Electric Power Systems Research 197 (2021) 107312

Contents lists available at ScienceDirect

Electric Power Systems Research

journal homepage: www.elsevier.com/locate/epsr

A multi-star synchronous machine model for real-time digital simulation

and its applications
A. B. Dehkordi *, T. L. Maguire
RTDS Technologies Inc., Winnipeg, Canada

A R T I C L E I N F O A B S T R A C T

Keywords: Poly-phase electric machines have the advantages of low ripple in torque, better fault tolerance and high reli­
Poly-phase ability in operation. Such properties make them attractive for applications in renewable energies, electric ship
Multi-star synchronous machine propulsion, traction, aircraft systems and electric vehicles [1,2]. Failure of the drive system through a few phases
Symmetrical components
could be tolerated by operating the remaining phases of the machine. One popular group amongst poly-phase
VSD
Two reaction theory
electric machines is the category of multi-star machines.
Real-time digital simulation This paper presents development and validation of a multi-star synchronous machine model for a real-time
Digital transient network analyzer digital simulator. The paper starts with the introduction and analysis of multi-star machines. A generalized
Wind generation method of Vector-Space Decomposition (VSD) is introduced for analyzing machines with an arbitrary number of
Electric ship stars. The method of incorporating the model into the network solution of real-time electromagnetic transient
program is described. For the purpose of demonstration, a typical circuit of a marine electric vessel, containing
poly-phase generators, power electronic converters, DC bus, hotel load, battery, and poly-phase propulsion
systems is simulated in real time.

1. Introduction through power electronic converters.

For the above reasons, poly-phase machines have applications in
In conventional power system applications, the number of phases is electric ship propulsion, traction, aircraft systems, electric vehicles [2,3]
three as it generally results in a lower transmission costs. Historically 3- and new schemes of wind turbines. There are variations of poly-phase
phase systems are selected for transmission and electric machines in machines such as split-phase, multi-star, higher order symmetrical,
both generation and loads. When it comes to applications where the etc., which can be studied in literature [2,3].
electric machine is connected to power electronic converters, restriction A real time digital simulator is a combination of computer hardware
to a system with 3 phases is not necessary [1]. and software designed specifically for the solution of power system and
Historically, electric machines with higher number of phases (poly- power electronic electromagnetic transients in real time. There are
phase machines) were introduced to lower the ripples in electric torque, many areas where this technology has been successfully applied.
and to lower current ratings per phase. In fact, the later was a reason that Simulation of electric networks in marine vessels is one of such appli­
promoted the development of multi-phase machines in early stages of cations. Modeling of poly-phase machines for real-time digital simula­
circuit breaker development. Additionally, lower current rating per tion is an important step in enhancing these simulators for marine
phase allows the deployment of power electronic converters with lower applications.
ratings. Modeling of poly-phase machines poses several challenges, espe­
In addition to lower torque ripple and current ratings, poly-phase cially when it comes to real-time digital simulation. One of the main
machines have other advantages such as higher force density, reduc­ problems is the diversity of these models; even if the phases are dis­
tion in the noise characteristic and the stator copper losses [2]. Addi­ placed in a symmetric manner, the number of phases may vary between
tionally, poly-phase machines have greater reliability and fault 3 to 18 or even more [4]. Developing and maintaining a large number of
tolerance. Several phases can be lost yet the machine can still be oper­ machine models in the library of any commercial software is challenging
ated utilizing the remaining phases by providing proper excitation and time-consuming.

* Corresponding author.
E-mail addresses: [email protected] (A. B. Dehkordi), [email protected] (T. L. Maguire).

https://doi.org/10.1016/j.epsr.2021.107312
Received 20 October 2020; Received in revised form 19 April 2021; Accepted 23 April 2021
Available online 15 May 2021
0378-7796/© 2021 Elsevier B.V. All rights reserved.
A. B. Dehkordi and T. L. Maguire Electric Power Systems Research 197 (2021) 107312

Fig. 1. Idealized winding configuration for a multi star 6-phase synchro­ Fig. 2. Idealized winding configuration for a fundamental arrangement 6-
nous machine. phase synchronous machine.

A comprehensive analysis in rotor frame of reference is necessary as 2.1. General description of a multi-star synchronous machine
it responds to the question of general analysis of poly-phase machines.
Note that, in practical applications, these machines are driven by power As described, in a multiple-star machine, stator is made of l winding
electronic converters with high frequency switching; therefore, models sets (stars) with k phases for each winding set. Naturally, number of
need to be implemented with a smaller simulation time-step of a few phases is N = l.k. In such machines, the angular displacement between
microseconds. This makes the modeling in rotor frame of reference more winding sets is commonly designed to be π /N. As an example, an
attractive. idealized winding diagrams for a 6-phase multi-star (dual-star) machine
As mentioned earlier, there are various types of poly-phase machines is shown in Fig. 1. The diagram shows the axis of each winding, which is
models; two main categories are symmetrical multi-phase machines and defined as the axis of MMF produced by that winding.
multiple-star machines. In symmetrical multiphase machines, N stator
phases are distributed symmetrically in the stator space with the angular 2.2. Multi-star winding configuration and transformation to fundamental
displacement of 2π/N. In multiple-star machines, stator is made of l winding configuration
winding sets (stars) with k phases for each winding set. Naturally,
number of phases is of k.l. In multiple star machines, the angular Consider the simplified arrangement of a symmetrical poly-phase
displacement between winding sets is commonly designed to be π /N. stator. Several sets of coils distributed symmetrically around the pe­
There are other types of multi-star machine where the angular riphery of the two-pole stator covering 360◦ . This is generally called two-
displacement between winding sets is arbitrary. Modeling of such ma­ pole symmetry. The fundamental winding arrangement with 180◦ pro­
chines is not in the focus of this paper. gression (shown in Fig. 2.) is referred to as single-pole symmetry, or more
This paper presents the development of a multi-star synchronous simply, pole-symmetry [9]. From design point of view, such a winding
machine model for real-time digital simulation. The paper starts with arrangement works as well as the full progression winding arrangement.
the introduction and analysis of multi-star synchronous machines. Dif­ In fact, a single-phase induction machine, which is a two-phase machine
ferential equations of a multi-star synchronous machine are described with 90◦ angular shift between the winding sets, has a pole-symmetry.
and challenges of modeling and analysis in these machines are outlined. It is shown in [7-9] that a multi-star arrangement can be transformed
The so-called method of Vector Space Decomposition (VSD) is pro­ to a symmetrically distributed winding arrangement with 180◦ pro­
posed for extending the two-reaction theory [5,6] to multi-star ar­ gression called fundamental winding configuration. Fig. 2 shows the
rangements [7-10]. A general equivalent circuit in rotor frame of fundamental winding arrangement for a 6-phase machine. The purpose
reference is presented from which a transient multi-star synchronous of this diagram is to show the axes of windings in this arrangement. As
machine model is developed and incorporated into the network solution can be seen in this figure, six windings are distributed with a spatial
of a real-time digital simulator. Capabilities of the model such as flexi­ angular shift of δ1 = 180◦ /6 = 30◦ , and a total progression of 180◦ .
bility in selecting the number of stars are described. Additionally, the Variables in multi-star configurations such as voltages, currents and
method of simulating power electronic converters based on a predictive flux linkages can be easily transformed into corresponding variables in
switching algorithm is introduced. To demonstrate the capabilities of fundamental winding arrangement [7,9]. As an example, for a dual-star
the model, a typical circuit of marine electric vessel, containing 6-phase machine (k=3, l=2), the relation between the multi-star vari­
poly-phase generators, power electronic converters, DC bus, hotel load, ables and fundamental winding variables are shown in (1).
battery, and poly-phase propulsion systems is simulated in real time ( )T
with a time-step of 3 µs, and the results are shown. fa fb fc fd fe ff = W2×3 ⋅(fA1 fB1 fC1 fA2 fB2 fC2 )T
where : ⎡ ⎤
1 0 0 0 0 0
2. Analysis of multi-star synchronous machines ⎢0 0 0 1 0 0 ⎥
⎢
⎢0 0 − 1 0 0 0 ⎥
⎥ (1)
This section briefly describes the analysis and modeling method of [W2×3 ] = ⎢ ⎢0 0 0 0 0 − 1⎥
⎥
⎢ ⎥
multi-star synchronous machines. Since the goal is incorporation of the ⎣0 1 0 0 0 0 ⎦
model into an electromagnetic transient program [11], the coupled 0 0 0 0 1 0
electric circuit approach is used for modeling of this type of machine.
The machine is considered as an idealized machine assuming sinusoidal
distribution for the magneto-motive forces and permeance.

2
A. B. Dehkordi and T. L. Maguire Electric Power Systems Research 197 (2021) 107312

2.3. Voltage and flux linkage equations For multiple star machines, a similar approach can be adapted. It is
shown in the previous section that, any multi-star winding arrangement
Based on the above description of a multiple-star synchronous ma­ can be transformed to a fundamental winding arrangement with 180◦
chine, voltage and flux linkage equations are developed. The following phase progression [7-9]. Based on the analogy in the previous para­
assumptions are made in this analysis: graph, a new symmetrical component transformation for windings with
180◦ phase progression can be developed, and consequently, αβ and dq
• A multi-pole synchronous machine is modeled as an equivalent two- transformations can be established [7,9]. These transformations diago­
pole machine. nalize the inductance matrix of machines with 180◦ phase progression
• It is assumed that the machine windings produce a sinusoidal MMF, windings. As an example, for a dual-star 6-phase machine (k=3, l=2),
and permeance is sinusoidally distributed; therefore, space har­ the general two-phase (αβ) transformation from fundamental winding
monics are ignored. Basically, the machine is assumed an idealized variables to αβ components are shown in (5). In this transformation, the
machine. first and the last row correspond to the fundamental αβ components.
• Although the magnetizing MMF is assumed sinusoidal, effects of Other rows correspond to harmonic components. In the absence of
space harmonics in cross-coupling leakage between stator winding, winding spatial harmonics, these components do not produce a
(i.e. differential leakage) is considered. That means that, the leakage revolving field; therefore, they can be considered zero sequences. Note
inductance matrix of stator may have off-diagonal elements that are that, if the number of phases is odd, there will be a row in this trans­
taken into consideration. formation corresponding to the homo-polar zero sequence.
• By definition, stator is made of l winding sets (stars) with k phases for ( )T
each winding set. Number of phases is N = l.k, and the angular (fα1 fα3 fα5 fβ5 fβ3 fβ1 )T = Cαa ⋅ fa fb fc fd fe ff
displacement between winding sets is π/N. Phase resistances are where :
equal, and leakage inductances are symmetrical with respect to each ⎡ ⎤
winding. 1 cos(δ1) cos(2δ1) cos(3δ1) cos(4δ1) cos(5δ1)
⎢ ⎥
⎢1 cos(3δ1) cos(6δ1) cos(9δ1) cos(12δ1) cos(15δ1) ⎥
⎢ ⎥
Voltage and flux linkage equations for a N-phase multi-star syn­ ⎢ ⎥
( )⎢ ⎥
chronous machine are presented in (2). Equation (2) shows the relation [ a] 2 ⎢⎢
1 cos(5δ1) cos(10δ1) cos(15δ1) cos(20δ1) cos(25δ1) ⎥
⎥
between vectors of voltages, currents and flux linkages of the machine. Cα = ⋅⎢ ⎥
N ⎢ sin(5δ1) sin(10δ1) sin(15δ1) sin(20δ1) sin(25δ1) ⎥
⎢0 ⎥
Here, vectors of stator voltages, currents and flux linkages are by ⎢ ⎥
⎢ ⎥
vA1B1…Kls , iA1B1…Kls and ψ A1B1…Kls respectively. A similar arrangement is ⎢0 sin(3δ1) sin(6δ1) sin(9δ1) sin(12δ1) sin(15δ1) ⎥
⎣ ⎦
made for rotor quantities.
0 sin(δ1) sin(2δ1) sin(3δ1) sin(4δ1) sin(5δ1)
⎛ ⎞ ( )⎛ ⎞ ⎛ ⎞
v [rs ] [0] i ψ A1B1…Kls π
⎝ A1B1…Kls ⎠ = ⎝ A1B1…Kls ⎠ + d ⎝ ⎠ δ1 =
vdqr [0] [rs ] idqr dt ψ dqr 6
(3)
⎛ ⎞ ( )⎛ ⎞
⎝
ψ A1B1…Kls
⎠=
[Lss ] [Lsr ]
⎝
iA1B1…Kls
⎠ Based on the above, the following methodology is used to analyze
ψ dqr [Lsr ]T [Lrr ] idqr multi-star machines: first transform the variables to fundamental
(2) winding variables, then use αβ or dq transformations to decouple the
where : equations. After solving the equations in the decoupled form (rotor
( )T
frame of reference), transform them back to fundamental winding
f A1B1…A2B2…AlBl…s = (fA1 fB1 … fA2 fB2 …… fAl fBl …)
quantities and then eventually obtain all multi-star quantities.
( )T
( ) Voltage and flux linkage equations for an N-phase synchronous
f dqr = fdr1 fdr2 …fqr1 fqr2 … machine with 180◦ phase progression are presented in (4). Here, vectors
of stator voltages, currents and flux linkages are shown by vab…ns , iab…ns
When the number of phases per set is 3 (i.e. k=3), the well-known 3- and ψ ab…ns respectively. Self-inductance matrices of stator and rotor are
phase Park transformation with suitable angular shifts may be used to shown by [Lss] and [Lrr]. [Lsr] is the matrix for the mutual inductances
transform the phase domain equations of (2) into rotor frame of refer­ between the stator and rotor windings. The values of these inductances
ence [12,13]. One of the drawbacks of this method is that due to Toeplitz are functions of rotor position θr and saturation. As an example, the
structure of the leakage inductance matrix, cross-coupling terms may stator inductance matrix [Lss] is shown in (5). Inductance values M0s and
appear between the d- and q- equivalent circuits [7,12]. Furthermore, L2s are functions of winding distribution, effective number of turns,
the method is not applicable if the number of phases per set is not 3 (k ∕
= effective air-gap length and saturation. As mentioned previously, higher
3). order space harmonics of the machine are ignored.
A more general approach used here, is the so-called vector space ⎛ ⎞ ( )⎛ ⎞ ⎛ ⎞
decomposition (VSD). The basis of this approach is the fact that sym­ v [rs ] [0] i ψ ab…ns
⎝ ab…ns ⎠ = ⎝ ab…ns ⎠ + d ⎝ ⎠
metrical component transformation introduced by Fortescue [14] is vdqr [0] [rs ] idqr dt ψ dqr
applicable to systems with any number of phases. It is also well known
⎛ ⎞ )⎛ ⎞
that, this transformation diagonalizes circulant (cyclic symmetric) ψ ab…ns
(
[Lss ] [Lsr ] iab…ns
matrices [15,16]. One can prove that, due to symmetry, the inductance ⎝ ⎠= ⎝ ⎠
ψ dqr [Lsr ]T [Lrr ] idqr
matrix of a symmetrical poly phase machine with 360◦ phase progres­ (4)
sion is cyclic symmetric. Therefore, symmetrical component trans­ where :
formation diagonalizes the stator inductance matrix of a symmetrical ( )T
poly phase machine with any number of phases. Ku and White extracted f ab…ns = (fas fbs …fns )
αβ and dq transformations based on this property for any number of ( )T
phases [15]. These transformations also diagonalize the inductance ( )
f dqr = fdr1 fdr2 …fqr1 fqr2 …
matrix of symmetrical poly-phase machines.

3
A. B. Dehkordi and T. L. Maguire Electric Power Systems Research 197 (2021) 107312

⎡ ⎤
l0 l1 l2 ⋯ − l3 − l2 − l1
⎢ l1 l0 l1 ⋯ − l4 − l3 − l2 ⎥
⎢ ⎥
⎢ l2 l1 l0 ⋯ − l5 − l4 − l3 ⎥
⎢ ⎥
[Lss− ⎢
l] = ⎢ ⋯ ⋯ ⋯ ⋯ ⋯ ⋯ ⎥ ⎥ (6)
⎢ − l3 − l4 − l5 ⋯ l0 l1 l2 ⎥
⎢ ⎥
⎣ − l2 − l3 − l4 ⋯ l1 l0 l1 ⎦
− l1 − l2 − l3 ⋯ l2 l1 l0
⎡ ⎤
Lls− h1 0 0 ⋯ 0 0 0
⎢ ⎥
⎢ 0 Lls− 0 ⋯ 0 0 0 ⎥
⎢ h3 ⎥
⎢ ⎥
⎢ ⎥
⎢ 0 0 Lls− h5 ⋯ 0 0 0 ⎥
⎢ ⎥
dq
⎢ ⎥
[Lss− l] =⎢
⎢ ⋯ ⋯ ⋯ Lls− hm ⋯ ⋯ ⋯ ⎥
⎥
⎢ ⎥
⎢ 0 0 0 ⋯ Lls− 0 0 ⎥
⎢ ⎥
(7)
h5
⎢ ⎥
⎢ ⎥
⎢ 0 0 0 ⋯ 0 Lls− h3 0 ⎥
⎣ ⎦
0 0 0 ⋯ 0 0 Lls− h1
( )
N− 1
truc
∑2
Lls− hm = l0 + 2⋅ lk ⋅cos(m⋅k⋅δ1)
k=1

2.4. Equivalent circuit in DQ frame of reference

The DQ Equivalent circuit of an N-phase synchronous machine with

180◦ phase progression is shown in Fig. 3. This circuit is effectively
applicable to a multi-star synchronous machine as well. Note that, in
addition to the d- and q- axis equivalent circuits, there exist other
equivalent circuits corresponding to harmonics (3, 5, ...). In an idealized
machine with perfectly sinusoidally wound stator, these harmonics do
not produce a revolving field. In this case, these circuits are effectively
Fig. 3. Equivalent circuit of an idealized N-phase synchronous machine with zero-sequence circuits, as they do not couple with the rotor. Note that if
180◦ phase progression. the number of phases is odd, there will also exist a homo-polar zero
sequence circuit. Leakage inductance values Lls-h1, Lls-h3, and Lls-h5 are
the leakage inductances corresponding to each harmonic. These values
2.3.1. Leakage inductance matrix are the diagonal elements of the leakage matrix after the application of
The leakage inductance matrix of the stator is shown by [Lss-l]. In αβ transformation. Note that, the model introduced in this paper re­
symmetrical multi-phase machines, this matrix usually has a circulant ceives these values as input, and prints the full phase-domain leakage
(cyclic symmetric) structure whereas for windings with single-pole matrix of the stator in a separate file for the users.
symmetry, it has a Toeplitz structure [7]. The existence of non-zero It is shown that, in multi-phase machines N>3, the method of VSD
off-diagonal elements may come as a surprise to some; although the can make a position-independent inductance matrix in dq frame of
magnetizing MMF is assumed sinusoidal, effects of space harmonics in reference, even in the presence of some of the space harmonics [1, 7,17].
cross-coupling leakage between stator winding, (i.e. differential In this case, these harmonics contribute to the revolving field, thus the
leakage) is considered. This causes the leakage inductance matrix to harmonic equivalent circuits in Fig. 3 need to be modified to account for
have off-diagonal elements [7]. The leakage values l0, l1 and so on in (6) the coupling of stator and rotor circuit. In the model presented here,
are functions of winding distributions and dimensions. these effects are ignored, since such data is not available to most users.
Application of symmetrical component or αβ transformations with
180◦ phase progression diagonalizes this Toeplitz-structured leakage
3. Implementation of the multi-star synchronous machine
inductance matrix [7]. This is demonstrated in (7). Note that, if the
model into the environment of the real-time digital simulator
off-diagonal elements of the leakage inductance matrix are not avail­
able, or if these elements are not intended for consideration, then users
Each unit of the real-time digital simulator used in this paper, con­
may consider a diagonal leakage inductance matrix. The elements of this
sists of processors or cores dedicated to the solution of the power system
diagonal matrix are all equal to the stator self-leakage.
networks, power system devices, and control components [18]. Each
[Lss ] = [Lss− l ] + [Lss− oh ] + [Lss− 2h ] simulation time-step is divided into computation intervals and
communication intervals. Similar to other electromagnetic transient
where :
⎧ ( ) programs such as EMTP and EMTDC, Dommel algorithm [11] is used for
2
⎪
⎪ [L ss− 0h ] = ⋅M0s ⋅cos[(k − i)δ] discretization of the electric system’s differential equation. In this
⎪
⎨ ik
N
method, passive RLC elements of the circuit, as well as, more sophisti­
(5)
⎪
⎪
( )
2 cated components such as, machines, transformers and power elec­
⎪
⎩ [Lss−
2h ]ik = ⋅L2s ⋅cos[2θr − (i + k − 2)δ] tronics are represented by companion circuits in the form of current
N
( ) sources and conductances [11]. Subsequently, the nodal solution is used
δ=
2π
i, k ∈ 1, 2…N for calculation of node voltages. Due to the fact that, multi-phase ma­
N chines are usually connected to power electronic converters with high
frequency switching, usually a small simulation time-step of a few

4
A. B. Dehkordi and T. L. Maguire Electric Power Systems Research 197 (2021) 107312

Table 1
Execution time vs number of phases
Number of Phase 3 6 9 12
Execution Time (ns) 446 690 997 1330

Fig. 5. Phase-phase fault on a 2-star (6-phase) synchronous machine. (a) Field

winding current. (b) Stator phase A1 current. Results overlap.

validity of the model for a synchronous machine with multiple stars, a 2-

star and a 3-star synchronous machine are modeled. Transient behavior
of the model is studied under both symmetrical and asymmetrical faults.
Fig. 4. 6-phase fault on a 2-star (6-phase) synchronous machine. (a) Field
winding current. (b) Stator phase A1 current. Results overlap. 4.1. Symmetrical fault on a 2-star (6-Phase) synchronous machine

microseconds needs to be used to model systems containing such In this section, the proposed multi-star synchronous machine model
machines. is configured to be a 2-star synchronous machine with 3 phases per star.
During the simulation, while the rotor speed is set to synchronous, the
machine is operating in open-circuit and its field voltage is adjusted such
3.1. Features and capabilities of the model
that the terminal voltage of 1 pu is achieved (1 normal field voltage).
After reaching steady state conditions, a solid 6-phase short-circuit is
This section describes the capabilities of the model proposed here.
applied to the terminals of the machine. The fault is applied at the time
The modeling technique can be applied to any number of stars and any
that phase A1 voltage crosses zero. Variations of the field current and the
number of phases per star. The current implemented model, however,
stator phase A1 current of the machine for the first 3.0 seconds of the
only considers 3-phase stars. The number of stars for the machine can be
short circuit is shown in Figures 4a and 4b respectively. In each figure,
selected by the user from a menu. Based on the selected number of
there are two curves, one corresponding to the real-time model labeled
phases, stator terminals and neutral will be available for external
as “Proposed Model”, and the other corresponding to the off-line phase
connection. Additionally, an option is available to access both ends of
domain solution labeled as “Off-Line Solution”.
each stator winding. This option allows users to connect both ends of
As shown in Fig 4, the two curves are indistinguishable, establishing
stator windings to power electronic converters as desired in particular
the fact that the proposed model is equivalent to the off-line phase-
configurations [2].
domain machine model. Consequently, the integrity of transformations
Mechanical swing equations can be solved either inside the model or
used in this paper is validated.
outside using a multi-mass model. Magnetic saturation is also incorpo­
rated into this model [19]. The machine accepts d- and q-axis data both
in the form of reactances and time constants. Another menu receives
machine zero sequence reactances and resistances in pu.
Table 1 shows the approximate execution time (on a IBM Power 8
processor [18]) for the machine with different number of phases.

4. Validation of the model and time-domain simulation results

This section provides validations for the multi-star synchronous

machine model presented in this paper. The focus of this section is to
provide validations for the capability of the proposed model in accu­
rately solving the differential equations of a symmetrical multi-phase
synchronous machine stated in (2) - (5). To validate the performance
of the proposed model in transient conditions, a small off-line stand-
alone electromagnetic transient program including a stand-alone multi-
star synchronous machine model is developed. The differential equa­
tions of the multi-star synchronous machine are solved with the more
accurate phase domain approach [20] with the simulation time step of
1.0 µs.
The data for the synchronous machine under the test is shown in
Appendix I. A larger salient-pole synchronous machine such as the one
used here provides more distinct transient and sub-transient time con­ Fig. 6. Single-phase fault on a 3-star (9-phase) synchronous machine. (a) Field
stants, and consequently a more thorough validation. To check the winding current. (b) Stator phase A1 current. Results overlap.

5
A. B. Dehkordi and T. L. Maguire Electric Power Systems Research 197 (2021) 107312

4.3. Asymmetrical fault on a 3-star (9-Phase) synchronous machine

The experiment is repeated with a 3-star synchronous machine with

3 phases per star. In steady-state, a solid phase-neutral short-circuit is
applied to terminal A1 of the machine at the zero-crossing of the phase
Fig. 7. The diagram for the simulated electric ship circuit. A1 voltage. Other phases were left open. The simulation results for the
field current and the stator phase A1 current are shown in Figures 6a and
4.2. Asymmetrical fault on a 2-star (6-Phase) synchronous machine 6b respectively. As can be seen again, the results are identical to the off-
line phase domain solution.
In this section, again the 2-star (6-phase) machine is used. The above
experiment in Section IV-A is repeated with a solid A phase-phase short- 5. An application example of the multi-star synchronous
circuit. Note that applying a phase-phase fault excites the zero sequence machine model in electric ship systems
circuits in addition to the d- and q-axes circuits.
In steady-state, a solid phase-phase short-circuit is applied to each In this section, a typical electric network of a marine vessel [21]
terminal A of the two stars at the zero-crossing of the phase A1 voltage. consisting of a dual star generator, rectifiers, a DC bus, battery storage,
Other phases were left open. Variations of the field current and the stator hotel load and propulsion system is simulated in real-time. The circuit
phase A1 current of the machine for the first 3.0 seconds of the short diagram is depicted in Fig. 7.
circuit is shown in Figures 5a and 5b respectively. As can be seen again, The generator is a 4.0 MVA, 1.6 kV, 50 Hz, dual-star 6-phase syn­
the results are matching. chronous machine. Its voltage is rectified using two 3-phase diode rec­
tifiers. The DC bus voltage is maintained at 1.8 kV in normal operation of
the system by means of the generator excitation system. The battery
system and a hotel load of 75 kW is connected to the DC bus. The pro­
pulsion system consists of two VSC-based converters driving a perma­
nent magnet synchronous motor and a mechanical load. The motor is a
1.6 MW, 0.69 kV, 6-phase dual-star PMSM. The torque and ultimately
the speed of the motor is controlled using the dq decoupling method.
The q- component of the stator current is used to control the torque,
while the d- component is used to control the reactive power absorbed
by the motor. The details of control strategy are out of the scope of this
paper and are not described here. The control principles are adapted
from the control techniques used in a typical PMSM wind turbine
generator [22]. The circuit, including multi-star machines, power elec­
tronic converters and local generation of gating signals is simulated in
real-time with the time step of 3 µs.

5.1. Switched resistance representation of power electronic converters

Earlier generations of real-time power electronic converters for high

frequency switching are usually emulated by using an L/C associated
discrete (L/C-ADC) in a fixed time-step simulation [23]. This method has
the advantage of maintaining constant Dommel equivalent admittance
for ON and OFF state of the switches. However, it causes unrealistically
Fig. 8. Dual star generator steady state signals. high virtual losses, especially during high PWM frequencies. The method
also causes fictitious current oscillations during switching due to the

Fig. 9. PMSM terminal voltages (A1 and A2) and converter gating signals

6
A. B. Dehkordi and T. L. Maguire Electric Power Systems Research 197 (2021) 107312

presence of inductive and capacitive components Table 2

Recently, resistive switching power electronic models have been Parameters of the synchronous MACHINE
used for real-time simulation [24]. One of the requirements for such Per-unit base Value
detailed simulation of power electronic converters is the capability to
Line-neutral rated voltage 7.97 kV
decompose the admittance matrix of the network in every time-step. Rated MVA 100 MVA
Fortunately, such capacities have recently become feasible with the Rated Frequency 60 Hz
introduction of higher performing processor cores [18]. D- and Q-axes Parameters Value (pu)
Another important performance enhancement of the new converter Stator Leakage Reactance 0.13
D-axis Unsaturated Magnetizing Reactance 1.66
models is their capability for reliable prediction of switching states in Field Winding Leakage Reactance 0.0618
each time-step. This algorithm allows the simulation results to be free of D-axis Damper Leakage Reactance 0.00546
erroneous spikes, even in the absence of iteration [24]. Q-axis Magnetizing Reactance 1.58
Q-axis Damper Leakage Reactance 0.3293
Stator Resistance 0.002
Field Resistance 0.001407
5.2. Performance of the multi-star generator in steady-state
D-axis Damper Resistance 0.00407
Q-axis Damper Resistance 0.01415
As mentioned earlier, the generator supplies the DC bus voltages Zero Sequence Parameters Value (pu)
through diode rectifiers. In steady state, the positive and negative rails of Homo-polar Zero Sequence Reactance 0.13
the DC bus are charged to +/-0.9 kV, maintaining the total voltage of 1.8 Homo-polar Zero Sequence Resistance 0.002
Third Harmonic Zero Sequence Reactance 0.0325
kV. The battery system and the hotel load is also connected to the DC Third Harmonic Zero Sequence Resistance 0.002
bus. The generator is continuously charging the battery and simulta­ Fifth Harmonic Zero Sequence Reactance 0.0195
neously provides power for the propulsion system. With the speed of Fifth Harmonic Zero Sequence Resistance 0.002
motor at 1.0 pu and the battery charged to 75%, the generator produces
around 2.2 MW of power. The steady state voltages, currents and electric
torque of the generator are shown in Fig. 8. As expected, generator
currents represent the dual star arrangement of the windings and elec­ References
tric torque contains the 12th harmonic.
[1] H.A Toliyat, Analysis and simulation of five-phase synchronous reluctance
machines including third harmonic of airgap MMF, IEEE Trans. Ind. Appl. 34 (2)
(1998) 332–339. April.
5.3. Performance of the motor drive system during the loss of a converter [2] L. Parsa, On advantages of multi-phase machines. Industrial Electronics Society,
leg 2005, IECON, Raleigh, 2005. Nov 2005.
[3] E. Levi, Multiphase electric machines for variable-speed applications, IEEE Trans.
Ind. Electron. 55 (5) (2008) 1893–1909. May.
As mentioned earlier, the PMSM is supplied through two 3-phase [4] E.A. Klingshirn, High phase order induction motors - Part I-description and
converters based on PWM switching. One of the advantages of multi­ theoretical considerations, IEEE Trans. Power Apparatus Syst. PAS-102 (1) (1983)
47–53. Jan.
phase machines is their capacity to operate even with the loss of one or
[5] R.H. Park, Two reaction theory of synchronous machines, Part 1, AIEE Trans. 48
more phases. Here in steady state, the speed of machine has reached 1.0 (1929) 716–730.
pu and it absorbs 1.5 MWs. The voltages from phase A of converters 1 [6] R.H. Park, Two reaction theory of synchronous machines, Part 2, AIEE Trans. 52
and 2 are depicted in Fig. 9(a). (1933) 352.
[7] A. Tessarolo, Modeling and Analysis of Multiphase Electric Machines for High-
This Figure also shows the firing pulses to each of the converter Power Applications," Ph.D. thesis, University of Trieste, 2011.
switches. The gating signals to phase A of converter 1 are suddenly [8] Tessarolo, On the modeling of poly-phase electric machines through vector-space
blocked, and the variation of voltages are shown in Fig. 9(a). Note that, decomposition: theoretical considerations. IEEE Int. Conf. on Power Engineering,
Energy and Electrical Drives POWERENG, Lisbon, Portugal, 2009, pp. 519–523.
even with the loss of a converter leg, the motor maintains its speed. [9] A. Rockhill, T.A. Lipo, A generalized transformation methodology for polyphase
electric machines and networks. IEEE Int. Electric Machines & Drives Conf. IEMDC,
Coeur d’Alene, ID, USA, 2015, pp. 27–34.
6. Conclusions and contributions [10] M.Jones Zoric, E. Levi, Vector space decomposition algorithm for asymmetrical
multiphase machines. International Symposium on Power Electronics (Ee), Novi
Analysis of multi-star synchronous machines is presented using a Sad, Serbia, 2017.
[11] H.W. Dommel, Digital computer solution of electromagnetic transients in single
generalized method of vector space decomposition (VSD).
and multiphase networks, IEEE Trans. Power Apparatus Syst. PAS-88 (4) (1969)
A detailed and flexible transient multi-star synchronous machine 388–399. Apr.
model was developed and validated for real-time digital simulation. The [12] R.F. Schiferl, C.M. Ong, Six phase synchronous machine with AC and DC stator
method is applicable to off-line electromagnetic transient programs as connections, part I: equivalent circuit representation and steady-state analysis,
IEEE Trans. Power Apparatus Syst. PAS-102 (8) (1983) 2685–2693. Aug.
well. A typical power system circuit of a marine vessel is simulated using [13] M.R. Aghaebrahimi, R.W. Menzies, A transient model for the dual wound
the introduced model. synchronous machine. International Power System Transient (CCECE 1997), St
John’s, 1997. May,.
[14] Charles L. Fortescue, Method of symmetrical Co-ordinates applied to the solution of
CRediT authorship contribution statement polyphase networks, AIEE Trans. 37 (part II) (1918) 1027–1140.
[15] Y.H. Ku, Transient analysis of AC machinery, AIEE Trans. 48 (1929) 707.
[16] S.A. Nasar, Electromechanical energy conversion in nm-winding double cylindrical
A. B. Dehkordi: . T. L. Maguire: Writing – review & editing. structures in presence of space harmonics, IEEE Trans. Power App. Syst. PAS-87 (4)
(1968) 1099–1106. April.
[17] P.L. Alger, Induction Machines Their Behavior and Uses, Science Publishes, New
Declaration of Competing Interest York, Gordon and Breach, 1970.
[18] Quick Reference Guide for Power8® Processor-based Servers, IBM® Power
SystemsTM, 2017.
The authors declare that they have no known competing financial [19] T.L. Maguire, An efficient saturation algorithm for real time synchronous machine
interests or personal relationships that could have appeared to influence models using flux linkages as state variables,” Electrimacs 2002, montreal, Canada,
June 2002, IEEE Trans. Power Apparatus Syst. PAS-103 (9) (1984) 2446–2451.
the work reported in this paper.
Sept.
[20] A.B. Dehkordi, P. Neti, A.M Gole, T.L. Maguire, Development and validation of a
Appendix I comprehensive synchronous machine model for a real-time environment, IEEE
Trans. Energy Conver. 25 (1) (2010) 34–48. Mar.

TABLE 2

7
A. B. Dehkordi and T. L. Maguire Electric Power Systems Research 197 (2021) 107312

[21] M.R. Patel, Shipboard Propulsion, Power Electronics, and Ocean Energy, CRC [24] T. Maguire, S. Elimban, E. Tara and Yi Zhang “Predicting switch ON/OFF statuses
Press, New York, 2012. in real time electromagnetic transients simulations with voltage source
[22] N.P.W. Strachan, D. Jovcic, Stability of a variable-speed permanent magnet wind converters”, The 2nd IEEE Conference on Energy Internet and Energy System
generator with weak AC grids, IEEE TPD 25 (4) (2010) 2779–2788. Oct. Integration (EI2), Beijing, China, Oct. 20-22, 2018.
[23] T. Maguire and J. Giesbrecht, “Small timestep (<2 μs) VSC model for the real time
digital simulator”, International Conference on Power System Transients (IPST ́ 05)
in Montreal, Canada, June 19-23, 2005, paper No. IPST05-168.