1

Fortescue Transformations for Three-Phase

Power Flow Analysis in Distribution Networks
I. Dzafic, Member, IEEE, T. Donlagic, Member, IEEE, and S. Henselmeyer

importance in nowadays.
Abstract—Power Flow solutions based on Fortescue’s Method Advantages of Fortescue's theory are well known, but this
of Symmetrical Components have been so far limited only to theory was never fully applied in any of the present power
three-phase transmission networks. Complete network solution flow solutions for distribution networks.
could be reached only by applying standard methods based on
Reference [13] describes a hybrid solution which first
phase coordinate system. This paper shows that it is possible to
apply Fortescue’s General Transformations for power flow decouples the network on balanced (three-phase laterals only)
solution on complete network. Special matrices for handling and unbalanced (presence of single and two-phase laterals as
distribution network configurations are derived. Presented well) part and then calculates power flow of balanced system
approach offers greater computational performances when using Newton-Raphson method in FC system and unbalanced
solving power flow problem in distribution networks. part of the network using Forward-Backward method in phase
Index Terms—Fortescue transform, Fortescue coordinates
coordinate system. As such, this method offers a significant
(FC), phase coordinates (PC), symmetrical components,
transformation matrices, power flow, distribution network improvement in memory requirements and calculations
savings.
I. NOMENCLATURE However, in order to gain full benefits of Fortescue’s theory,
it has to be uniformly applied on the complete network, i.e.
X Matrices are presented in uppercase letters
without separating it on balanced and unbalanced part.
X, x complex values are underlined
indices with lower case letters a, b, c and their This paper proposes a new approach to existing solutions of
abc combinations represent different phases: abc, ab, Power Flow problem. It defines special transformation
bc, ca, a, b, c matrices which are necessary to resolve difficulties with the
indices with numbers 0, 01, 012 represent unbalanced distribution networks. Using general Fortescue’s
012 theory it is possible to speed up solutions and reduce memory
Fortescue coordinates
P Indicates Phase coordinates requirements up to 3 times. In this way significant
F Indicates Fortescue coordinates improvements in performances are achieved.
Pn, Fn Indicates coordinates of dimension n (index)
APn a Fn General transformation matrix from n-phase phase III. FORTESCUE’S TRANSFORMATIONS THEORY
coordinates to n-phase Fortescue coordinates and In his published paper [16] from 1918., Fortescue showed that
AFn a Pn vice versa. The arrow a shows direction. any unbalanced set of n phasors can be transformed into n-1
AFn a Fm Fortescue transform from n- to m-phase Fortescue balanced n-phase systems of different phase sequence and one
coordinates zero-phase sequence system. By his definition a zero-phase
an Fortescue complex operator for n-phase system sequence system is one in which all phasors are of equal
magnitude and angle or they are all identical [17]. Fortescue
II. INTRODUCTION Transform is accomplished using transformation matrices that

P erformance demands have caused development of many

power flow techniques over last few decades.
Conventional solutions such as Y- and Z-matrix [1]-[4],
differ in dependence of the system dimension, i.e. the number
of system phases, n.
Any n-phase system defined in PC as:
Newton-Raphson and its variants [5]-[9], and those for radial V Pn = ⎡⎣V a V b V c K V n ⎤⎦
T
(1)
and weakly meshed networks [10]- [12], are all carried out in
is transformed into FC using the following matrix equation:
phase coordinate system. There are also several methods
which use non-conventional solutions partially based on ⎡ V0 ⎤ ⎡1 1 1 K 1 ⎤ ⎡V ⎤
⎢V ⎥ ⎢ n −1 ⎥⎢ a⎥
Fortescue's Symmetrical components [13]-[15]. All mentioned ⎢ 1 a a
2
K a ⎥ ⎢V b ⎥
methods give excellent results in the term of accuracy, but ⎢ 1 ⎥ 1 n n n
⎢
⎢ V 2 ⎥ = 1 an 2 4 2 ( n −1) ⎥
⎢V ⎥
an K an (2)
computational costs are those to be considered with great ⎢ ⎥ n⎢ ⎥⎢ c⎥
⎢ M ⎥ ⎢M M M K M ⎥⎢ M ⎥
⎢V ⎥ ⎢ n −1 2( n −1)
⎥
( n −1)( n −1) ⎢V ⎥
Authors are with Siemens AG, IC SG EA SOL, Humboldtstr. 59, 90459
⎣ n −1 ⎦ ⎢⎣1 a n an K an ⎥⎦ ⎣ n ⎦
Nuremberg, Germany (e-mail: [email protected],
In previous equation complex operator a is equal to e j 2π n and
[email protected], [email protected])

978-1-4673-2729-9/12/$31.00
Authorized licensed use limited©2012 IEEENouvelles. Downloaded on March 04,2024 at 09:43:12 UTC from IEEE Xplore.
to: IFP Energies Restrictions apply.
 2

rotates one’s phasor by 2π/n radians. changes when transforming one-phase system into Fortescue
Equation (2) can also be written as: components. In theory, the transformation matrix for single-
V Fn = APn a Fn V Pn (3) phase system is a single element matrix with its value of one.
It can be shown that A is nonsingular matrix, and therefore the
Fortescue transformation is unique: V. IMPEDANCE MATRIX IN FORTESCUE COORDINATES
V Pn = AFn a Pn V Fn (4) Voltage drop on n-phase lateral can be expressed as:
ΔV Pn = Z Pn I Pn (10)
where
T
⎡1 1 1 K 1 ⎤ where I Pn =⎡⎣I a I b I c K I n ⎤⎦ and Z Pn is the line
⎢ n −1 n−2 ⎥
⎢1 a n an K an ⎥ impedance matrix.
AFn a Pn = ⎢1 a nn − 2 a nn − 4 K a n2 ⎥
Considering (4), the voltage drop can now be expressed as:
⎢ ⎥ (5)
AFn a Pn ΔV Fn = Z Pn AFn a Pn I Fn (11)
⎢M M M K M ⎥
⎢ 2 n −1
⎥ where ΔV Fn is the vector of voltage drops and I Fn is the
⎣⎢1 a n a n K a n ⎦⎥
vector of currents, both in Fortescue coordinate system.
It is clear that when transforming n-phase system to FC, the Multiplying (11) with APn a Fn it becomes:
transformation matrix has dimensions of the system, i.e. n×n.
ΔV Fn = APn a Fn Z Pn AFn a Pn I Fn (12)
IV. FORTESCUE’S TRANSFORMATIONS IN POWER SYSTEMS Voltage drop in FC can be written as:
Fortescue’s Method of Symmetrical Components is suitable V Fn = Z Fn I Fn (13)
for application on the symmetrical three-phase power systems. which follows:
As shown in previous section, this theory can be applied on Z Fn = APn a Fn Z Pn AFn a Pn(14)
any n-phase system because it is only a mathematical linear
transformation of coordinates. This means that also one and where Z Fn is the line impedance matrix in Fortescue’s
two-phase laterals, which are a subject of analysis in the coordinate system.
power flow problem in distributive networks, can be For three-phase line, the impedance matrix in PC is
transformed into FC. The overview of transformation of one, represented as:
two and three-phase systems into FC (and vice-versa) will be ⎡ Z 11 Z 12 Z 13 ⎤
outlined at this point. ⎢ ⎥
Z abc = Z P3 = ⎢ Z 21 Z 22 Z 23 ⎥ , (15)
For three-phase laterals (n=3), well known transformation
matrices are used: ⎢⎣ Z 31 Z 32 Z 33 ⎥⎦
⎡1 1 1 ⎤ ⎡1 1 1 ⎤ where Zij is the mutual coupling impedance between phase i
1⎢ 2⎥ ⎢ 2 ⎥ and j, and Zii is the impedance of the phase i.
AP3 a F3 = ⎢1 a a ⎥ AF3 a P3 = ⎢1 a a⎥ (6)
3⎢ If the line is symmetrical, all off-diagonal elements
2 ⎥ ⎢ 2⎥
⎣1 a a⎦ ⎣1 a a ⎦ (couplings) are equal, so the matrix is described as:
where a = a 3 = e j 2π 3
. ⎡ Zl Zm Zm ⎤
⎢ ⎥
For example, the symmetrical components, zero, positive and Z abc = ⎢ Z m Z l Z m ⎥ , (16)
negative sequence, of the three-phase system voltages are ⎢⎣ Z m Z m Z l ⎥⎦
calculated as: where Zl is the phase impedance, and Zm is the mutual
V 012 = AP3 a F3 V abc (7) coupling impedance of the line.
where V 012 = V F3 = ⎡⎣V 0 V 1 V 2 ⎤⎦ ,
T Applying (14) to (16), it becomes:
T
⎡ Z l + 2Z m 0 0 ⎤
and V abc = V P3 = ⎡⎣V a V b V c ⎤⎦ . 1⎢ ⎥
Z 012 = Z F3 = ⎢ 0 Zl − Zm 0 ⎥ (17)
Two-phase system in FC consists of only two sequences, and 3
⎢⎣ 0 0 Z l − Z m ⎥⎦
its transformation matrices are as follows:
or
1 ⎡1 1 ⎤ ⎡1 1 ⎤
AP2 a F2 = ⎢ ⎥ AF2 a P2 = ⎢ ⎥ (8) ⎡ Z (012) 0 0 ⎤
2 ⎣1 −1⎦ ⎣1 −1⎦ ⎢
0
⎥
(012)
Here, because of two-phases, the complex operator is equal Z 012 = ⎢ 0 Z1 0 ⎥ (18)
⎢ (012) ⎥
to e jπ = −1 . ⎢⎣ 0 0 Z 2 ⎥⎦
Two-phase system written in FC will now be: Off diagonal elements are lost if the three-phase line is
V 01 = AP2 a F2 V ab (9) symmetrical. Exponent (012) denotes dimension of Fortescue
where V 01 = V F3 = ⎣⎡V 0 V 1 ⎦⎤ ,
T coordinate system.
To calculate voltages of the three-phase line, (18) is multiplied
T
and V ab = V P2 = ⎣⎡V a V b ⎦⎤ . with the current vector of the line:
In comparison with the phase coordinate system, nothing

Authorized licensed use limited to: IFP Energies Nouvelles. Downloaded on March 04,2024 at 09:43:12 UTC from IEEE Xplore. Restrictions apply.
 3

transformed, depending on the used matrix, lines that are of

⎡V 0 ⎤ ⎡ Z 0 0 ⎤ ⎡I 0 ⎤ ⎡Z 0 I 0 ⎤
(012)
0
⎢ ⎥ ⎢ ⎥ different type are of different coordinate system. This is
0 ⎥ ⎢⎢ I 1 ⎥⎥ = ⎢⎢ Z 1 I 1 ⎥⎥
(012)
⎢V 1 ⎥ = ⎢ 0 Z1 (19) because complex operators used in transformation matrices
⎢⎣V 2 ⎥⎦ ⎢ 0 0
(012) ⎥ ⎢
Z 2 ⎦⎥ ⎣ I 2 ⎦⎥ ⎣⎢ Z 2 I 2 ⎦⎥ differ. Power flow solutions require algebraic operations
⎢⎣
between line currents or node voltages. To achieve this, these
The line is decoupled into three independent sequences, zero, need to be in the same coordinate system in order to make
positive and negative, and thus described with three separate these operations legible. The adaptation to the unique
equations. coordinate system is explained in the next section.
The same works with two-phase line (e.g. bc line), which is Table 1 provides all possible combinations of phase
always symmetrical. configurations in a distribution network.
Impedance matrix of two-phase line, in PC is written as: TABLE 1
⎡Z Zm⎤ PHASE DISTRIBUTION IN TYPICAL DISTRIBUTION NETWORK
Z bc = ⎢ l ⎥, (20) Node from Node to
⎣ m Zl ⎦
Z 012(abc) 012(abc) v 01(ab) v 01(bc) v 01(ac) v 0(a) v 0(b) v 0(c)
01(ab) 01(ab) v 0(a) v 0(b)
and if transformed into FC it becomes:
01(bc) 01(bc) v 0(b) v 0(c)
0 ⎤ ⎡Z 0 0 ⎤
(01)
⎡Z + Z m 01(ac) 01(ac) v 0(a) v 0(c)
Z 01(bc ) = ⎢ l = ⎢ ⎥ (21)
Z l − Z m ⎦⎥ ⎣⎢ 0
0(a) 0(a)
⎣ 0
(01)
Z 1 ⎦⎥ 0(b) 0(b)
When multiplied with current vector, the solution for the 0(c) 0(c)
voltages is given by:
( 01) VII. PASSING FROM ONE LINE TYPE TO ANOTHER
⎡V 0 ⎤ ⎡ Z 0 0 ⎤ ⎡ I 0 ⎤ ⎡ Z (0 ) I 0 ⎤
01

⎢V ⎥ = ⎢ ⎥ ⎢ ⎥ = ⎢ ⎥. (22) There are two possibilities that need to be considered when

⎣ 1 ⎦ ⎢⎣ 0 Z 1( ) ⎥⎦ ⎣ I 1 ⎦ ⎢⎣ Z 1( ) I 1 ⎥⎦
01 01
working on adaptation to the same coordinate system - either
Therefore, in this case, the line is decoupled into two the adaptation is in the direction from lower to higher type of
independent systems. line, e.g. summation of line currents in backward sweep of
When transformed into FC, diagonal impedance matrix of Forward-Backward Algorithm ([2], [3], [5]), or vice-versa,
lateral is achieved. The advantages of such transformation are e.g. summation of node voltages in forward sweep of the same
clear. Firstly, it is required to keep only non-zero elements in algorithm.
the computers memory, and secondly, multiplication process
A. Passing from lower to higher type of line
is reduced from nine to three when dealing with three-phase
lines, and from four to two, when dealing with two-phase When transforming an n-phase line from phase coordinate
lines. In the case of single-phase lines, performances cannot system into FC, n×n transformation matrix with the
j 2π n
be improved. operator e is used. When transforming m-phase line from
If network is not symmetrical, off-diagonal elements will be PC to Fortescue one’s m×m transformation matrix with the
present, and calculation will be done using dense matrices, operator e
j 2π m
is used, where n<m.
thus no improvements in calculation and memory costs are
In the case from Fig. 1, “hot spot 1” connects two lines, and
possible in this case as well.
thus current vectors have different dimensions – m-phase line
is described with m-element and n -phase line with n-element
VI. POWER FLOW AND FORTESCUE TRANSFORM ISSUES IN vector.
DISTRIBUTIVE GRID Current vector of the n-phase line has to be adapted so that it
Previous principle can be applied easily to uniform three, two can be summed with the current vector of the m-phase line.
or single-phase networks. However, real power system The only way to do this is to transform the n-element vector
network is uniform only in its transmission part, where it is into coordinate system in which it can be represented as m-
mostly of three-phase, less often two-phase and when handled element vector.
with long distance transmission, single-phase type as HVDC
system. In distribution networks, different configuration of
network is met, that is, network is unbalanced, and phase
configuration varies in every part of it. One example of
network configuration in distribution systems is shown on Fig.
1.
It can be seen (“hot spot 1”) that three-phase branch is
continued with two-phase lateral which is again continued
with only single-phase lateral. Nodes which connect laterals
with different number of phases are called “hot spots”.
Exactly these places need attention when solving power flow Fig. 1 Example of Distribution Network configuration
problem, since additional steps have to be applied in order to
make calculations possible. Generally, there is no possibility to make a transformation
For known distribution network configuration all line from n-phase Fortescue to m-phase Fortescue coordinate
impedances are transformed into FC using (6) and (8). When

Authorized licensed use limited to: IFP Energies Nouvelles. Downloaded on March 04,2024 at 09:43:12 UTC from IEEE Xplore. Restrictions apply.
 4

system, but to first return to the original n-phase PC system Because of the unnecessary calculations with the zero-element
using (5). from the current vector, modification of the transformation
After the line was transformed to phase coordinate system, matrix can be made such that it does not do the unnecessary
transformation back to FC needs to be done, but this time to calculations with the zero-element. Dumping the column with
FC with the same number of coordinates as the m-phase line. the index of the zero element(s) in the current vector is
To apply this approach, multiplication of the current vector of sufficient, and in this way the current vector in Fortescue’s
the n-phase line with the matrix APm a Fm is necessary. coordinates, that is ready to be summed with the current
vector from the above line, is calculated. The following
However, their dimensions do not match.
equation demonstrates previous explanation:
Since it is not possible to multiply m×m matrix with the n-
phase line current vector, either a zero elements have to be ⎡I 0 ⎤ ⎡1 1⎤
⎢ ⎥ 1⎢ 2 ⎥ ⎡I ⎤
added in the n-element vector in place of missing phase(s) to I 012 = ⎢ I 1 ⎥ = ⎢ a a ⎥ ⎢ b ⎥ . (27)
match up with m, or the transformation matrix has to be ⎢⎣ I 2 ⎥⎦
3⎢ 2 ⎥ ⎣I c ⎦
reduced in a way that it becomes m×n i.e. m-n columns that ⎣a a⎦
corresponds to the missing phases have to be dumped out. Now it is possible to sum the currents of two branches which
Latter is better choice from the performance point of view. are with the different number of phases. The modified
Transformation matrix can be modified so that when transformation matrix used in the previous equation will be
multiplied with current vector of n-phase line it results in m- marked as AP2(bc) a F3 and it states the transformation from ab
element vector which is practically the representation of the n-
coordinate system to 012 coordinate system. Equation (27)
phase line in m-phase Fortescue coordinate system. Now it is
can be written as:
possible to make algebraic operation with this new
representation of current vector of n-phase line and the current ⎡I 0 ⎤
vector of the m-phase line because they have the same I 012 = AP2(bc ) a F3 I bc = ⎢⎢ I 1 ⎥⎥ . (28)
dimensions. ⎢⎣ I 2 ⎥⎦
The stated explanations can be applied to the example from
Since the transformation of the current vector had to be done
Fig. 1.
two times in order to make it suitable for summation, and
The current vector for two-phase line (bc) is:
since in both times the current vector has been multiplied with
⎡I ⎤ the transformation matrix from the left side, then the same can
I 01 = ⎢ 0 ⎥ = AP2 a F2 I bc (23)
⎣ I1 ⎦ be done in one step using only one matrix which is
And for three-phase line: multiplication of the two necessary transformation matrices:
⎡I 0 ⎤ ⎡I 0 ⎤
I 012 = ⎢⎢ I 1 ⎥⎥ = AP3 a F3 I abc (24) I 012 = ⎢⎢ I 1 ⎥⎥ = AP2(bc ) a F3 AF2 a P2 I 01 =
⎢⎣ I 2 ⎥⎦ ⎢⎣ I 2 ⎥⎦
It is necessary to return back to phase coordinate system first ⎡1 1⎤
by using the adequate transformation matrix. This is done for 1⎢ 2 ⎥ ⎡1 1 ⎤ ⎡I 0 ⎤
= ⎢ a a ⎥⎢ ⎥⎢ ⎥ = (29)
the two-phase line as the final goal is to adapt for the three- 3⎢ 2 ⎥ ⎣1 −1⎦ ⎣ I 1 ⎦
phase line. ⎣a a⎦
⎡I ⎤ ⎡2 0 ⎤
I bc = ⎢ b ⎥ = AF2 a P2 I 01 (25) 1⎢ ⎥ ⎡I ⎤
⎣ Ic ⎦ = ⎢ −1 j 3 ⎥ ⎢ 0 ⎥ = AF2(bc ) a F3 I 01
The three-phase line is in FC, and returning back to FC with
3⎢ ⎥ ⎣ I1 ⎦
the two-phase line is necessary in a way that the appropriate ⎣ −1 − j 3 ⎦
summation with the three-phase current vector of the line In the similar manner it is possible to derive matrices for all
above, which is defined in FC, can be done. Hence, the same other cases that may occur when passing from the lower to
transformation which was used for the three-phase line needs higher type of line (see appendix).
to be used, in order to make the summation in its coordinate
system. B. Passing from higher to lower type of line
In order to do this, a zero element in the current vector in the If the same example from Fig. 1 is considered (“hot spot 1”),
place of the missing phase can be added, and then multiplied voltage vector of node 1 has three, and the voltage vector of
with the corresponding matrix that transforms three-phase line node 2 has two elements. In order to make these two vectors
currents into FC. This can be done using the next equation: compatible for algebraic operations, they have to be written in
⎡I 0 ⎤ ⎡1 1 1 ⎤ ⎡ 0 ⎤ the same coordinate system. The procedure which will allow
⎢ ⎥ 1⎢ 2⎥⎢ ⎥ mutual operations between these two vectors differs from the
I 012 = ⎢ I 1 ⎥ = ⎢1 a a ⎥ ⎢ I b ⎥ (26)
3⎢ ⎥ one with the currents in backward sweep. Currents have had
⎢⎣ I 2 ⎥⎦ 2 ⎢ ⎥
a ⎦ ⎣I c ⎦
⎣1 a to be adjusted for the higher type of line, but with the
Equation (26) shows how two-phase line can be transformed voltages, situation is opposite. Voltage of node 2 is calculated
into FC such that it could be summed with the three-phase FC using voltage of node 1. Voltage vector of node 2 can be of
line. higher dimension. This means that the voltage vector of node

Authorized licensed use limited to: IFP Energies Nouvelles. Downloaded on March 04,2024 at 09:43:12 UTC from IEEE Xplore. Restrictions apply.
 5

1 has to be transformed in the coordinate system in which the with all other cases that can happen in the distributive
voltage vector of node 2 is already written. In order to do this, network, as stated in Table 1, are presented.
voltage vector of node 1 has to be transformed to the PC with
the dimension of the voltage vector of node 2, and then, it is to VIII. RESULTS
be transformed to FC. Generally, for a line with two nodes, if The presented approach has been tested using Forward-
node 1 is described with the m-element voltage vector, and Backward sweep power flow algorithm based on the presented
node 2 with the n-element voltage vector, where n<m, then Fortescue Transformations.
the m-element vector needs to be transformed into n-phase For the test network IEEE 13-bus network has been chosen.
Fortescue coordinate system. This can be done with usage of Table 2 lists the results of the voltage magnitudes for all
the Fortescue transformation matrices. The process is network buses.
described with the following steps: For this test, power flow calculations have been executed
Step 1: The voltage vector of the node 1 is transformed back without any voltage regulation, and the spot loads between
into phase coordinate system using the modified nodes 632 and 671 have been modeled as 5 identical loads at
AFm a Pn transformation matrix. This modification is nodes 632, 671 and 3 additional virtual nodes added between
needed because node 1 has more phases attached that nodes 632 and 671. For all buses the difference between the
are not present on the node 2. Because of this, when voltage magnitudes and angles from [20] and the proposed
transforming to PC this phase(s) need to be eliminated. approach is very small (respectively, < 0.009 p.u. and < 0.245
This is done by modifying the transformation matrix deg.). The slight difference in modeling explains these small
in the way that the row(s) with index of the missing differences. This comparison of the results shows that the
phase(s) are dumped out of the matrix. In this case, Fortescue-based power flow leads indeed to the same results
after the transformation, the voltage vector of the node as those presented in [20] where a non-symmetrical line model
1 will have only elements for those phases that the is used.
voltage vector of the node 2 has. TABLE 2 POWER FLOW RESULTS FOR IEEE 13 TEST NETWORK
|V| (p. u.) / angle (degree) |V| (p. u.) / angle (degree) |Delta|
Step 2: This new voltage vector now needs to be transformed IEEE13 Phase IEEE13 Fortescue
coordinates coordinates
into n-phase FC. For this, it should be multiplied with Bus A B C A B C A B C
the n×n transformation matrix APn a Fn . 650
1 1 1 1 1 1 0.000 0.000 0.000
0 -120 120 0 -120 120 0.000 0.000 0.000
Applying the forward step to the example from Fig. 1, process 632
1.021 1.042 1.0174 1.020 1.041 1.016 0.001 0.001 0.001
is described with the following equations. -2.49 -121.72 117.83 -2.485 -121.708 117.594 0.005 0.012 0.236
1.018 1.0401 1.0148 1.017 1.038 1.013 0.001 0.002 0.002
The voltage vector of the node 1 is transformed from three- 633
-2.56 -121.77 117.82 -2.555 -121.526 117.584 0.005 0.244 0.236
phase FC to two-phase PC: 634
0.994 1.0218 0.996 0.992 1.019 0.994 0.002 0.003 0.002
-3.23 -122.22 117.34 -3.224 -122.098 117.223 0.006 0.122 0.117
⎡V ⎤
⎡V ⎤ ⎡1 a a ⎤⎢ 0⎥
2 1.0329 1.0155 1.030 1.012 0.003 0.003
645
V bc = ⎢ b ⎥ = ⎢ ⎥ V
2 ⎢ 1⎥
(30) -121.9 117.86 -121.778 117.742 0.122 0.118

⎣V c ⎦ ⎣⎢1 a a ⎦⎥ ⎢V ⎥ 646
1.0311 1.0134 1.028 1.009 0.003 0.004
⎣ 2⎦ -121.98 117.9 -121.858 117.664 0.122 0.236
0.99 1.0529 0.9778 0.986 1.048 0.974 0.004 0.005 0.004
In order to transform this vector into two-phase Fortescue, the 671
-5.3 -122.34 116.02 -5.274 -122.095 115.788 0.027 0.245 0.232
following equation is used: 692
0.99 1.0529 0.9777 0.984 1.048 0.972 0.006 0.005 0.006

⎡V ⎤ -5.31 -122.34 116.02 -5.283 -122.218 115.788 0.027 0.122 0.232

V 01 = ⎢ 0 ⎥ = AP2 a F2 V bc (31) 675
0.9835 1.0553 0.9758 0.978 1.050 0.970 0.006 0.005 0.006

⎣V 1 ⎦ -5.56
0.9881
-122.52 116.03
0.9758
-5.527
0.981
-122.397 115.914
0.970
0.033
0.007
0.123 0.116
0.006
684
In this way the voltage vector of node 1 is prepared for the -5.32 115.92 -5.283 115.688 0.037 0.232
algebraic operations with the voltage vector of the node 2, 611
0.9738 0.965 0.009
115.78 115.664 0.116
thus the voltage of this node can be calculated. Forward sweep 0.99 1.0529 0.9778 0.982 1.046 0.970 0.008 0.007 0.008
680
can be described with only one equation as well: -5.3 -122.34 116.02 -5.263 -122.095 115.904 0.037 0.245 0.116
0.9825 0.974 0.009
⎡V ⎤ 652
-5.25 -5.198 0.053
V 01 = ⎢ 0 ⎥ = AP2 a F2 AF3 a P2(bc ) V 012 =
⎣V 1 ⎦
A. Performance test
⎡V ⎤
1 ⎡1 1 ⎤ ⎡1 a a ⎤⎢ 0⎥
2
Generally, the gain on performance of the described
= ⎢ ⎥⎢ ⎥ V =
2 ⎢ 1⎥
2 ⎣1 −1⎦ ⎣⎢1 a a ⎦⎥ ⎢ ⎥ (32) approach is that for the three-phase main and laterals the
⎣V 2 ⎦ number of multiplications is reduced to one third compared to
⎡V ⎤ the calculation in phase coordinates, and for two-phase
1 ⎡2 −1 −1 ⎤ ⎢ 0 ⎥ laterals, it is halved. On the average, a gain on performance
= ⎢ ⎥ V 1 = AF3 a F2(bc ) V 012
2 ⎢⎣ 0 − j 3 j 3 ⎥⎦ ⎢ ⎥ between 2 and 3 can be therefore expected.
⎢⎣V 2 ⎥⎦
“Hot spot 1” was only one example of handling phase IX. CONCLUSION
transition in the distribution networks, and in the appendix of
This paper presented a complete Power Flow solution based
this paper the complete list of matrices that needs to be used
on Fortescue transformations that gives better performances to
a great extend compared to the conventional methods used for

Authorized licensed use limited to: IFP Energies Nouvelles. Downloaded on March 04,2024 at 09:43:12 UTC from IEEE Xplore. Restrictions apply.
 6

solving power flow in distribution networks. The method used more accurate power flow calculations. Since the forward-
in this paper is Forward-Backward sweep algorithm which backward method is only suitable for radial and weakly
gives excellent results with the radial networks that are met in meshed networks, the same principal can be applied on the
distributive systems. The number of float point calculations is conventional power flow methods. “Hot Spot”
reduced by three in case of three-phase laterals and by two in transformations matrices are defined in advance and can be
case of two-phase laterals. This is the contribution to many used easily during the algorithm execution.
existing power flow techniques and a step forward to enabling

X. APPENDIX
Table 3 Complete list of Fortescue’s transformation matrices necessary for handling “hot spots”
Case Backward sweep Forward sweep

⎡ 2 0 ⎤
1 ⎡ 2 −a −a ⎤
2
1⎢ 2 ⎥
AF2( ab) a F3 = ⎢−a 1− a ⎥ AF3 a F2( ab) = ⎢ ⎥
3⎢ 2⎥
2 ⎣⎢ 0 1 − a 2 1 − a ⎦⎥
⎣ −a 1 − a ⎦

⎡2 0 ⎤
1⎢ ⎥ 1 ⎡2 −1 −1 ⎤
AF2(bc ) a F3 = ⎢ −1 j 3 ⎥ AF3 a F2(bc ) = ⎢ ⎥
3⎢ ⎥ 2 ⎢⎣ 0 − j 3 j 3 ⎥⎦
⎣ −1 − j 3 ⎦

⎡ 2 0 ⎤
1 ⎡ 2 −a −a ⎤
2
1⎢ 2⎥
AF2( ca ) a F3 = ⎢ −a 1 − a ⎥ AF3 a F2( ca ) = ⎢ ⎥
3⎢ 2 2 ⎢⎣ 0 1 − a 1 − a 2 ⎥⎦
⎣ −a 1 − a ⎥⎦

⎡1⎤
1⎢ ⎥ AF3 a F1( a ) = [1 1 1]
AF1( a ) a F3 = ⎢1⎥
3
⎢⎣1⎥⎦

⎡1⎤
1⎢ ⎥
AF3 a F1(b) = ⎡⎣1 a a ⎤⎦
2
AF1(b ) a F3 = ⎢a⎥
3⎢ 2⎥
⎣a ⎦

⎡1⎤
1 ⎢ 2⎥
AF3 a F1(c) = ⎡⎣1 a a ⎤⎦
2
AF1( c) a F3 = ⎢a ⎥
3⎢ ⎥
⎣a⎦

1 ⎡1⎤ AF2( ab) a F1( a ) = AF2(bc) a F1(b ) = AF2( ca ) a F1( a ) = [1 1]

AF1( a ) a F2( ab) = AF1(b) a F2(bc) = AF1( a ) a F2(ca ) = ⎢⎥
2 ⎣1⎦

1⎡1⎤ AF2( ab) a F1(b ) = AF2(bc ) a F1(c ) = AF2( ca ) a F1( c ) = [1 −1]

AF1(b) a F2( ab) = AF1(c) a F2(bc) = AF1( c) a F2( ca) = ⎢ ⎥
2 ⎣ −1⎦

Authorized licensed use limited to: IFP Energies Nouvelles. Downloaded on March 04,2024 at 09:43:12 UTC from IEEE Xplore. Restrictions apply.
 7

1999 and 2002, respectively. Currently, he is with Siemens AG, Nuremberg,

Germany. He is the head of the department for Smart Grid Applications R&D.
XI. REFERENCES His research interest includes systems modeling and simulation and applying
[1] K. A. Birt, J. J. Graff, J. D. McDonald, and A. H. El-Abiad, “Three advances in software engineering to power systems simulations. He is a
phase load flow program”, IEEE Trans. on Power Apparatus and member of IEEE Power & Energy and Computer societies.
Systems, vol. 95, no. 1, pp 59-65, Jan/Feb, 1976.
[2] D. I. H. Sun, S. Abe, R. R. Shoults, M. S. Chen, P. Eichenberger, and D. Tarik Donlagic (M’07) received his M.Sc. in Power Engineering from
Farris, “Calculation of energy losses in a distribution system”, IEEE University of Sarajevo, Sarajevo, Bosnia in 2010. He is currently with
Trans. on PAS, vol. 99, no. 4, pp. 1347-1356, July-August 1980 Siemens AG where he works on development of smart grid applications for
[3] B. K. Chen, M. S. Chen, R. R. Shoults, and C. C. Liang, “Hybrid three distribution networks. His interests include power system modeling and
phase load flow”, Proc. IEE Generation, Transmission, and Distribution, control and electrical machines and drives. He is a member of the IEEE Power
vol. 137, no. 3, pp177-185, May 1990. & Energy and Industry Applications societies.
[4] Chen T. H., Chen M. S., Hwang K. J., Kotas P., and Chebli E. A. (1991),
“Distribution system power flow analysis- a rigid approach”, IEEE Sylwia Henselmeyer is a Software Developer with Siemens AG, Nuremberg,
Trans. Power Delivery, vol. 6, no. 3, July, pp. 1146-1152 Germany, working in distribution system applications. Her interests are state
[5] R. G. Wasley and M. A. Shlash, “Newton-Raphson algorithm for 3- estimation, numerical programming, computer graphics, object oriented
phase load flow”, proc. IEE, vol. 121, pp. 630-638, July 1974 analysis and design. She received her Diploma in Computer Science from the
[6] H. L. Nguyen, “Newton–Raphson method in complex form-power University Erlangen-Nuremberg, Germany in 2005.
system load flow analysis”, IEEE Trans. Power Systems, vol. 12, no. 3,
pp. 1355–1359, August 1997
[7] P. A. N. Garcia, J. L. R. Pereria, S. Cameiro J. R, V. M. Da Costa, and
N. Martins, “Three-phase power flow calculation using the current
injection method”, IEEE Trans. on Power Systems, vol. 15, no. 2, pp
508-514, May 2000.
[8] J. Arrillaga and C. P. Arnold, “ Fast decoupled three phase load flow”,
Proc. IEE, vol. 125, no. 8, pp 734-740, 1978.
[9] A. V. Garcia and M. G. Zago, “Three-phase fast decoupled power flow
for distribution networks”, Proc IEE. Generation, Transmission, and
Distribution, vol. 143, no. 2, pp. 188-192, March 1996
[10] C. S. Cheng and D. Shirmohammadi, “A three-phase power flow method
for real-time distribution system analysis”, IEEE Trans. on Power
Systems, vol. 10, pp. 671–679, May 1995.
[11] J. H. Teng, “A direct approach for distribution system load flow
solutions”, IEEE Trans. on power system, vol. 18, no. 3, pp. 882-887,
July 2003
[12] Shirmohammadi, D.; Hong, H.W.; Semlyen, A.; Luo, G.X.; , “A
compensation-based power flow method for weakly meshed distribution
and transmission networks”, Power Systems, IEEE Transactions on ,
vol.3, no.2, pp.753-762, May 1988
[13] Abdel-Akher, M.; Nor, K.M.; Abdul-Rashid, A.-H.; , “Development of
unbalanced three-phase distribution power flow analysis using sequence
and phase components”, Power System Conference, 2008. MEPCON
2008. 12th International Middle-East, vol., no., pp.406-411, 12-15
March 2008
[14] Jen-Hao Teng; , “A direct approach for distribution system load flow
solutions”, Power Delivery, IEEE Transactions on, vol.18, no.3, pp.
882- 887, July 2003
[15] Kang Xiaoning; Suonan Jiale; Song Guobing; Fu Wei; Zhiqian Bo; , “A
novel three phase load flow algorithm based on symmetrical
components for distribution systems”, Universities Power Engineering
Conference, 2008. UPEC 2008. 43rd International , vol., no., pp.1-4, 1-4
Sept. 2008
[16] C. L. Fortescue, “Method of Symmetrical Co-ordinates Applied to the
Solution of Polyphase Networks”, Presented at the 34th Annual
Convention of the American Institute of Electrical Engineers, Atlantic
City, N. J., June 28, 1918.
[17] Anderson, P., " Analysis of Faulted Power Systems," vol., no., pp.540
[18] Zhang, X.-P.; Chen, H.; , “Asymmetrical three-phase load-flow study
based on symmetrical component theory”, Generation, Transmission and
Distribution, IEE Proceedings , vol.141, no.3, pp.248-252, May 1994
[19] Lo, K.L.; Zhang, C.; , "Decomposed three-phase power flow solution
using the sequence component frame," Generation, Transmission and
Distribution, IEE Proceedings C , vol.140, no.3, pp.181-188, May 1993
[20] 13-node IEEE Test Feeder – Data and Release 2006. Available:
http://ewh.ieee.org/soc/pes/dsacom/testfeeders/index.html
[21] J. L. Blackburn, Symmetrical Components for Power System
Engineering, New York: Marcel Dekker Inc., 1993.

XII. BIOGRAPHIES
Izudin Dzafic (M’05) received M.Sc. and Ph.D. degrees from the University
of Tuzla, Tuzla, Bosnia, and the University of Zagreb, Zagreb, Croatia, in

Authorized licensed use limited to: IFP Energies Nouvelles. Downloaded on March 04,2024 at 09:43:12 UTC from IEEE Xplore. Restrictions apply.