ybus(t,t);
end
end
22 July 2011 46
Gauss-Seidel
Matlab code 2
%calculos los valores en coordenadas cuadrangulares de la tension
V=zeros(n,1);
for a=1:n
V(a)=Vmodini(a)*exp(i*Angini(a));
end
ang=Angini;
%empieza el bucle:
error=1; %valores iniciales para poder entrar en el bucle
errorQ=1;
nite=0;
tic;
while max(abs(error))>tol | max(abs(errorQ))>tol
nite=nite+1;
Vmod=abs(V);
Vini=V;
ang=angle(V);
Qini=Q;
%calculo las reactivas para los nudos PV
if m>1
for l=2:m
AQ=0;
for k=1:n
AQ=AQ+Ybus(l,k)*V(k);
end
Q(l)=-imag((V(l)')*AQ);
end
22 July 2011 47
%calculo las A para todos los nudos:
for a=1:n
A(a)=(P(a)-i*Q(a))/Ybus(a,a);
end Gauss-Seidel
%calculo los angulos nudos PV:
for l=2:m
Matlab code 3
Aang=0;
for k=1:n
Aang=Aang+B(l,k)*V(k);
end
ang(l)=angle(A(l)/((V(2))')-Aang+B(l,l)*V(l));
end
end
for a=1:n
A(a)=(P(a)-i*Q(a))/Ybus(a,a);
end
%ahora actualizo los voltajes otra vez:
for a=1:n
V(a)=Vmod(a)*exp(i*ang(a));
end
%ahora calculo los nudos PQ
AV=zeros(n,1);
for p=m+1:n
for k=1:n
AV(p)=AV(p)+B(p,k)*V(k);
end
V(p)=A(p)/((V(p))')-AV(p)+B(p,p)*V(p);
end
error=Vini-V;
errorQ=Qini-Q;
errorplot(1,nite)=norm(abs(error));
Vfinal(:,nite)=abs(V);
angfinal(:,nite)=ang*180/pi; %paso a grados
22 July 2011 end 48
Gauss-Seidel
Matlab code 4
%calculos los valores de potencia para el nudo slack:
S1=0;
for t=1:n
S1=S1+(V(1)')*(Ybus(1,t)*V(t));
end
P(1)=real(S1);
Q(1)=-imag(S1);
tiempo=toc;
%alerta por si se sobrepasan valores aceptables:
if max(Vmod)>Vmax | min(Vmod)<Vmin
disp('¡¡SE SOBREPASAN LIMITES TENSIONES!!')
end
if max(Q)>Qmax | min(Q)<Qmin
disp('¡¡SE SOBREPASAN LIMITES REACTIVA!!')
end
ang=(180/pi)*angle(V);
Vmo=abs(V);
%represento los errores por iteración:
plot(1:nite,errorplot,'o');grid on;xlabel('iteración');ylabel('error por iteración');title('evolución error tesión');
%------------------------------------------------------------%
% %
% Realizado por Carlos Ruiz Mora octubre 2006 %
% %
%------------------------------------------------------------%
22 July 2011 49
Gauss-Seidel example
22 July 2011 50
G-S Example
Solution tolerance is set to 0.1 MVA.
ONE TWO
THREE
22 July 2011 51
G-S Example
Data below. Base power is SB=100 MVA:
Bus Voltage (p.u.) Power Lin Impedance (p.u.)
(slack)
1 1.02 1-2 0.02+0.04j
2 1.02 PG=50 MW 1-3 0.02+0.06j
3 - PC=100 MW 2-3 0.02+0.04j

QC=60 MVAr (each)
22 July 2011 52
Solution procedure
1. Data and unknown:
Bus Type Data Unknown
1 Slack V1=1.02 1=0.0 P1 Q1
2 PV V2=1.02 P2=0.5 δ 2 Q2
3 PQ P3=-1.0 Q2=-0.6 δ3 V3
22 July 2011 53
Solution procedure
YBUS
ONE TWO
 15  35 j  10  20 j  5  15 j 
YBUS   10  20 j 30  60 j  20  40 j
 
  5  15 j  20  40 j 25  55 j 
THREE
22 July 2011 54
Solution procedure
3. Voltage magnitude initialization (iteration 0):
PQ bus
V3  V30  1
2  02  0 All buses but the reference one

3  03  0
Vector form:
1.02 0
V 0  1.02 0  0
   
 1  0
22 July 2011 55
Solution procedure
Per iteration:
- PV buses (Q, δ) i=2,...,m (bus 2)
- PQ buses (V, δ) i=m+1,...,n (bus 3)
- Stopping criterios:
a) convergence  Sslack and power flows;
b) no converge  new iteration
22 July 2011 56
Solution procedure
4. PV buses: iteration (r+1) :

bus 2:

Q(2r 1)   ( V2(r ) )* Y21V1( r 1)  ( V2(r ) )* Y22 V2( r )  ( V2(r ) )* Y23 V3( r ) 
P2  jQ(2r 1)
2  angle [ V2 ] A (2r 1) 
Y22
22 July 2011 57
Solution procedure
 A (2r 1) (r ) 
(2r 1) ( r 1)
 angle (r 1) *  B21V1  B23 V3 
( V2 ) 
where
Yik
Bik  is a constant
Yii
22 July 2011 58
Solution procedure
5. Buses PQ iteration (r+1):
bus 3:
A3
V3(r 1)  (r ) *  B31V1(r 1)  B32 V2(r 1)
( V3 )
where
P3  jQ3
A3 
Y33
Are constants for PQ buses
Yik
Bik 
Yii
22 July 2011 59
Solution procedure
6. Stopping criterion:
  10 3
| Vi(r 1)  Vi(r ) |

&  i  2, 3; j  2
| Q(jr 1)  Q(jr ) |
22 July 2011 60
Solution procedure
If convergence:
6.1) Slack power
*
Sslack  P1  jQ1  V1* ( Y11V1  Y12 V2  Y13 V3 )
6.2) Power flows  S12 , S21, S13 , S23 , S31, S32
Sik  Vi ( Vi - V *k )YLik
* *
22 July 2011 61
Solution procedure
6.3) Line losses:
Slosss ,ik  Sik  Ski k, i  1,2,3
Sloss  Sloss ,12  Sloss ,13  Sloss ,23
7. If no coveergence, the procedure continues in Step 4.
22 July 2011 62
Implementation
MATLAB:
22 July 2011 63
Solution
11 iterations needed to attain the solution
Iteration (pu) 1 2 3 ... 10 11

0.5083
P1 ,Q1(slack) - - - ... -
0.0716
P2 0.5 0.5 0.5 ... 0.5 0.5
Q2 0.81 0.4084 0.4696 ... 0.5493 0.5501
P3 -1.0 -1.0 -1.0 ... -1.0 -1.0
Q3 -0.6 -0.6 -0.6 ... -0.6 -0.6
22 July 2011 64
Solution
Iteration 1 2 3 ... 10 11
V1 1.02 1.02 1.02 ... 1.02 1.02
1 (º ) 0 0 0 ... 0 0
V2 1.02 1.02 1.02 ... 1.02 1.02
2 (º ) 0.0675 -0.1596 -0.2885 ... -0.4667 -0.4685
V3 1.0041 1.0042 1.0043 ... 1.0043 1.0043
3 (º ) -0.5746 -0.7336 -0.8278 ... -0.9580 -0.9593
22 July 2011 65
Solution
Errors for |V| & |Q|:
Iteration 1 2 3 ... 10 11
Max. Error V 0.0041 9.68·10-5 5.05·10-5 ... 1.15·10-6 6.74·10-7
Max. Error Q 0.8160 0.4076 0.0612 ... 0.0014 8.11·10-4
22 July 2011 66
Final Solution
Bus P (MW) Q (MVAr ) V (pu)

1 50.83 7.16 1.02  0º
2 50.00 55.01 1.02 -0.4685º
3 -100 -60.00 1.0043  -0.9593º
22 July 2011 67
Checking
Checking using Power-World:
ONE TWO
THREE
22 July 2011 68
Checking
Bus 1 Bus 2 Bus 3

Solution G-S PW G-S PW G-S PW
V (pu) 1.02 1.02 1.02 1.02 1.0043 1.0043
δ(º) 0 0 -0.4685 -0.4710 -0.9593 -0.9612
P (MW) 50.83 50.98 50.00 50.00 -100 -100
Q (MVAr) 7.16 7.10 55.01 55.12 -60 -60
22 July 2011 69
Checking
Variable V δ P Q
Error Max. (%) 0% 0.53 % 0.29% 0.84 %
22 July 2011 70
4. The Newton-Raphson solution
technique
22 July 2011 71
The Newton-Raphson solution technique
Introduction
• One variable
f ( x )  0, f ( x ( 0 ) )  0, f ( x (0 )  x ( 0 ) )  0
(0)
( 0 )  df 
f ( x  x )  0  f ( x )  x  
(0) (0) (0)
 ...
 dx 
f ( x(0) )
x ( 0 )   (0)
; x (1)  x ( 0 )  x ( 0 )
 df 
 
 dx 
f ( x (r ) )
x (r 1)  x (r )  (r )
 df 
 
 dx 
22 July 2011 72
Introduction
• Example
df ( x )
f ( x )  x 2  5 x  4  0;  2x  5
dx
( r 1) ( x ( r ) )2  5 x ( r )  4 ( 0 )
x x (r )
 ; x  0 .5
2x  5
(r )
0 .5 2  5  0 . 5  4
x (1)
 0 .5   0.9375
2  0 .5  5
0.9375 2  5  0.9375  4
x ( 2)
 0.9375   0.9988
2  0.9375  5
0.9988 2  5  0.9988  4
x (3)
 0.9988   1.0000
2  0.9988  5
Fast!
22 July 2011 73
General formulation
• General case
f ( x )  0 : Rn  Rn f ( x ( 0 ) )  0, f ( x ( 0 )  x ( 0 ) )  0
f ( x ( 0 )  x )  f ( x ( 0 ) )  J( 0 )  x ( 0 )  0
x (0)
J  ( 0 ) 1
f ( x(0) )
x (1)  x ( 0 )  x ( 0 )
x (r 1)
x (r )
 
J(r ) 1
f ( x (r ) )
22 July 2011 74
General formulation
 f1 f1 f1 

 x 
 x1   f1  x 2 x n 
 1 
x  f   f2 f2 f2 

x   2 f   2 J   x1 x 2 x n 
   
x  f    
 n  n  fn fn fn 
  
 x1 x 2 x n 
22 July 2011 75
Load flow case
 (2r ) 
 ( r ) 
 3 
   n
(r )  fiP ()  Vi   Vk (Gik cos ik  Bik sinik )
x  (r )
  (r ) 
n
 Vm1 
k 1
n
  fiQ ()  Vi   Vk (Gik sinik  Bik cos ik )

   k 1
 Vn(r )1 
 (r ) 
 Vn 
Pi ()  fiP, Pi  Pi,sp
Qi ()  fiQ, Qi  Qi,sp
 2 ,..., n ; Vm1,..., Vn
22 July 2011 76
Load flow case
Using Taylor:
(r ) (r ) (r ) (r )
 f   f   f   f 
Pi,sp  fiP(r )   iP  (2r )  ...   iP  (nr )   iP  Vm(r)1  ...   iP  Vn(r )
  2   n   Vm1   Vn 
(r ) (r ) (r ) (r )
 f   f   f   f 
Qi,sp  fiQ(r )   iQ  (2r )  ...   iQ  (nr )   iQ  Vm(r)1  ...   iQ  Vn(r )
  2   n   Vm1   Vn 
The increments below should be 0:
Pi(r )  Pi,sp  fiP(r ) ; Q(ir )  Qi,sp  fiQ(r )
22 July 2011 77
Load flow case
Matrix notation:
 P2(r )    f  (r )  f2P 
(r )
   (r )
2

    2P  ... ...     
      2   Vn    
 P2(r )   ... ... ... ...    (r ) 
 (r )     (r ) 
n

Qm1   ... (r ) ... ... ...  Vm1 
     fnQ   fnQ     
(r )
    ... ...     (r ) 

 Qn    2 
(r )   n    Vn 
22 July 2011 78
Jacobian matrix
• PQ buses generate 2 Jacobian rows

corresponding to ΔP and ΔQ
• PV buses generate 1 Jacobian row

corresponding to ΔP
22 July 2011 79
Jacobian elements
• Jacobian dimension
2  NPQ  NPV
NPQ Number of PQ buses
NPV Number of PV buses
22 July 2011 80
Jacobian elements
• Jacobian dimension
P(r )  H(r ) N(r )   (r 1) 

 (r )    (r ) (r ) 
 V (r 1) 
Q   J L   V (r ) 
V (r 1) (r 1)
instead of V
V (r )
for improving computational efficiency
22 July 2011 81
Jacobian elements
 m  k
Pk
Hkm   Vk Vm (Gkm sinkm  Bkm cos km )
m
Pk
Nkm  Vm  Vk Vm (Gkm cos km  Bkm sinkm )
Vm
Qk
Jkm    Vk Vm (Gkm cos km  Bkm sinkm )
m
Qk
Lkm  Vm  Vk Vm (Gkm sinkm  Bkm cos km )
Vm
22 July 2011 82
Jacobian elements
k  m
Pk
Hkk   Qk  Bkk Vk2  Note :
k
Pk
Nkk  Vk  Pk  Gkk Vk2 km  k  m
Vk
Qk
Jkk   Pk  Gkk Vk2 Y km  Gkm  jBkm
k
Qk
Lkk  Vk  Qk  Bkk Vk2
Vk
22 July 2011 83
Solution outline
1. Build YBUS
2. Specify 1  0
Pi i  2,...,n
Qi i  m  1,...,n
Vi i  2,...,m
V1
i , i  2,...,n
3. Initialize 
 V i , i  m  1,...,n
22 July 2011 84
Solution outline
4. Compute Pi(r ) (for PV & PQ Buses), Q(ir ) (for PQ Buses)

5. If Pi(r )  P & Q(ir )  Q
1) P1  jQ1

then compute 2) line flows
3) Stop

else go on
6. Compute submatrices H(r ), N(r ), J(r ), L(r )

22 July 2011 85
Solution outline
7. Solve  (r 1)  H(r ) N(r )  1  P(r ) 

V (r 1)    (r ) (r )   (r ) 
 V (r )   J L  Q 
8. Update
(r 1)  (r )  (r 1) PV & PQ Buses
V (r 1)  V (r )  V (r 1) PV Buses
9. Go to step 4
22 July 2011 86
The Newton-Raphson, Matlab Code
function [V_modulo, V_fase, P, Q, N_iter, T_calculo, Error_p] = nraphson(V_modulo_ini, V_fase_ini, P_ini, Q_ini, Y, npv)
%
% [V_modulo, V_fase, P, Q, N_iter, T_calculo, Error_p] = nraphson(V_modulo_ini, V_fase_ini, P_ini, Q_ini, Y, npv)
%
% Obtencion de indices de nodos: 1 -> SLACK;
% 2,...,M -> PV;
% M+1,...,N -> PQ
n = length(V_modulo_ini);
m = npv + 1;
% Parametros del Metodo

Tol = 0.0001; % Tolerancia del metodo (Perdida de potencia).
Error_p = 1; % Error inicial (A un valor mayor que Tol).
N_iter = 0; % Numero de iteraciones.
% Valores iniciales
V_modulo= V_modulo_ini';
V_fase = V_fase_ini';
P = P_ini';
Q = Q_ini';
j=sqrt(-1);
22 July 2011 87
tic;
while (Error_p > Tol) % Bucle principal (Se permiten 50 iteraciones como mucho).
if (N_iter >50)
error('Demasiadas iteraciones');
break
end
V = V_modulo.*exp(j*V_fase); % Expresion compleja de la tension
S = V.*conj(Y*V); % Expresion compleja de la potencia
DP = P(2:n)-real(S(2:n)); % Incremento de potencia activa (nudos PV y PQ)
DQ = Q(m+1:n)-imag(S(m+1:n)); % Incremento de potencia reactiva (nudos PQ)
PQ = [DP ; DQ];
Error_p = norm(PQ,2); % Error en esa iteracion
DS_DA = diag(V)*conj(Y*j*diag(V)) + diag(conj(Y*V))*j*diag(V);

DS_DV = diag(V)*conj(Y*diag(V./V_modulo)) + diag(conj(Y*V))*diag(V./V_modulo);
% Construccion del Jacobiano

J = [real(DS_DA(2:n , 2:n)) real(DS_DV(2:n , m+1:n))
imag(DS_DA(m+1:n , 2:n)) imag(DS_DV(m+1:n , m+1:n))]
dx=J\PQ; % indices: 1...n-1 = fases en PV y PQ; n...final = modulos en PQ
22 July 2011 88
V_fase (2:n) = V_fase(2:n) + dx(1:n-1); % Actualizamos la fase de las tensiones (nudos PV y PQ)
V_modulo (m+1:n)= V_modulo(m+1:n) + dx(n:end); % Actualizamos el modulo de las tensiones (nudos PQ)
N_iter = N_iter + 1; % Incremento el numero de iteraciones
disp('Pulse una tecla para continuar')

pause
end
P=real(S); % Calculo de la potencia activa
Q=imag(S); % Calculo de la potencia reactiva
V_fase=V_fase*180/pi; % Paso de Radianes a grados
T_calculo=toc;
% **** ENTRADAS ****

%
% V_modulo_ini = Modulo de la tension para comenzar a iterar (conocidos en SLACK y PV).
% V_fase_ini = Fase de la tension para comenzar a iterar (conocido en SLACK).
% P_ini = Potencia activa en los nodos (conocido en PV y PQ).
% Q_ini = Potencia reactica en los nodos (conocido en PQ).
% Y = Matriz de admitancias nodales.
22 July 2011 89
% **** SALIDAS ****
%
% V_modulo = Modulo de la tension en todos los nodos.
% V_fase = Fase de la tension en todos los nodos.
% P = Potencia activa en todos los nodos.
% Q = Potencia reactiva en todos los nodos.
% N_iter = Numero de iteraciones.
% T_calculo = Tiempo de calculo.
% Error_p = Error.
%
%
% **** OBSERVACIONES ****
%
% Los nodos deben estar ordenador asi:
%
% *1 : SLACK.
% * 2...m : PV.
% * m+1...n : PQ.
%
%
% **** FECHA Y AUTOR ****
%
% Laura Laguna - Octubre de 2005
22 July 2011 90
Ejemplo resuelto por el método
de Newton-Raphson
22 July 2011 91
Newton-Raphson Example
• Checking with Matlab and PowerWorld
22 July 2011 92
ONE TWO
Bus Voltage p.u Power

1 1.02 -
THREE
2 1.02 PG=50 MW
3 - PC= 100 MW
QC=60 MVAr
22 July 2011 93
Data:
Line Impedance p.u.

1-2 0.02+0.04j
1-3 0.02+0.06j
2-3 0.02+0.04j each
22 July 2011 94
• 3 buses:
• Bus1: Slack
• Bus 2: PV
• Bus 3: PQ
• Voltage magnitude at bus 3 initialized at 1.02.
• Angles intitialized to zero.
22 July 2011 95
Y bus:
 15.0000 - 35.0000j - 10.0000 + 20.0000j - 5.0000 + 15.0000j 
 
Y   - 10.0000 + 20.0000j 30.0000 - 60.0000j - 20.0000 + 40.0000j 
 - 5.0000 + 15.0000j 25.0000 - 55.0000j 
 - 20.0000 + 40.0000j
Intitialization:
 2  3  0
V3  1.02
22 July 2011 96
Residuals:
3
Pi  Piesp - Vi ∑Vj (Gij cos ij  Bijsen(ij )) i  2,3
j1
3
Qi  Qiesp - Vi ∑Vj (Gijsenij - Bij cos(ij )) i  3
j1
22 July 2011 97
Checking:
P2cal  0.3624·1014   P2   0.5  P2   0.5 

cal
 cal       cal   1 
P
 3    0   P
 3    1  P3 
  
Q3  7.2475·10 
cal 15
Q3   0.6  Q3   0.6
cal
   
No tolerance satisfied: the process continues.
22 July 2011 98
Jacobian:
 H22 H23 N23   62.4240  41.6160  20.4000 

   
J   H32 H33 N33  J    41.6160 57.2220 25.5000 
M   20.8080
 32 M33 L 33 
  26.0100 56.1000 
22 July 2011 99
First iteration:
 
  2  - 0.4557 1.0200  0 
 3   - 0.9406    
V  1.0200 ;    - 0.4557
V   
1.0046  - 0.9406
 3  - 0.0154    
 V3 
22 July 2011 100

Ejemplo por Newton-Raphson
Residuals:
P2cal   0.4958   P2   0.5  0.4958   0.0042 

 cal      P     1  ( 0.9835)    0.0165
P
 3     0.9835  3    
Q3   0.5874
cal
Q3   0.6  ( 0.5874)  0.0126
 
No convergence.
22 July 2011 101

Jacobian for iteration 2:
 61.8860  41.1614  20.0540 

 
J    40.8145 56.0994 24.1371 
 20.8409  26.2162 54.6707 

22 July 2011 102

State variables at iteration 2
  1.0200   0 
 2   - 0.0154     
 3    - 0.0206  V  1.0200 ;    - 0.4710 
V    1.0043   - 0.9612 
 3  - 3.0024·10     
-4
 V3 
22 July 2011 103

Residuals:
P2cal   0.5000   P2   0.5  0.5000   0.0664·10 

5
 cal         5 
P3     1.0000   P3     1  ( 1.0000)     0.5454·10 
Qcal   0.6000 
 Q 
   0.6  ( 0.6000)
   4.9698·10 6 
 3  3 
Tolerance OK.
22 July 2011 104

Jacobian iteration 3:
 61.8860  41.1614  20.0540 

 
J    40.8145 56.0994 24.1371 
 20.8409  26.2162 54.6707 

 
  2 
 - 0.6458·10-5
 1.0200   0 
   
 3   - 0.7534·10-5  V  1.0200 ;    - 0.4710 
V   
1.0043   - 0.96810 
   - 1.1106·10-7 
   
3

 V3
22 July 2011 105

Final power values:
P1   0.5098   Q1   0.0710 

P    0.5000  Q    0.5513 
 2    2  
P3   1.0000 Q3   0.6000
22 July 2011 106

convergence
22 July 2011 107

convergence
22 July 2011 108

convergence
22 July 2011 109

convergence
22 July 2011 110

Power World
5 0 ,9 7 7 M W 5 0 ,0 0 0 M W
7 ,0 9 5 M var 5 5 ,1 2 6 M var
UNO
C ERO
1 ,0 2 0 pu
1 ,0 2 0 pu
0 ,0 0 0 Deg
-0 ,4 7 1 Deg
DO S
1 ,0 0 4 3 pu
-0 ,9 6 1 Deg
100 MW
6 0 M var
22 July 2011 111

Bus V(p.u) δ P Q
0 (PW) 1.02 0 50.9763 7.0955
0 1.02 0 50.9755 7.0954
1 (PW) 1.02 -0.4710 50 55.1256
1 1.02 -0.4710 50.0006 55.1228
2 (PW) 1.0043 -0.9612 -100 -60
2 1.0043 -0.9612 -99.9989 -59.9970
22 July 2011 112

• Largest error below 0.1 MVA.
• More effective technique than Gauss Seidel.
• Convergence is fast (if adequate initialization).
22 July 2011 113

Including tap-changing transformers
22 July 2011 114

Tap-Changing transformer
• A tap changing transformer makes the

admitance matrix dependent on the
transformer parameter t.
• The Jacobian matrix also depends on t.
22 July 2011 115

Admitance matrix
Equivalent circuit:
t:1 Ycc
Ycc i t j
i j
1 t t 1
Ycc 2 Ycc
t t
t 1
22 July 2011 116

Admitance matrix
• General building rules

1. Self admittance of node i, Yii , equals the
algebraic sum of all the admittances
connected to node i
2. Mutual admittance between nodes i and
k, Yii , equals the negative of the sum of
all admittances connecting nodes i and k
3. Yik  Yki
22 July 2011 117

Tap-changing
Example
22 July 2011 118

Tap-changing
Example
1 Z13 3 Z 23 2
Yshunt 13 Yshunt 13 Yshunt 23 Yshunt 23
Ycc
t 4
The tap-changing
transformer
controls voltage 1 t t 1
Ycc 2 Ycc
of bus 4 t t
22 July 2011 119

Tap-changing
Example
1
Y11   Yshunt 13 ; Y12  0 ;...
Z13
1 1 1  t Ycc
... Y33   Yshunt 13   Yshunt 23  Ycc 2 
Z13 Z 23 t t
Ycc
Y34  
t
Ycc t 1
Y44   Ycc  Ycc
t t
It does not depend

22 July 2011
on t! 120
Tap-changing
Example
 Y11 Y12 Y13 Y14 

Y Y22 Y23 Y24 
Y 21 
 Y31 Y32 Y33 ( t ) Y34 ( t )
Y 
 41 Y42 Y43 ( t ) Y44 
22 July 2011 121

Tap-changing transformer
Load flow equations:

n
Pi  Vi   Vk Gik cos ik  Bik sin ik 
k 1
n
Qi  Vi   Vk Gik sin ik  Bik cos ik 
k 1
22 July 2011 122

Taylor Expansion:
(r ) (r ) (r )
 fiP   fiP   fiP 
Pi dato
f (r )
   (r )
 ...     (r )
   Vm 1  ...
(r )
  2   n   Vm1 
iP 2 n
(r )
 fiP 
(r )
 fiP 
...    V (r )
  t  t (r )

n
 n
V  
(r ) (r ) (r )
 f   f   f 
Qidato  fiQ(r )   iQ  (2r )  ...   iQ  (nr )   iQ  Vm(r)1  ...
  2   n   Vm1 
(r ) (r )
 fiQ   fiQ 
...    V (r )
  t  t (r )

n
 n
V  
22 July 2011 123
• The admitance matrix depend on t.
• The Jacobian matrix has a new column.
• The new variable (t) replace the voltage value

at the corresponding PQ bus.
22 July 2011 124

• Increment Δt is divided by t to improve

computational efficiency.
• The Jacobian matrix maintains its # of column

& rows.
• The variable t is considered in the last place.
22 July 2011 125

The Jacobian matrix blocks are:

H(r ) N(r ) C(r ) 
J   (r ) (r ) (r ) 
M L D 
H, N, M & L are the standard blocks.

Submatrix C & D have as many columns as
the # of tap-changing transformers.
22 July 2011 126

Derivatives with respect to t:

fiP n
 Gik Bik 
Ci  t   t  Vi  ∑Vk  cos ik  sinik 
t k 1  t t 
fiQ n
 Gik Bik 
Di  t   t  Vi  ∑Vk  sinik cos ik 
t k 1  t t 
C & D multiplied by t!
22 July 2011 127
Iterative procedure analogous, but:
• If t hit any of its limit, it is fixed to the limit &

the corresponding bus becomes PQ.
• If the procedure converge, fix t at its closest

integer value & continues the iteration with
that t fixed.
22 July 2011 128

Tap-changing
Example
1.- Slack
2.- PV
3.- PQ
4.- PQ
5.- PQV
22 July 2011 129

Tap-changing
Example
Data:
Line impedances: 0.001+j0.05 p.u.

Shunt admitances: j0.05 p.u. (2 per line).
Transformer: 0.9<t<1.1, steps of 0.005; Zcc=j0.1.
22 July 2011 130

Tap-changing
Example
Bus Voltage (pu). Power
1 1.00 Slack
2 1.00 Pg=150MW
3 -- Pc=50MW, Qc=10MVAr
4 -- Pc=0MW, Qc=0MVAr
5 1.00 Pc=100MW,Qc=50MVAr
22 July 2011 131

Tap-changing
Example
Ylínea Ylínea Ylínea
1 2 4 3
Yshunt Yshunt Yshunt

Yshunt Yshunt
Yshunt
Ycc 5
t
1 t t 1
Ycc 2 Ycc
t t
22 July 2011 132

Tap-changing
Example
Admitance matrix:
 Yl  Ys Yl 0 0 0 
 Y 2Yl  2Ys 0 Yl 0 
 l 
 0 0 Yl  Ys Yl 0 
YBUS  Ycc (1 t ) Ycc 
 0 Yl Yl 2Yl  2Ys   Ycc 2 
 t t t 
 0 Ycc
0 0 Ycc 
 t 
22 July 2011 133

Tap-changing
Example
 0.3998  j19.942  0.3998  j19.992 0 0 0 
 0.3998  j19.992 0.7997  j39.884 0  0 .3998  j19 .992 0 
 
Ybus   0 0 0.3998  j19.942  0.3998  j19.992 0 
 
 0  0 .3998  j19.992  0 .3998  j19. 992 0 .7997  j49 . 884 j10 
 0 0 0 j10  j10
Variable initial values:

2  3   4  5  0;
V3  V4  1;
t  1;
22 July 2011 134
Tap-changing
Example
Increment calculations:
 P2   1.5  P2cal   1.5 
 P   cal    0 .5 
 3   0 . 5  P 3   
 P4    P4cal   0 
   cal   
 
 5    1  P5    1 
P 
 Q3    0.1  Qcal    0.1
   3
  
Q 4    Q 4   0 
cal
 Q   0.5  Qcal   0.5

 5  5   
22 July 2011 135

Tap-changing
Example
Jacobian
 39.984 0  19.992 0 0  0.3998 0 

 0 19.992  19.992 0 0.3998  0.3998 0 
 
 19.992  19.992 49.984  10  0.3998 0.7997 0 
J 0 0  10 10 0 0 0 

 0  0.3998 0.3998 0 19.892  19.992 0 
 
 0.3998 0.3998  0.7997 0  19.992 49.754  10 
 0 0 0 0 0  10 10 
22 July 2011 136

Tap-changing
Example
First iteration:
  2 
  
 3  
0
  4   0.0993   1   0 
    
  0 . 0744   1   0 
 5    
 
 V3    0.1744  V  0.9623 ;    0.0993 ; t  0.918
 V3   
 0.968   0.0744 
 V    0 . 0377 
     
 4
  0.032   1   0.1744 
 V4  
 t    0.082 
 t 
22 July 2011 137
Tap-changing
Example
Admitance matrix:
 0.3998  j19.942  0.3998  j19.99 0 0 0 

 0.3998  j19.992 0.7997  j39.884 0  0.3998  j19.992 0 
 
Ybus  0 0 0.3998  j19.94  0.3998  j19.992 0 
 0  0 . 3998  j19 . 992  0 . 399  j19 . 99 0 . 7997  j51 . 750 j10 . 8933 
 
 0 0 0 j10.893  j10 
22 July 2011 138

Tap-changing
Example
Power computation:
 P1   0   Q1    0.05 
P   1.4521  Q   0.5647 
 2    2  
P3    0.4657  Q3    0.1407 
P    Q   
 4  0 . 069   4  0 . 1027 
P5    1.0527  Q5    0.492 
22 July 2011 139

Tap-changing
Example
Power increments:
 P2   1.5  1.4521   0.0479 
 P   0.5  0.4657   0.0343 
 3    
 P4    0.069    0.069 
 P       0.0527 
 5   1  1 . 0527   
 Q3    0.1  0.1407   0.0407 
     
Q 4    0.1027   0.1027 
 Q5    0.5  0.492    0.008 
22 July 2011 140

Tap-changing
Example
Jacobian matrix:
 39.9193 0  19.3273 0 0 1.0523 0 

 0 18.6075  18.6075 0  0.0954  0.8359 0 
 
 19.2698  18.6261 48.3879  10.492 0.0912 0.8183  1.0527 
J 0 0  10.492 10.492 0  1.0527 1.0527 

 0  0.8359 0.8359 0 18.3261  18.6075 0 
 
 1.8242  0.0912  0.6803  1.0527  18.6261 48.5934  11.746 
 0 0 1.0527  1.0527 0  10.492 10.492 
22 July 2011 141

Tap-changing
Example
Second iteration
  2 
    0.0002 
 3 
  4    0.005   1   0 
    
  0 . 003   1    0.0002 
 5    
  
 3   0.0021  V  0.9607 ;    0.1043 ; t  0.9144
V
 V3   
0.9644   0.0744 
 V    0 . 0016 
     
 4
  0.0037   1   0.1723 
 V4  
 t   0.0039 
 t 
22 July 2011 142
Tap-changing
Example
Admitance matrix:
 0.3998  j19.942  0.3998  j19.992 0 0 0 

 0.3998  j19.992 0.7997  j39.884 0  0 .3998  j19 .992 0 
 
Ybus  0 0 0.3998  j19.942  0.3998  j19.992 0 
 
 0  0 .3998  j19.992  0.3998  j19.992 0 . 7997  j51 . 8447 j10.9365 
 0 0 0 j10.9365  j10 
22 July 2011 143

Tap-changing
Example
Power calculations:
 P1   0.0031   Q1    0.0501
P   1.4998  Q   0.6391 
 2    2  
P3    0.4998  Q3    0.1000 
P    Q   
 4  0 . 000   4  0 . 0006 
P5    1.0000  Q5   0.4999 
22 July 2011 144

Tap-changing
Example
Power increments:
 P2   1.5  1.4998   0.2065 
 P   0.5  0.4998   0.1969 
 3    
 P4   0.000   0.0057 
 P       0.0258   10 3
 5   1  1 . 0000   
 Q3    0.1  0.1000   0.0339 
     
Q
 4   0 . 0006    0 . 6375 
 Q5   0.5  0.4999   0.1218 
22 July 2011 145

Tap-changing
Example
Tap to the closest feasible value:
t  0.9144  t  0.915
Admitance matrix:
 0.3998  j19.942  0.3998  j19.992 0 0 0 
 0.3998  j19.992 0.7997  j 39.884 0  0.3998  j19.992 0 

Ybus  0 0 0.3998  j19.942  0.3998  j19.992 0 
 
 0  0.3998  j19.992  0.3998  j19.992 0.7997  j51.8282 j10.929
 0 0 0 j10.929  j10 
22 July 2011 146

Tap-changing
Example
Power calculation:
 P1   0.0031   Q1    0.0501
P   1.4998  Q   0.6391 
 2    2  
P3    0.4998  Q3    0.1000 
P    Q   
 4  0 . 0007   4   0 . 0075 
P5   0.9993  Q5   0.4926 
22 July 2011 147

Tap-changing
Example
Power increment calculations:
  2 
 
 P2   1.5  1.4998   0.0002   3 
 P   0.5  0.4998    0.0002    
 3      4
 P4    0.0007   0.0007   5 
 P       0.0007  ;  V3 
 5   1  0 . 9993    V 
 Q3    0.1  0.1000   0.0000   3 
       V4 
Q 4   0.0075   0.0075   V4 
 Q5   0.5  0.4926   0.0074   V 
5
V 
 5 
22 July 2011 148
Tap-changing
Example
Jacobian:
 39.2449 0  19.253 0 0 1.103 0 

 0 18.5072  18.5072 0  0.1307  0.8689 0 
 
 19.1935  18.5271 48.2132  10.4926 0.1282 0.7431 0.9993 
J 0 0  10.4926 10.4926 0  0.9993  0.9993 

 0  0.8689 0.8689 0 18.3071  18.5072 0 
 
 1.872  0.1282  0.7445  0.9993  18.5271 48.1983  10.4926 
 0 0 0.9993  0.9993 0  10.4926 9.5074 
22 July 2011 149

Tap-changing
Example
Final values
  2 
  
 3   0.0007 
  4   0.0289   1   0 
    
  0 . 0152   1    0.0002 
 5     
V   
 3    0.1699  10  V  0.9607 ;    0.1043 
3
 V3   
0.9644   0.0744 
 V    0 . 0552 
     
 4
  0.0558   0.9991  0.1725 
 4 
V 
 V5    0.8522 
V 
 5 
22 July 2011 150
Tap-changing
Example
Final power values:
 P1   0.0031   Q1    0.0501
P   1.5000  Q   0.6402 
 2    2  
P3    0.5000  Q3    0.1000 
P    Q   
 4  0 . 0000   4  0 . 000 
P5    1.0000  Q5   0.5000 
22 July 2011 151

Power World
22 July 2011 152

5. Fast decoupled AC load flow
22 July 2011 153

Fast decoupled AC load flow
Two simplifications:
– Do not build Jacobian at each iteration (small error
introduced, then, the procedure needs more
iterations to reach the solution)
– Decoupling between P-δ and Q-V (not

recommended in system highly loaded and/or with
low voltage levels)
22 July 2011 154

 P(r )  H(r ) N(r )    

( r 1)
 (r )    (r )  (r 1)
( r )  V

Q   J L   V (r ) 
Assume:
i) V i  1 .0 i
ii) Gik  Bik i, k
iii) cos(i  k )  1.0
sin(i  k )  0.0
iv) Qk  Bkk
22 July 2011 155
We have:
H  B~1 , L  B~2 , N  0 , J  0
 ~
 ik  Bik 
  
~ ~
B1 , B2  ~
B
  Elements of YBUS
Bii  Bii 
22 July 2011 156

 P(r )  B~   ( r 1) 
0 
 (r )   
1
~   (r 1) 
Q   0 B 2  V 
~
 
P(r )  B1 (r 1)  (1)
~
 
Q(r )  B 2 V (r 1)  ( 2)
     
( r 1) ~ 1
 B1 P(r )
V     
( r 1) ~ 1
 B 2 Q(r )
Use Newton-Raphson Iteration

22 July 2011 157
Fast decoupled load flow. Flow diagram
DATA INPUT
Calculate Delta P
Real Reactive YES

power YES power
coverged coverged
? ?
NO NO
Solve (1) and update angles
Calculate Delta Q
NO
Reactive YES Real YES

power power
coverged coverged
? ?
NO NO
Solve (2) and update V
OUTPUT RESULTS
22 July 2011 158
Fast decoupled load flow. Matlab Code
function [V_modulo, V_fase, P, Q, N_iter, T_calculo, Error_p] = desacoplado(V_modulo_ini, V_fase_ini, P_ini, Q_ini, Y,
npv)
%
% [V_modulo, V_fase, P, Q, N_iter, T_calculo, Error_p] = desacoplado(V_modulo_ini, V_fase_ini, P_ini, Q_ini, Y, npv)
%
% Obtencion de indices de nodos: 1 -> SLACK;
% 2,...,M -> PV;
% M+1,...,N -> PQ
n = length(V_modulo_ini);
m = npv + 1;
% Parametros del Metodo
Tol = 0.0001; % Tolerancia del metodo (Perdida de potencia).
Error_p = 1; % Error inicial (A un valor mayor que Tol).
N_iter = 0; % Numero de iteraciones.
% Valores iniciales
V_modulo= V_modulo_ini';
V_fase = V_fase_ini';
P = P_ini';
Q = Q_ini';
DS_DA = diag(V)*conj(Y*j*diag(V)) + diag(conj(Y*V))*j*diag(V);

DS_DV = diag(V)*conj(Y*diag(V./V_modulo)) + diag(conj(Y*V))*diag(V./V_modulo);
22 July 2011 159

%J = [real(DS_DA(2:n , 2:n)) real(DS_DV(2:n , m+1:n)) % Construccion del Jacobiano
% imag(DS_DA(m+1:n , 2:n)) imag(DS_DV(m+1:n , m+1:n))];
H = imag(-Y(2:end , 2:end))
L = imag(-Y(m+1:end,m+1:end))
j=sqrt(-1);
tic;
while (Error_p > Tol) % Bucle principal (Se permiten 50 iteraciones como mucho).
if (N_iter >50)
error('Demasiadas iteraciones');
break
end

S = V.*conj(Y*V); % Expresion compleja de la potencia
DP = P(2:n)-real(S(2:n)); % Incremento de potencia activa (nudos PV y PQ)
DQ = Q(m+1:n)-imag(S(m+1:n)); % Incremento de potencia reactiva (nudos PQ)
Dfase = H\DP;
Dmodulo = L\DQ;
PQ = [DP ; DQ];
Error_p = norm(PQ,2); % Error en esa iteracion
22 July 2011 160

V_fase (2:n) = V_fase(2:n) + Dfase; % Actualizamos la fase de las tensiones (nudos PV y PQ)
V_modulo (m+1:n)= V_modulo(m+1:n) + Dmodulo; % Actualizamos el modulo de las tensiones (nudos PQ)
N_iter = N_iter + 1; % Incremento el numero de iteraciones
% disp('Pulse una tecla para continuar')

% pause
end
P=real(S); % Calculo de la potencia activa

Q=imag(S); % Calculo de la potencia reactiva
V_fase=V_fase*180/pi; % Paso de Radianes a grados
T_calculo=toc;
% **** ENTRADAS ****

%
% V_modulo_ini = Modulo de la tension para comenzar a iterar (conocidos en SLACK y PV).
% V_fase_ini = Fase de la tension para comenzar a iterar (conocido en SLACK).
% P_ini = Potencia activa en los nodos (conocido en PV y PQ).
% Q_ini = Potencia reactica en los nodos (conocido en PQ).
% Y = Matriz de admitancias nodales.
22 July 2011 161

% **** SALIDAS ****
%
% V_modulo = Modulo de la tension en todos los nodos.
% V_fase = Fase de la tension en todos los nodos.
% P = Potencia activa en todos los nodos.
% Q = Potencia reactiva en todos los nodos.
% N_iter = Numero de iteraciones.
% T_calculo = Tiempo de calculo.
% Error_p = Error.
%
%
% **** OBSERVACIONES ****
%
% Los nodos deben estar ordenador asi:
%
% *1 : SLACK.
% * 2...m : PV.
% * m+1...n : PQ.
%
%
% **** FECHA Y AUTOR ****
%
% Laura Laguna - Nobiembre de 2005
22 July 2011 162

Fast decoupled load flow: Exaple
• Tolerance 0.1 MVA.
ONE TWO
THREE
22 July 2011 163

• Power base SB = 100MVA

Bus Voltage p.u. Power
1 1.02 -
2 1.02 PG=50 MW
3 - PC=100 MW
QC=60 MVAr
Line Impedance p.u.

1-2 0.02+0.04j
1-3 0.02+0.06j
2-3 0.02+0.04j
(cada una)
22 July 2011 164

• Admitance matrix
 15  35 j  10  20 j  5  15 j 
Ybus   10  20 j 30  60 j  20  40 j
 
  5  15 j  20  40 j 25  55 j 
22 July 2011 165

• Data and unknown:

Bus Type Data Unknown
Slack and V1 = 1.02
1 P1, Q1
reference δ1=0.0
P2 =0.5
2 PV δ2, Q2
V2 = 1.02
P3 = -1.0
3 PQ δ3, V3
Q3 = -0.6
• Inicialization:
δ2 = δ3 = 0.0 ; V3 = 1.02 p.u.
22 July 2011 166

• Flow diagram
Vi(0) , δi(0), Piesp, Qiesp Compute: Delta P
Compute: Pical, Qical

Yes Yes
Convergence P? Convergence Q?
NO
Solve subproblem 1 and update angles NO
Compute: Delta Q
Yes Yes
Convergence Q? Convergence P?
NO
NO
Solve subproblem 2 and update voltages
Final results
22 July 2011 167

 60  40 
Compute H ; L  55 
  40 55 
Calculate P and Q:
P2cal  0.3624·1014   P2   0.5  P2   0.5 

cal
 cal       cal   1 
P
 3    0   P
 3    1  P3 
  
Q3  7.2475·10 
cal 15
Q3   0.6  Q3   0.6
cal
   
No convergence
22 July 2011 168

• First iteration:
 
  2  - 0.4213 1.0200   0 
 3    - 1.3481    
V  1.0200 ;    - 0.4213 
V   
 3  - 0.0109  1.0091   - 1.3481 
   
 V3 
22 July 2011 169

Residuals:
P2cal   0.7385   P2   0.5  P2   0.2385

cal
 cal      P     1  Pcal    0.3003 

P
 3     1. 3003  3  3   
Qcal           cal 
 0.4583
 3   0 .1417  3 
Q 0.6 Q 3 
Error = 0.5976 → no convergence
22 July 2011 170

• System state (second iteration)

 
  2  - 0.0372 1.0200   0 
 3    0.2857     
V    V  1.0200 ;    - 0.4585 
 3  - 0.0083 1.0008   - 1.0264 
 V3     
22 July 2011 171

•Residuals:
P2cal   0.6578   P2   0.5  P2   0.1578

cal
 cal      P     1  Pcal    0.1936 

P
 3     1.1936  3  3   
Q3   0.7444
cal
Q3   0.6  Q3   0.1444 
cal
 
•Error = 0.2885 → no convergence
22 July 2011 172

• System state (third iteration):

 
  2  - 0.0315 1.0200   0 
 3    0.1788     
V  1.0200 ;    - 0.4900 
V   
 3   0.0026  1.0034   - 0.8836 
 V3     
• The process continues.
22 July 2011 173

•After 9 iterations:
P2cal   0.5006   P2   0.5  P2   0.5954·10 

cal 3
 cal      P     1  Pcal    0.7006·103 

P
 3     1. 0007   3  3   
Qcal           cal    3 
 3   0 . 5992  3 
Q 0. 6 Q 3   8.3346·10 
•Error = 0.0012 → convergence attained
22 July 2011 174

• System state at iteration 10:
 
  2 
 - 0.1592·10-3
 1.0200   0 
 3    0.6141·10-3     
V    V  1.0200 ;    - 0.4710 
   - 1.5154·10-5 
1.0043   - 0.9614 
3
    
 V3
22 July 2011 175

•Residuals:
P2cal   0.5003   P2   0.5  P2   0.2863·10 

cal 3
 cal       cal   3 
P3     1.0004   P3     1  P3    0.3516·10 
Qcal           cal   4 
 3   0. 6003  3 
Q 0.6 Q 3   3.3349·10 
•Error = 5.6285·10-4 → convergence attained
22 July 2011 176

• State at iteration 11:
 
  2 
 - 0.0567·10-3
 1.0200   0 
 3    0.3251·10-3     
V  1.0200 ;    - 0.4711
V   
1.0043   - 0.9611
   6.0634·10-6 
   
3

 V3
 0.5098   0.0711 
   
P   0.5003  Q   0.5515 
  1.0004    0.6003 
   
22 July 2011 177
convergence
22 July 2011 178

convergence
22 July 2011 179

convergence
22 July 2011 180

convergence
22 July 2011 181

6. Variable limits
22 July 2011 182

Variable limits
Physical considerations
1. Voltage magnitudes out of limits

a PQ BUS does not meet
v min
i  v i  v max
i
Action: Warning! voltage problem
2. Flow magnitudes out of limits

A line does not meet
- S max ≤ S ≤ S max
ij ij ij
Action: Warning! overloading of lines problem
22 July 2011 183

Variable limits
Physical Considerations
3. Reactive Power out of limits

A PV BUS does not meet
min max
Qi  Qi  Qi
Action: Wrong Formulation!
Specified voltage cannot be attained
Formulate the problem properly
22 July 2011 184

Variable limits
Computational Considerations
Changing PV to PQ
no yes
Q hits
limit?
no Yes Change to
Q=Qmax ? QG>Qmax? P-Q
Q=Qmax
Yes No
yes
Change to P-V Yes Change to
V >Vdato? V=Vdato P-Q
QG<Qmin?
Q=Qmin
No No
Stay as
P-Q Stay as
P-V
Yes Change to P-V
V <Vdato? V=Vdato
No
Stay as
P-Q
22 July 2011 185

7. DC load flow
22 July 2011 186

DC load flow
• Approximate solution.
• Two simplifications:
– In network model: do not consider series
resistences and shunt admittances
– Assume Vi=1 at all buses
22 July 2011 187

DC load flow
Approximate analytical solution
V V
i j
P   sin
ij X ij
ij
2
V V V
i i j
Q  -  cos 
ij X X ij
ij ij
Assume
i) Vi ≈ 1.0 ∀i
ii) sinij ≈ ij
iii) cos ij ≈ 1.0
22 July 2011 188
DC load flow
Approximate analytical solution
ij
Pij   Bijij  Biji  Bij j
Xij
 
n
 n
 n
Pi   Pij   Bij  i   Bij j
j1 j1 j1
j i 
 ji  j i
P  B´
where n
Bii'   Bij
j1
ji
22 July 2011
Bij'   Bij 189
DC load flow
Continuation
Vi  V j
Qij   Bij ( V i  V j )  Bij V i  Bij V j
Xij
n
n  n
 
Qi   Qij   Bij V i   Bij V j
j1
 j1  j1
ji  
 ji  ji
Q  B' V 
22 July 2011 190

DC load flow
Continuation
Solution
P  B'  PQ & PV Buses
Q  B' V  PQ Buses
 1   V1  P2  Qm1 
 
     ; V      ; P     ; Q    
n   Vn  Pn   Qn 
 
 ' n

 ii  ij 
B  B
 
B 
' j1
ji  ; B ij 
1
Xij
 ' 
Bij  Bij 
22 July 2011 191
DC Power flow: example
22 July 2011 192

ONE TWO
THREE
22 July 2011 193

Data
0 1.02 -
1 1.02 PG=50MW
2 - PC=100MW, QC=60MVAr
Line Impedance p.u.

0-1 0.02+0.04j
0-2 0.02+0.06j
1-2 0.02+0.04j (both)
22 July 2011 194

 41.67  25  16.67 
B    25 75  50 
 
 16.67  50 66.67 
 P0   41.67  25  16.67   0 
 0.5     25 75  50   1 
    
 1.0  16.67  50 66.67   2 
 Q0   41.67  25  16.67  1.02 
 Q     25 75  50  1.02 
 1    
 0.6  16.67  50 66.67   V2 
22 July 2011 195

 P0   41.67  25  16.67  0

     
 0.5     25 75  50   1 
  1.0    16.67  50 66.67   
     2
22 July 2011 196

Po= -25δ1 - 16.67δ2

0.5= +75δ1 - 50δ2
-1.0= -50δ1 + 66.67δ2
Q0= 41.67*1.02 - 25*1.02 - 16.67V2

Q1= -25*1.02 + 75*1.02 - 50V2
-0.6= -16.67*1.02 - 50*1.02 + 66.67V2
22 July 2011 197

Solution:
P0 0.5p.u.=50MW
Q0 0.15p.u.=15MVAr
Q1 0.45p.u.=45MVAr
δ1 -1.1459º
δ2 -0.3819º
V2 1.011p.u.
22 July 2011 198

PowerWorld comparison:
Var DC PowerWorld DC PowerWorld G-S

P0 0.50p.u. 0.50p.u. 0.509p.u.
Q0 0.15p.u. 0.00p.u. 0.07p.u.
Q1 0.45p.u. 0.00p.u. 0.55p.u.
δ1 -0.3819 -0.3820 -0.47
δ2 -1.1459 -1.1459 -0.96
V2 1.0056p.u. 1.0000p.u. 1.0043p.u.
22 July 2011 199

0,00 MW 50,00 MW
0,00 Mvar 0,00 Mvar
CERO
UNO
DOS
TRES
100,00 MW
22 July 2011 60,00 Mvar 200
Data.
0 1.02 -
1 1.02 PG=50MW
Line Impedance p.u.

0-1 0.02+0.04j
0-2 0.02+0.06j
1-2 0.02+0.04 (both)
2-3 0.1j
22 July 2011 201

 41.67  25  16.67 0 
  25 75  50 0 
B 
 16.67  50 76.67  10 
 0  
 0 10 10 
 P0   41.67  25  16.67 0  0 
 0.5    25 75  50 0   1 
   
 0.0   16.67  50 76.67  10   2 
  1 .0   0  
   0  10 10  3 
 Q0   41.67  25  16.67 0  1.02
 Q    25 75  50 0  1.02
 1   
 0.0   16.67  50 76.67  10   V2 
  0 .6   0  
   0  10 10   V3 
22 July 2011 202
P0 = -25δ1 -16.67δ2-0*δ3
0.5 = 75δ1 - 50δ2-0*δ3
0.0 = -50δ1 + 76.67δ2-10δ3
-1.0 = -0*δ1 - 10δ2+10δ3
Q0 = 41.67*1.02 - 25*1.02 - 16.67V2 + 0*V3

Q1 = -25*1.02 + 75*1.02 - 50V2 + 0*V3
0.0 = -16.67*1.02 - 50*1.02 + 76.67V2-10*V3
-0.6 = 0*1.02-0*1.02 - 10*V2 + 10*V3
22 July 2011 203

PowerWorld comparison:
Var DC PowerWorld G-S PowerWorld DC

P0 0.50p.u. 0.51p.u. 0.50p.u.
Q0 0.00p.u. 0.11p.u. 0.00p.u.
Q1 0.00p.u. 0.67p.u. 0.00p.u.
δ1 -0.3819º -0.48º -0.382º
δ2 -1.1459º -0.91º -1.1459º
δ3 -6.8755º -7.05º -6.8755º
V2 1.011p.u. 1.002p.u. 1.000p.u.
V3 0.951p.u. 0.932p.u. 1.000p.u.
22 July 2011 204
8. Comparison of load flow
solution methods
22 July 2011 205

Comparison of load flow methods
1. Gauss-Seidel (G-S)
 Simple technique
 Iteration time increases linearly with the
number of buses. Lower iteration time than N-
R. Seven times faster in large systems
 Linear rate of convergence. Many iterations
required for getting close to the solution
 Number of iteration increases with the
number of buses
22 July 2011 206

(Continuation)
2. Newton-Raphson (N-R)
 Widely used
 Iteration-time increases linearly with the number of
buses
 Quadratic rate of convergency. A few iterations for
getting close to the solution
 Number of iterations independent of the number of
buses of the system
 The Jacobian is a very sparse matrix
 Method non-sensitive to slack bus choice and the
presence of series capacitors
 Sensitive to initial solution
22 July 2011 207
(Continuation)
3. AC decoupled
 B has to be computed and factorized only once

~
It requires more iterations than Newton-

Raphson method
Iteration time is 5 times lower than Newton-
Raphson´s iteration time
Useful for analyzing topology changes because B
~
can be easily modified
Used in planning and contigency analyses
22 July 2011 208

(Continuation)
4. DC Decoupled
 Analytical, approximate and non-iterative method

 Good approximation for  , not that good
approximation for V 
 Used in reliability analyses
 Used in optimal pricing calculations
 Good for getting an initial point
22 July 2011 209

01 LoadFlow R1 PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

01 LoadFlow R1 PDF

Uploaded by

Copyright:

Available Formats

LOAD FLOW

Univ. Castilla – La Mancha

3. The Gauss-Seidel solution technique

4. The Newton-Raphson solution technique

5. Fast decoupled AC load flow

1. A. Gómez Expósito, A. J. Conejo, C.

2. A. R. Bergen, V. Vittal. “Power Systems

• Most used tool in steady state power system

• Knowning the demand and/or generation of

• The problem is described throught a non-

• Need of iterative solution techniques

• Solution technique: accuracy vs. computing

We want to find out the relationship between

Using Kirchhoff laws:

• Complex power injected in each bus:

Therefore, the no lineal equations for the 2

• Two bus case

2. Mutual admittance between nodes i and

(>90% for more than 100 buses)

Shunt element YG  j 0.01 (two per line)

 j19.98 j10 j10 

• 2n equations (static load flow equations)

Vi, i,Pi, Qi ; i  1,...,n

• If 2n variables are specified, the other 2n are

Pi known ( PDi known, PGi zero)

Pi known ( PGi specified, PDi known)

Qi unknown ( QGi unknown, QDi known)

3. Slack bus, generator with large capacity.

 Voltage magnitude Vi min  Vi  Vi max

 Power angle i  k  i  k max (every existing line)

PGi,min  PGi  PGi,max

At iteration (r+1) and bus i, the available values of

 A (r 1) i1 (r 1) n (r ) 

(more on this below)

Vi(r 1)  Vi(r )   ; i  2,...,n

6.1) Slack bus power (after convergence)

6.2) Compute line currents (after convergence)

Sik Iik 0 Iki0 Sik

Sik Iik 0 Iki0 Sik

6.3) Compute losses (after convergence)

• Acceleration factor (in order to decrease

plot(1:nite,errorplot,'o');grid on;xlabel('iteración');ylabel('error por iteración');title('evolución error tesión');

Data below. Base power is SB=100 MVA:

Bus Voltage (p.u.) Power Lin Impedance (p.u.)

2 1.02 PG=50 MW 1-3 0.02+0.06j

3 - PC=100 MW 2-3 0.02+0.04j

Bus Type Data Unknown

1 Slack V1=1.02 1=0.0 P1 Q1

3. Voltage magnitude initialization (iteration 0):

2  02  0 All buses but the reference one

- PV buses (Q, δ) i=2,...,m (bus 2)

- PQ buses (V, δ) i=m+1,...,n (bus 3)

4. PV buses: iteration (r+1) :

| Vi(r 1)  Vi(r ) |

6.1) Slack power

6.2) Power flows  S12 , S21, S13 , S23 , S31, S32

Slosss ,ik  Sik  Ski k, i  1,2,3

Sloss  Sloss ,12  Sloss ,13  Sloss ,23

7. If no coveergence, the procedure continues in Step 4.

Iteration (pu) 1 2 3 ... 10 11

Q2 0.81 0.4084 0.4696 ... 0.5493 0.5501

P3 -1.0 -1.0 -1.0 ... -1.0 -1.0

Q3 -0.6 -0.6 -0.6 ... -0.6 -0.6

DS_DA = diag(V)conj(Yjdiag(V)) + diag(conj(YV))jdiag(V);

% ENTRADAS