Sie sind auf Seite 1von 10

IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 29, NO.

1, JANUARY 2014

193

Generalized Fortescue Equivalent Admittance


Matrix Approach to Power Flow Solution
Izudin Dafi, Senior Member, IEEE, Bikash C. Pal, Fellow, IEEE, Michel Gilles, Senior Member, IEEE, and
Sylwia Henselmeyer, Member, IEEE

AbstractThis paper develops a generalized admittance matrix


approach in Fortescue coordinate system to solve unbalanced/unsymmetrical distribution networks including different number of
phases. This generalized Fortescue equivalent is defined in this
paper for solving the heterogeneous phase, and thus Fortescue, network model. The performance of the approach is demonstrated in
different model networks with number of nodes ranging between
168 and 14200. It is found that the current iteration method exploiting the decoupling in admittance matrix in Fortescue coordinate is substantially faster than the typical unbalanced three-phase
solution in phase domain. The method has a significant potential
for application in real time active power network management.
Index TermsDistribution system, Fortescue transformations,
symmetrical components, three-phase power flow.

NOMENCLATURE
0, 1, 2

Explicit suffix for symmetrical components.


Explicit suffix for phase coordinates.
Voltage and current complex vectors.
Complex element of a vector or matrix.
Complex Fortescue operator

Nodal admittance submatrices in -Fortescue


domain for node .
Branch admittance submatrices in -Fortescue
domain for an branch.
Fortescue transformation from
to -phase domain.

-Fortescue domain

Fortescue transformation from


-Fortescue domain.

-phase domain to

th sequence element or sequence voltage vector.


Current of node in -Fortescue domain.
Current of node in -phase domain.
Fortescue domain of node .

Manuscript received January 07, 2013; revised April 30, 2013 and August
03, 2013; accepted August 23, 2013. Date of publication September 13, 2013;
date of current version December 16, 2013. Paper no. TPWRS-01337-2012.
I. Dafi, M. Gilles, and S. Henselmeyer are with Siemens AG, Nuremberg 90450, Germany (e-mail: idzafic@ieee.org; nbgh_007@ieee.org;
sylwia.henselmeyer@siemens.com).
B. C. Pal is with the Electrical and Electronic Engineering Department, Imperial College London, London SW7 2BT, U.K. (e-mail: b.pal@imperial.ac.uk).
Digital Object Identifier 10.1109/TPWRS.2013.2279755

Voltage magnitude for phase .


Voltage angle for phase .
I. INTRODUCTION

OWER distribution networks are undergoing unprecedented transitions in recent time. The driver is the need
to utilize the network capacity through modern comprehensive
distribution management system (DMS) while accommodating
more local generation and demand-side management. This is
essentially how a smart distribution grid is expected to function.
Fast computation and control are at the core of such DMS. The
effectiveness of network automation is very much dependent
on the speed of computation. Power flow, state estimation,
and contingency analysisall of these critical functions rely
on fast computation of network voltage magnitudes and phase
angles. These are routine tasks in a transmission system that is
balanced and symmetric. On the other hand, the power distribution network in practice is rarely symmetric and balanced.
The asymmetry originates from nonsymmetrical and/or nontransposed three-phase mains, two and single-phase laterals.
The demand is largely unbalanced, even in symmetrical distribution networks, requiring three-phase approach to power flow
solution. One way to deal with the symmetrically constructed
unbalanced system is the method of symmetrical components
(SC) [1][3] where positive, negative and zero sequence network models are prepared and simultaneously solved. This
reduces significantly the size of the problem because of the
decoupling between the sequence network models. When
applied to a NewtonRaphson (NR) power flow the usage of
symmetrical components allows to replace the original
Jacobian matrix by a
Jacobian matrix for positive
admittance matrices for
sequence power flow and two
negative sequence and zero sequence nodal current equations.
The decoupling enables parallel computations as well.
Wasley and Shlash [4] proposed a generic three-phase power
flow solution framework in phase coordinates and solved
through NR method which was enhanced for faster solution
through fast decoupled load flow (FDLF) technique [5], [6].
A complex version of the Jacobian has been implemented for
faster convergence in [7]. The sparsity of the Jacobian has
been nicely exploited for faster convergence in the distribution
network model in [8][11]. Reference [12] has proposed a
direct solution using the bus and branch incidence matrix
approach. A branch current approach is derived to decouple
three phases in [13]. The symmetrical components are used in
[14] to decouple the network into three sequence components
and solve them simultaneously and iteratively. The couplings

0885-8950 2013 IEEE

194

between the admittance matrices are ignored, resulting in the


loss of accuracy of the solution. The representation of the
phase shift in the transformer in the sequence domain is simplified and made symmetric with a scaling concept [15], [16].
The advantage is fast solution of three-phase unbalanced and
unsymmetrical transmission and distribution systems. These
couplings between the sequence admittance matrices because
of the untransposed lines have been modeled as compensating
current injections in positive, negative, and zero sequence
network and solved in parallel with NR [17] and FDLF NR
method [18]. This approach has been developed comprehensively [19] to include transformer of various vector groups for
network of different R/X ratios. In [20], the positive sequence
part is solved as a power mismatch solution to keep the standard
structure of a balanced single-phase power flow solution. In
[19] and [20], the authors use a combination of symmetrical
and phase components depending on the type of the nodes. The
recent advancement on three-phase unbalanced power flow
models includes the voltage unbalance factor at the point of
common coupling of network and wind farm [21]. It is found
that the power output of each phase is very much dependent on
the voltage unbalance factor, and the simple positive sequence
approach does not lead to robust convergence and accurate
solution. In the context of micro-grid with many electronically
interfaced distributed energy resources, the distributed slack
bus approach has been adopted considering slack bus real and
reactive power as state variables [22] in the sequence components framework. The three-phase power flow formulation is
extended to a quasi-Newton optimization framework, where
the reactive power output from distributed micro generator
and real power demand are control variables. The concept
is demonstrated to be a compact and integral part of a DMS
system with convincing results for 123-node distribution feeder
model [23].
In a situation where the number of phases changes through
single- and two-phase laterals, the usage of the symmetrical
component method does not offer much advantage because of
the presence of the mutual coupling. In fact, new couplings
are observed between the parts with different sequence systems. The formulation and interface between three-phase and
two-phase are not generic and straightforward. Some authors
have used a hybrid approach such as a phase approach to laterals and component approach to three-phase mains [24], [25].
Reference [26] has solved balanced parts with a single-phase
approach and unbalanced parts with symmetrical component
method with an iterative interface matrix of sequence component admittances. Some authors [27][30] have dealt with the
laterals by forward/backward sweep algorithms.
In [31], the authors proposed forward backward sweep power
flow using only voltage and current Fortescue transformations
between different Fortescue domains not to the node admittance. This paper used the application of Fortescue transformation to any matrix-based power flow algorithm where the node
admittance matrix is also factored into the generalized Fortescue
domain. The entire power solution with voltage, current, and admittance in vector and matrix form is given in the Fortescue domain. It is achieved by connecting a equivalent for any combination of its terminal phase types. It is very generic and called

IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 29, NO. 1, JANUARY 2014

the generic Fortescue equivalent approach. The solution of


equations in vector and matrix form in the Fortescue domain
is very fast. For the complete system, the existing couplings
between the phases of symmetrical elements are eliminated in
the Fortescue domain. As an important consequence, the computational complexity is significantly reduced. Unlike the iterative forward backward ladder network solution approach in
[31], a direct nodal voltage solution is obtained through iterative nodal power mismatch minimization. The solution is very
fast because of the near decoupling nature of admittance matrix
in generalized Fortescue domain and it works very well for radial and meshed system model. The coupling is only present in
the element of the Y-bus matrix corresponding to phase transition nodes. The contribution of this paper is in representing
the different number of phases in the generalized Fortescue domain and fast obtaining the solution in the Fortescue domain
with comparable accuracy but at higher speed. It is shown that
a generalized Fortescue equivalent can expedite the solution of
matrix-based power flow problems.
Next, the Fortescue transformation for phase system is
briefly described and is utilized to develop various interface
matrices to connect lines of different number of phases.
II. FORTESCUE TRANSFORMATION THEORY
In a seminal paper [1], Fortescue proved that any unbalanced
balanced -phase
set of phasors can be transformed into
system of different phase sequences and one zero phase sequence system by the following transformation:

..
.

..
.

..
.

..
.

..

..
.

..
.
(1)

where
voltages in Fortescue coordinates;
voltages in phase coordinates.
Because of the nonsingularity of the transformation, it is also
proved that the quantities from the Fortescue domain to the
transphase domain can be recovered through an inverse
formation. In the following paragraphs, the transformation ma,
, and
will be
trices for the special cases
briefly reviewed.
into
The three-phase system is transformed through
. The transformed
Fortescue coordinates using
quantities (voltage and current) are recovered in the phase dogiven as follows:
main through the inverse transformation

(2)
Electric current through a three-phase element is given by
(3)

DAFI et al.: GENERALIZED

FORTESCUE EQUIVALENT ADMITTANCE MATRIX APPROACH TO POWER FLOW SOLUTION

195

Transforming voltage and current into Fortescue domain, (3)


can be rewritten as

Three-phase admittance matrix is transformed into the


Fortescue domain of order 3 as shown in
(4)
The Fortescue transformation (4) when applied to a symmetrical
three-phase element (stationary or rotating) results in a 3 3
diagonal matrix.
The two-phase system in Fortescue coordinates will only
, the operator
have 0 and 1 sequences. Here, because of
. The respective transformation matrices are

(5)
Two-phase admittance matrix is transformed into Fortescue domain of order 2 as shown in
(6)

Fig. 1. IEEE-13 network with marked PTNs.

as phase transition nodes (PTNs). All PTNs are marked with


a solid dot surrounded by a diamond. There are three buses in
Fig. 1 with this feature: buses 632, 671, and 684. Bus 632 of
phase-type ABC is connected to a branch of phase-type BC.
Bus 671 of phase-type ABC is connected to a branch of phasetype AC. Bus 684 of phase-type AC is connected to a branch of
phase-type A and a branch of phase-type C.
For branches connected at both terminals to buses with
matching phase type, the standard Fortescue coordinate
transformation matrices are used on the phase-domain bus
admittance matrix. The choice of the transformation matrix
depends on the number of branch phases. For three-phase
branches, the matrices as described by (2) and (4) are used; for
two-phase branches, the transformation is executed based on (5)
and (6). For the branches connected with at least one terminal
to a PTN, a careful consideration is required to build a proper
bus admittance matrix required for power flow execution.

The admittance matrix in case of a two-phase element is


(7)
Substituting the expression for

from (7) into (6) results in


(8)

In comparison with the phase coordinate system, nothing


changes when transforming a one-phase system admittance
into a Fortescue coordinate. The transformation matrix for a
single-phase system is a 1 1 identity matrix. This is because the
, which in this case equals 1.
value of the operator
III. APPLICATION OF FORTESCUE COORDINATES TO AN
UNSYMMETRICAL/UNBALANCED DISTRIBUTION NETWORK
Distribution networks are typically unsymmetrical and unbalanced due to, respectively, the use of asymmetrical equipment such as two-phase and single-phase laterals [19] drawn
from a three-phase feeder and the presence of unbalanced loads.
Further asymmetry results from nontransposed overhead lines.
One example of a typical distribution network is presented in
Fig. 1. It shows the IEEE-13 distribution network model [32].1,
For each branch, the phase type information is provided. The
branches connected with at least one terminal to a node of different number of phases are drawn in bold. These are defined
1[Online]. Available: http://ewh.ieee.org/soc/pes/dsacom/testfeeders/index.
html

A. Handling PTNs in Phase Domain


The equivalent definition, in phase domain, is based on
number and type of branch phases. However, even in that case,
there exists phase mapping on PTNs. Let us first explain the required mathematical operations for phase-domain mapping on
PTNs using node 632 in Fig. 1. Node 632 is connected to other
nodes 633, 650, and 671 through three-phase lines and to node
645 through a two-phase line (BC). Each element of the admittance matrix of these lines equivalents is represented as 3
3 admittance submatrix for the three-phase lines and as 2
2 admittance submatrix for the two-phase line. For clarity,
a two-phase node 632 has been introduced. Fig. 2 illustrates
these two-phase domains and the need to define their interface
via the mapping between nodes 632 and 632 . In phase coordinates, the equivalent of the two-phase line between nodes
632 and 645 is given by

(9)

The contribution of this line to the corresponding system admittance matrix can be obtained by simple addition and proper
phase mapping of the admittance matrix described in (9). This
, is easily implemented
mapping, used to eliminate node
by simply adding one zero row/column vector in place of the

196

IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 29, NO. 1, JANUARY 2014

Fig. 3. Three-phase to two-phase lateral.


Fig. 2. Interconnection of nodes of different phase combinations.

Contribution

can be written as
(17)

missing phase A at node


and then substituting node
with node 632. Equation (9) then becomes

Using two-phase voltage transformations from (5), we have

(10)

Using (10), it is now easy to apply Kirchhoffs current law


without any additional phase mapping for building of the system
nodal admittance matrix.
B. Handling PTNs in the Fortescue Domain

and two-phase to Fortescue


(14) yields

(18)
domain current transformation
(19)

Similar transformations can now be applied on node 645, to


form full Fortescue equivalent of two-phase line between
three-phase node 632 and two-phase node 645 as follows:

Applying Kirchoffs current law in Fortescue coordinates requires all node currents to be in the same Fortescue domain. Applying Fortescue transformation to the above example requires
a Fortescue-specific step to integrate the mapping between the
gives
two Fortescue domains. Rewriting (9) for

(11)
The contribution of node

to the three-phase node 632 is


(12)

Mapping between three-phase node voltage 632 in Fortescue


domain and two-phase node voltage
in phase domain [31]
is defined as
(13)

(20)

equivalent of this
Equation (20) represents the Fortescue
branch connecting two different Fortescue domains. Note that,
is unaffected
as would be expected, the submatrix
by the mapping transformation. It is also important to note
equivalent will be needed only for
that such a Fortescue
one- or two-phase branches that connect to different Fortescue
domains. However, a formulation as given in (20) will enable
one to build a Fortescue system admittance matrix in a similar
way to that used in the phase domain.
In Section IV, the mathematical formulation of Fortescue
equivalents for all possible mapping cases are described.
IV. FORTESCUE

EQUIVALENTS

A. Three-Phase to Two-Phase Lateral


Similarly, using current mapping from two-phase to Fortescue
domain [31] yields

(14)
Substituting (13) and (14) into (12) yields

(15)
and thus
(16)

Let us consider a network segment as shown in Fig. 3. It


,
, and
. The
shows two buses and and three lines
phase types are shown in Fortescue and in phase coordinates.
The voltages at buses and are shown in Fortescue coordinates
and
, respectively. The bold two-phase
and marked as
line
is connected to the three-phase bus at which line phase
and bus phase types do not match. For all remaining elements in
this example, there is no mismatch between the bus and equipment phase type.
Bus is a three-phase bus which requires a Fortescue domain of third order . Bus is a two-phase bus which requires
Fortescue domain of second order . In order to apply Kirchhoffs current law, all currents at bus must be formulated in the
in this case. Although line
is
same Fortescue order, i.e.
a two-phase line, its current contribution at bus side needs to

DAFI et al.: GENERALIZED

AND

FORTESCUE EQUIVALENT ADMITTANCE MATRIX APPROACH TO POWER FLOW SOLUTION

197

TABLE I
SUBMATRICES FOR DIFFERENT PHASE TYPES
Fig. 4. Three-phase to single-phase lateral.

The Fortescue equivalent sub-matrix is given by


(24)
Equation (24) can be written in general form as
be calculated in domain . We will now derive the Fortescue
equivalent described in

(21)

and already introduced in the previous section.


All four submatrices will be derived directly, i.e. bypassing
the fictitious node introduced in the previous section, from
the current equations of the branch equivalent in the phase
domain.
Submatrix: The current contribution corresponding
1)
to this (self-admittance) submatrix in the Fortescue domain is
derived by transforming first the phase voltage to the Fortescue
voltage and then the phase current to the Fortescue current as
follows:

(25)
where
is obtained from Table I and
is obtained from
(5).
Submatrix: The current contribution cor3)
responding to this (mutual admittance) submatrix in the
Fortescue domain is derived as follows, transforming first the
phase voltage to the Fortescue voltage, then the phase current
to the Fortescue current:

The Fortescue equivalent submatrix is therefore given by


(26)
Equation (26) is written in general form as
(27)
The Fortescue equivalent sub-matrix is therefore given by
(22)
Equation (22) can be written in general form as
(23)
Table I shows all possible phase combinations for a two-phase
line connected between three- and two-phase buses.
Submatrix: The current contribution cor2)
responding to this (mutual admittance) sub-matrix in the
Fortescue domain is derived as

where
is obtained from Table I and
is obtained from
(5).
Submatrix: This (self-admittance) submatrix is not
4)
affected by the mapping at bus and can therefore be derived
directly from (6).
B. Three-Phase to Single-Phase Lateral
This case is illustrated in Fig. 4. Bus is three-phase and bus
is single-phase. A special Fortescue transformation is again required for this branch between two different Fortescue domains.
Similarly to the previous case, the Fortescue equivalent for
this case can be written as follows:

(28)

The sub-matrix terms of this equation are derived in the following subsections.

198

IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 29, NO. 1, JANUARY 2014

AND

TABLE II
SUBMATRICES FOR DIFFERENT PHASE TYPES
Fig. 5. Two-phase to single-phase lateral.

AND

TABLE III
SUBMATRICES FOR DIFFERENT PHASE TYPES

1)
Sub-Matrix: The current contribution corresponding
to this (self admittance) submatrix in the Fortescue domain is
similarly derived as follows:

Equation (33) can be written in the general form as


The Fortescue equivalent submatrix is therefore given by

(34)
(29)

Equation (29) can be written in the general form as


(30)
Table II shows all possible combinations for a single-phase line
connected between three- and single-phase buses.
Submatrix: The current voltage relationship is
2)
expressed through admittance as

where
is obtained from Table II and
.
Submatrix: This (self-admittance) submatrix is not
4)
affected by the mapping at bus and can therefore be derived
directly from the phase domain.
C. Two-Phase to Single-Phase Lateral
This case is illustrated in Fig. 5, where bus is two-phase
and bus is single-phase. A special Fortescue transformation
is again required for this branch between these two different
Fortescue domains.
Similarly to the previous cases, the Fortescue equivalent
for this case can be written as follows:

(35)
The Fortescue equivalent submatrix is therefore given by
(31)

1)
Submatrix: Repeating similar steps to those used for
the three-phase to single-phase derivation, the Fortescue (selfadmittance) submatrix at bus is therefore given by

(32)

(36)

Equation (31) can be written in the general form as

is obtained from Table II and


.
where
Submatrix: The current contribution cor3)
responding to this (mutual admittance) submatrix in the
Fortescue domain is similarly derived as follows:

Equation (36) can be expressed in the general form as


(37)
Table III shows all possible combinations for a two-phase
branch connected between two and single phase nodes.
Submatrix: Following the similar approach for
2)
in Section IV-C-1, we have
(38)
and (38) can be written in the generic form as

The Fortescue equivalent submatrix is therefore given by

(39)
(33)

where

is obtained from Table III and

DAFI et al.: GENERALIZED

3)

FORTESCUE EQUIVALENT ADMITTANCE MATRIX APPROACH TO POWER FLOW SOLUTION

199

Submatrix: Similarly, we have


(40)

Equation (40) in general is expressed as


(41)
where
is obtained from Table III and
.
Submatrix: This (self-admittance) submatrix is not
4)
affected by the mapping at bus and can therefore be derived
directly from the phase domain.

Fig. 6. Fortescue

equivalent.

For branches connected to PTN nodes or special transformer connections, appropriate transformation matrices
as illustrated throughout Sections III and IV are necessary.

D. Special Transformer Connections


Transformer modeling using symmetrical components is
explained in [33]. The
equivalent in the Fortescue domain for open-wye and open-delta type transformer deserves
careful treatment. Typically, this transformer is used to supply
three-phase load connected to two-phase lateral. Phase domain
equivalent and dimensions of corresponding sub-matrices is
shown in

VI. POWER FLOW COMPUTATION


The generalized Fortescue approach can be applied to any
matrix-based power flow methods. The power flow calculation
based on the Fortescue modeling proposed in this paper was
tested using the IEEE test network model [32], using current
mismatches (iterations) method. The bus current equation can
be written as follows:
(45)

(42)

The Fortescue submatrices


and
are calculated from
corresponding phase domain submatrices using (6) and (4), respectively. The other two submatrices are calculated using a
similar approach to that used for three to two-phase connection,
as explained in Section IV-A.
V. GENERALIZED FORTESCUE

EQUIVALENT

The generalized nodal current injection equation for the


-phase branch connected between the -phase node and the
-phase node , where
, as shown in Fig. 6, is
(43)
where all the quantities are in Fortescue coordinates of appropriate order. Admittance submatrices are calculated in
(44)
Several important remarks are appropriate here.
Dimensions of submatrices
are defined by the
number of phases at nodes and .
Phase-domain branch equivalent is taken as input. Dimensions of these submatrices are defined by the number
.
of phases of the branch
The self-admittance and mutual admittance matrices of
each node are diagonal except for PTNs.
When the nodal admittance matrix of the entire system is
formed, it requires less storage compared with its phase
counterpart and LU factorization is therefore faster.

The nodal voltage magnitude and phase are described in the


flow chart shown in Fig. 7 and calculated as follows.
Step 1) Read input data and detect PTNs.
. FacStep 2) Create Fortescue equivalents and build
into
and
.
torize
as 1.0, depending on
Step 3) Assume initial node voltage
the number of phases in node . Choose the angle
or
for phase ,
appropriately, i.e. 0 ,
, or respectively.
Step 4) Where needed, compute the current injection from
. Conthe power injection, i.e.
vert
into Fortescue domain
; set
.
Step 5) Solve
for
; convert
into
and, where needed, compute the power:
, and the power error:
;
, stop; else go to step 4).
Step 6) If
Note that, for simplicity, regulation (e.g., LTC) has not been
included here, but it can be easily integrated.
VII. RESULTS AND DISCUSSIONS
Table IV lists the results for voltage magnitudes and angles of
all network nodes (Fig. 1) after conversion back to phase coordinates. The power flow calculations have been executed without
voltage regulation, and spot loads between nodes 632 and 671
have been modeled as five identical loads. For all buses, the differences between the voltage magnitudes and angles are very
). The comparison
small (respectively, 0.01 p.u. and
of the results shows that the Fortescue-based power flow yields
virtually the same results obtained through the ladder iterative
method [31] and the phase-coordinate method [34], which are
widely used in industry. Table IV shows results compared with
IEEE-13 reference results [32].

200

IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 29, NO. 1, JANUARY 2014

TABLE IV
BUS VOLTAGES FOR IEEE-13 NODE NETWORK

Fig. 7. Iterative power flow solution process.

The computation time for power flow calculation using


Fortescue and phase coordinates [34] have also been compared. The computation is carried out in a high-performance
PC (Intel i7-2600 K, 8 GB RAM). Table V summarizes the
results for different networks. Five test network models having,
respectively, 168, 1505, 2986, 8500, and 14 200 nodes were
considered. The test network models differ in number of buses
and the number of branches connected to PTN nodes. As can be
seen, the Fortescue solution takes 0.78 times that taken by the
three-phase solution for Net. I. For all remaining test networks,
the proposed approach takes less than 0.55 times the time taken
by the three-phase approach. The difference in performance
results for Net. I and the other test networks is related to the
proportion of branches which are connected to PTNs.
Generally, the gain of performance of the proposed approach
is for the three-phase main and three-phase laterals. This can be
observed on heavily loaded Net. IV, which is European urban
distribution network with few unsymmetrical lines. Minimum
node voltage for this network is 0.81 p.u.. The number of multiplications is reduced to a third compared with a calculation
in phase coordinates. In two-phase laterals, it is halved. On average, a reduction in computation time, for medium and large
networks, in the range of 30%50%, can be obtained. It is noticed that the number of iterations required for Fortescue power
flow is similar to the number of iterations required by the equivalent phase-domain power flow.

TABLE V
PERFORMANCE TEST RESULTS

VIII. CONCLUSION
The delivery of a smart distribution grid operation requires
fast computation and time-critical automation. Since the power
flow solution is the starting point to every network control decision, fast power flow solution in a large unbalanced distribution system is essential. A three-phase solution is good, but for
a system with many single- and two-phase laterals, a more effective algorithm such as the generalized Fortescue method is
necessary. Using the proposed algorithm based on Fortescue coordinates of dimension 1, 2 and 3, the properties of symmetrical
components are extended to power systems with one-, two-, and
three-phase components. The initialization of admittance matrix
of the Fortescue system only needs additional transformations at
the network location, to which components of different phases
are connected. The required transformations between the phase

DAFI et al.: GENERALIZED

FORTESCUE EQUIVALENT ADMITTANCE MATRIX APPROACH TO POWER FLOW SOLUTION

coordinates and Fortescue coordinates are defined. The adaptations, which are needed for Fortescue operations, have also been
discussed. The main advantage of the approach presented in this
paper is the increase in speed of power flow solution. This approach has been validated in the IEEE-13 node test feeder model
and a variety of other distribution network models of different
sizes. A 14 200-node distribution system model is solved by the
method in less than 500 ms. This clearly suggests that the proposed technique holds great potential for real-time control of the
distribution network.

ACKNOWLEDGMENT
The authors would like to acknowledge all of the anonymous
reviewers and editor for their valuable comments and suggestions that have helped to improve the quality of the content of
this manuscript and have enhanced the authors insight into this
research topic.

REFERENCES
[1] C. L. Fortescue, Method of symmetrical co-ordinates applied to the
solution of polyphase networks, Trans. Amer. Inst. Electr. Eng., vol.
37, no. 2, pp. 10271140, Jun. 1918.
[2] P. M. Anderson, Analysis of Faulted Power Systems. New York, NY,
USA: IEEE, 1995.
[3] J. L. Blackburn, Symmetrical Components for Power Systems Engineering. Boca Raton, FL, USA: CRC, 1993.
[4] R. G. Wasley and M. A. Shlash, Newton-Raphson algorithm for
3-phase load flow, Proc. Inst. Electr. Eng.Gen, Transm. Distrib.,
vol. 121, no. 7, pp. 630638, Jul. .
[5] J. Arrillaga and B. J. Harker, Fast-decoupled three phase load flow,
Proc. Inst. Electr. Eng.Gen, Transm. Distrib., vol. 125, no. 8, pp.
734740, Aug. 1978.
[6] A. V. Garcia and M. G. Zago, Three-phase fast decoupled power flow
for distribution networks, Proc. Inst. Electr. Eng.Gen, Transm. Distrib., vol. 143, no. 2, pp. 188192, Mar. 1996.
[7] H. Le Nguyen, Newton-Raphson method in complex form [power
system load flow analysis], IEEE Trans. Power Syst., vol. 12, no. 3,
pp. 13551359, Aug. 1997.
[8] K. A. Birt, J. J. Graffy, J. D. McDonald, and A. H. El-Abiad, Three
phase load flow program, IEEE Trans. Power App. Syst., vol. PAS-95,
no. 1, pp. 5965, Jan. 1976.
[9] B. K. Chen, M. S. Chen, R. R. Shoults, and C. C. Liang, Hybrid three
phase load flow, Proc. Inst. Electr. Eng.Gen, Transm. Distrib., vol.
137, no. 3, pp. 177185, May 1990.
[10] T. H. Chen, M. S. Chen, K. J. Hwang, P. Kotas, and E. A. Chebli,
Distribution system power flow analysisA rigid approach, IEEE
Trans. Power Del., vol. 6, no. 3, pp. 11461152, Jul. 1991.
[11] P. A. N. Garcia, J. L. R. Pereira, S. Carneiro, Jr., V. M. da Costa, and
N. Martins, Three-phase power flow calculations using the current injection method, IEEE Trans. Power Syst., vol. 15, no. 2, pp. 508514,
May 2000.
[12] J. H. Teng, A direct approach for distribution system load flow
solutions, IEEE Trans. Power Del., vol. 18, no. 3, pp. 882887,
Jul. 2003.
[13] W. M. Lin and J. H. Teng, Phase-decoupled load flow method for
radial and weakly-meshed distribution networks, Proc. Inst. Electr.
Eng.Gen, Transm. Distrib., vol. 143, no. 1, pp. 3942, Jul. 1996.
[14] K. L. Lo and C. Zhang, Decomposed three-phase power flow solution
using the sequence component frame, Proc. Inst. Electr. Eng.Gen,
Transm. Distrib., vol. 140, no. 3, pp. 181188, May 1993.
[15] V. C. Strezoski and L. D. Trpezanovski, Three-phase asymmetrical
load-flow, Electr. Power Energy Syst., vol. 22, no. 7, pp. 511520,
Oct. 2000.

201

[16] V. C. Strezoski, Advanced symmetrical components method, IET


Gen., Transm. Distrib., vol. 5, no. 8, pp. 833841, Aug. 2011.
[17] X. P. Zhang and H. Chen, Asymmetrical three-phase load-flow
study based on symmetrical component theory, Proc. Inst. Electr.
Eng.Gen, Transm. Distrib., vol. 141, no. 3, pp. 248252, May
1994.
[18] X. P. Zhang, Fast three phase load flow methods, IEEE Trans. Power
Syst., vol. 11, no. 3, pp. 15471554, Aug. 1996.
[19] M. A. Akher, K. M. Nor, and A. H. A. Rashid, Development of unbalanced three-phase distribution power flow analysis using sequence and
phase components, in Proc. 12th Int. Middle-East Power Syst. Conf.,
Aug. 2008, pp. 406411.
[20] B. C. Smith and J. Arrillaga, Improved three-phase load flow using
phase and sequence components, Proc. Inst. Electr. Eng.Gen,
Transm. Distrib., vol. 145, no. 3, pp. 245250, Mar. 1998.
[21] M. Abdel-Akher and K. Mahmoud, Unbalanced distribution
power-flow model and analysis of wind turbine generating systems, Eur. Trans. Electr. Power Mar. 2012 [Online]. Available:
http://dx.doi.org/10.1002/etep.1634
[22] M. Z. Kamh and M. R. Iravani, A three-phase power flow method for
real-time distribution system analysis, IEEE Trans. Smart Grid, vol.
3, no. 2, pp. 828836, May 2012.
[23] S. Bruno, S. Lamonaca, G. Rotondo, U. Stecchi, and M. La Scala, Unbalanced three-phase optimal power flow for smart grids, IEEE Trans.
Ind. Electron., vol. 58, no. 10, pp. 45044513, Oct. 2011.
[24] M. A. Akher, K. M. Nor, and A. H. A. Rashid, Improved three-phase
power-flow methods using sequence components, IEEE Trans. Power
Syst., vol. 20, no. 3, pp. 13891397, Aug. 2005.
[25] S. Ghazali, K. M. Nor, and M. A. Akher, Parallel sequence decoupled
full Newton-Raphson three phase power flow, in Proc. IEEE Region
10 Conf., Jan. 2009, pp. 16.
[26] J. M. T. Marinho and G. N. Taranto, A hybrid three-phase single-phase
power flow formulation, IEEE Trans. Power Syst., vol. 23, no. 3, pp.
10631070, Aug. 2008.
[27] C. S. Cheng and D. Shirmohammadi, A three-phase power flow
method for real-time distribution system analysis, IEEE Trans.
Power Syst., vol. 10, no. 2, pp. 671679, May 1995.
[28] D. Shirmohammadi, H. W. Hong, A. Semlyen, and G. X. Luo, A compensation-based power flow method for weakly meshed distribution
and transmission networks, IEEE Trans. Power Syst., vol. 3, no. 2,
pp. 753762, May 1988.
[29] W. H. Kersting, Distribution System Modeling and Analysis. Boca
Raton, FL, USA: CRC, 2002.
[30] E. R. Ramos, A. G. Expsito, and G. A. Cordero, Quasi-coupled threephase radial load flow, IEEE Trans. Power Syst., vol. 19, no. 2, pp.
776781, May 2004.
[31] I. Dzafic, H.-T. Neisius, M. Gilles, S. Henselmeyer, and V. Landerberger, Three-phase power flow in distribution networks using
Fortescue transformation, IEEE Trans. Power Syst., vol. 28, no. 2,
pp. 10271034, May 2013.
[32] W. H. Kersting, Radial distribution test feeders, in Proc. IEEE/Power
Eng. Soc. Winter Meeting, May 2001, vol. 2, pp. 908912.
[33] H. T. Neisius and I. Dzafic, Three-phase transformer modeling using
symmetrical components, in Proc. IEEE PES Innovative Smart Grid
Technol., Jan. 2011, pp. 16.
[34] J. C. Das, Power System AnalysisShort Circuit, Load Flow and Harmonics. New York, NY, USA: Marcel Dekker, 2002.

Izudin Dafi (M05SM13) received the Ph.D. degree from the University of
Zagreb, Zagreb, Croatia, in 2002.
He is currently with Siemens AG, Nuremberg, Germany, where he is Head
of the Department and Chief Product Owner (CPO) for Distribution Network
Analysis (DNA) R&D. His research interests include power system modeling,
development and application of fast computing to power systems simulations.
Dr. Dafi is a member of the IEEE Power and Energy Society and the IEEE
Computer Society.

202

Bikash C. Pal (M00SM02F13) received the B.Eng degree from Jadavpur


University, Calcutta, India, in 1990, the M.Eng degree from The Indian Institute of Science Bangalore, India, in 1992, and the Ph.D. degree from Imperial
College London, London, U.K., in 1996, all in electrical engineering.
He is currently a Professor of power systems with Imperial College London,
London, U.K. His research interest focuses on robust control of power system,
wide-area monitoring, and analysis and distribution system state estimation and
FACTS controllers.
Prof. Pal is Editor-in-Chief of the IEEE TRANSACTIONS ON SUSTAINABLE
ENERGY.

IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 29, NO. 1, JANUARY 2014

Michel Gilles (SM90) received the M.Sc. and Ph.D. degrees from Wayne State
University, Detroit, MI, USA, in 1979 and 1981, respectively.
He is currently a Senior Principal/Control Center with Siemens AG, Nuremberg, Germany, where his responsibilities include the development and implementation of product line and technology strategies. Prior to joining Siemens, he
held various engineering and management positions in the operation and R&D
domains of the SCADA/DMS/EMS and Energy Market Management (EMM)
business in the United States (20 years) and in Europe (10 years).

Sylwia Henselmeyer (M13) received the Diploma in computer science from


the University of Erlangen-Nuremburg, Nuremberg, Germany, in 2005.
She is a Senior Software Developer with Siemens AG, Nuremberg, Germany,
working in distribution system applications. Her research interests are state estimation, numerical programming, computer graphics, and object-oriented analysis and design.

Das könnte Ihnen auch gefallen