Professional Documents
Culture Documents
Introduction - 1
Contents
Introduction - 2
Background (I)
A classical problem of circuit theory is to find all branch currents and all node voltages
of an assigned circuit.
Typical input data are generator voltages as well as the impedances of all branches.
If all impedances are constant, the resulting set of equations that describe the circuit is
linear.
Example 1 (I)
i2
jx12
jx13
jx23
+
3
v1
v2
+
v3
i3
z3
Example 1 (II)
v1 v2
v1 v3
+
i1
jx12
jx13
v2 v3
v2 v1
+
i2
jx12
jx23
v3 v1
v3 v2
+
i3
jx13
jx23
(1)
Example 1 (III)
In vector form:
i1
0
v1
i2 = Y + I 3 0 v2 = Y tot v
0
1/
z3
v3
(2)
Example 1 (IV)
1/jx12 + 1/jx13
1/jx12
1/jx13
Y =
1/jx12
1/jx12 + 1/jx23
1/jx23
1/jx13
1/jx23
1/jx13 + 1/jx23
(3)
Background (II)
The power flow problem is conceptually the same problem as solving a steady-state ac
circuit.
Example 2 (I)
s2
jx
12
v1
v2
jx
23
jx
13
v3
3
s3
Example 2 (II)
0 =
0 =
0 =
v2 v1
sin(2 1 ) p2
x12
v3 v2
v3 v1
sin(3 1 ) +
sin(3 2 ) p3
x13
x23
v32
v32
v3 v1
v3 v2
+
cos(3 1 )
cos(3 2 ) q3
x13
x23
x13
x23
(4)
Loads at high voltage level are modelled as constant PQ due to under-load tap
changers.
Observe that one bus has to be the phase angel reference. Typically, one generator is
used as slack bus.
(5)
which leads to write the power flow problem as the complex power injections at buses:
= V i = V Y v
s
where V
(6)
= diag(
v1 , v2 , . . . , vnb ) and nb is the number of network buses.
Rewriting (6):
V Y v
0=s
(7)
0 = g(y)
(8)
Variables
Data
Slack generator
p, q
v,
PV generator
q,
p, v
PQ load
v,
p, q
The origins of the formulation of the power flow problem and the solution based on the
Newtons method date back to the late sixties.
Since then, a huge variety of studies have been presented about the solution of the
power flow problem, addressing:
Starting initial guess
Computational efficiency
Ill-conditioned cases
Robustness
Multiple solutions
Unsolvable cases
It is relevant to classify the power flow problems into the following categories:
Well-conditioned case
Ill-conditioned case
Bifurcation point
Non-physical solution
Unsolvable case
Well-conditioned case
The power flow solution exists and is reachable using a flat initial guess (e.g., all load
voltage magnitudes equal to 1 and all bus voltage angles equal to 0) and a standard
Newtons method.
Ill-conditioned case
The solution of the power flow problem does exist, but standard solvers fail to get this
solution starting from a flat initial guess.
This situation is due to the fact that the region of attraction of the power flow solution is
narrow or far from the initial guess.
In this case, the failure of standard power flow procedure is due to the instability of the
numerical method, not of the power flow equations.
Robust power flow methods have proved to be efficacious for solving ill-conditioned
cases.
Bifurcation Point
The solution of the power flow exists but it is either a saddle-node bifurcation or a
limit-induced bifurcation.
Saddle-node bifurcations are associated with the maximum loading condition of a
system. The solution cannot be obtained using standard or robust power flow
methods, since the power flow Jacobian matrix is singular at the solution point.
Limited-induced bifurcations are associated with a physical limit of the system, such
as the shortage of generator reactive power. The solution point is typically a
well-conditioned case and does not show convergence issues.
Non-physical Solution
Solutions that cannot be accepted since some variable is out of its technical limits.
Typically these solution are characterized by very low voltage levels.
These solutions are also known as: extraneous, false, lower, or unstable.
Unsolvable Case
The i-th iteration of the classical Newtons method for (8) is as follows:
where g (i)
y (i)
1 (i)
]
g
[g (i)
y
y (i+1)
y (i) + y (i)
(9)
(i)
The geometrical interpretation of the Newtons method is well-known. For the actual
value y (i) , one computes the tangent of g (i) as:
(i)
(y) = g (i) + g (i)
y (y y )
(10)
g(y)
g(y)
gy
gy
gy
y
y
(b)
(a)
g(y)
g(y)
gy
gy
(c)
gy
gy
(d)
There are idiosyncratic cases for which the Newtons technique fails to converge.
A variety of robust variations of the basic Newtons method have been proposed in the
literature for solving ill-conditioned cases.
The majority of these techniques mainly consist in modifying the first equation of (9) as
follows:
1 (i)
y (i) = [g (i)
g
y ]
(11)
where is a factor that improves the convergence properties of the iterative process.
It is important not to confuse ill-conditioned cases with those that are unsolvable since
the solution does not exist.
Robust solvers are useful in case of ill-conditioned systems but do not generally work
well for unsolvable cases.
Unsolvable cases are better tackled using the continuation power flow technique.
g(y)
gy
gy
gy
gy
(a)
y
(b)
One of the most relevant drawbacks of the Newtons method is the need of factorizing
the full Jacobian matrix at each iteration.
The computational effort can be reduced to N 1.5 if using sparse matrices techniques,
which allows saving a considerable time for large systems (e.g., thousands of buses).
However, the Jacobian matrix factorization remains the most critical issue of the
Newtons method (about 85% of the total CPU time for networks with thousands of
buses).
The GMRES is a particular case of Krylovs subspace. The main issue is to properly
pre-conditioning the Jacobian matrix.
Dishonest methods compute the Jacobian matrix factorization only the first one or two
iterations and then use the previous factorization for the remaining iterations.
g(y)
gy
gy
y
y
(a)
(b)
The FDPF requires much more iterations than the NR method but proved to be more
robust.
Newton
Jacobi
Gauss-Seidel
Bus #
Iter. #
time [s]
Iter. #
time [s]
Iter. #
time [s]
14
0.0050
76
0.0217
56
0.0288
118
0.0287
580
0.505
388
2.738
1228
0.210
454
5.120
224
112.4
11856
3.15
340
399.0
173
9112
GMRES
Dishonest
FDPF
Bus #
Iter. #
time [s]
Iter. #
time [s]
Iter. #
time [s]
14
0.4339
0.0040
0.0053
118
53.53
15
0.0183
0.0117
1228
n.a.
n.a.
26
0.207
12
0.160
11856
n.a.
n.a.
10
3.820
5.174
Example 3 (I)
2
z = 0.01 + j0.1 pu
v1 = 1.0 + j0 pu
s2 = 0.9 + j0.6 pu
Example 3 (II)
Since the power flow equations are basically quadratic equations, the number of real
solutions increases as the bus number increases.
For the two-bus system of Example 3, the solutions are two, both real.
For big networks, the number of real solutions is unkown a priori.
A simple method to find all solutions is using the continuation power flow (CPF)
analysis.
However, to use the CPF, one solution has to be found before being able to apply such
method.
7
4
Transformer Equivalent
G
1
Compensators
Synchronous
Generators
Three Winding
G
12
11
13
Example 4 (I)
10
14
Example 4 (II)
Bus
Sol. #1
Sol. #2
Sol. #3
Sol. #4
vh
vh
vh
vh
1.0600
0.0000
1.0600
0.0000
1.0600
0.0000
1.0600
0.0000
1.0450
-0.0869
1.0450
-0.2329
1.0450
-0.2291
1.0450
-0.1506
1.0100
-0.2220
1.0100
-0.5447
1.0100
-0.5578
1.0100
-0.3439
1.0186
-0.1802
0.6202
-0.4833
0.5588
-0.4720
0.9306
-0.3199
1.0203
-0.1533
0.5340
-0.3756
0.4903
-0.2834
0.9436
-0.2966
1.0700
-0.2482
1.0700
-2.8782
1.0700
-3.1994
1.0700
-0.6762
1.0620
-0.2333
0.6120
-1.0675
0.5756
-1.4094
0.8851
-0.4953
1.0900
-0.2333
1.0900
-1.0675
1.0900
-1.4094
1.0900
-0.4953
1.0564
-0.2609
0.4017
-1.5433
0.4500
-1.9780
0.7352
-0.6144
10
1.0513
-0.2636
0.4077
-2.0123
0.4620
-2.3752
0.7841
-0.6338
11
1.0571
-0.2582
0.6756
-2.6346
0.7043
-2.9451
0.9198
-0.6576
12
1.0552
-0.2631
0.9427
-2.9136
0.9860
-3.1992
0.9389
-0.7138
13
1.0504
-0.2646
0.8225
-2.8861
0.9224
-3.1533
0.8142
-0.6870
14
1.0358
-0.2799
-0.0628
-0.0597
0.5177
-2.7767
-0.0356
1.7273
Example 4 (III)
Solution of the full CPF analysis for the voltage at bus 14.