TECHNICAL

NOTES

Design optimization of a round-rotor synchronous

generator for enhancing short-circuit ratio (SCR)
Elissa Soares De Carvalho
WEG
Brazil
elissac@weg.net

P. Kuo-Peng
GRUCAD/EEL/UFSC

F. Wurtz
G2ELab Laboratory, INPG/UJF/CNRS,ENSE3
LIA MAXWELL

Brazil
patrick@grucad.ufsc.br

France / Brazil
frederic.wurtz@g2elab.grenoble-inp.fr

INTRODUCTION

In the last decade, the Brazilian energy market has

witnessed a rise of independent energy producers
entering into the national electrical grid, coming
mainly from sugarcane segment [1]. In these
plants the bagasse, a sugarcane by-product, is
used as fuel for steam turbines in a cogeneration
system, providing economical and environmental
benefits. The Figure 1 shows one of these
turbogenerators installed in a sugarcane plant.

stronger [5]. As a direct consequence, the

generator cost will increase for the same installed
capacity, once larger SCR often means larger airgaps, rotor volume [5] and consequently generator
size.
The goal of this work is to take an existing
standard generator (Table 1) and, keeping its main
external dimensions (diameter and iron stack
length), to optimize its parameters to obtain a
larger SCR. This way, both technical and
commercial objectives could be attended and
generator manufacturer will meet its objective.
2

Figure 1: A Synchronous generator manufactured by

WEG in a sugarcane plant

In past years this system was exclusively used to

provide energy for self-consumption. Seasonality
and lack of regulation discouraged, by the
beginning, the small producers to sell exceeding
energy due the below market prices [1]. With a
new electricity sector regulation approved in 2004,
turning energy sales into a profitable opportunity
for small energy producers, some technical
requirements for turbogenerators from this
segment became more recurrent.
The requirement of larger short circuit ratio (SCR)
for generators from sugarcane generation plants is
correlated to static stability limits. However, in
general, it has not been considered the gain or the
impact of this improvement. Nowadays, modern
excitation systems for instance, using digital
regulators would allow the turbogenerators to
operate safely with SCR values between 0.4 and
0.5. Besides that, the enhancement of SCR tends
to lose its advantage depending on the grid
connection, where the plant is settled, becomes

PROBLEM DEFINITION

This work consists in optimizing the geometry

(Figure 2) of an existing standardized
turbogenerator to achieve the maximum short
circuit ratio (SCR) with no alteration on external
dimensions, taking into account the
electromagnetic and thermal constraints which
define a properly generator operation.

Figure 2: Detail of Rotor and Stator Sheet

Design optimization of a round-rotor synchronous generator for enhancing short-circuit ratio (SCR) - January/2013

TECHNICAL
NOTES
The purpose of preserving the external dimensions
of the machine is to reduce the impact on the
existing manufacturing process, enabling to the
company maintaining its production cost with an
improved generator. So that, the parameters are
handled within their boundaries in order to achieve
a satisfactory result taking into account such limits.
Whereas the optimization of an electric machine
deals with discrete parameters or due to
constraints of the process, as the variation of the
diameters and dimensions of wires and slots or
due to the intrinsic nature of the parameter as the
number of slots, for example, it is advisable to use
genetic algorithms.
Mathematically, a set of parameters [3],

r
r
y = ( f ( x ))

(1)

will be evaluated by a simplified analytical

calculation, starting by an existent generator. For
the optimization, it is proposed to apply the method
MOGAII, a variation of Multi-Objective Genetic
Algorithm MOGA with a directional crossover
operator and multisearch elitism [4], on single
objective-problem [6]. The use of this method will
be explained in section 3.
2.1

The aspect of sheets from the stator and rotor is

shown in Figure 2 where the stator's internal
diameter and rotors outer diameter, the number of
slots and the height of the yoke of the stator are
among the parameters which are handled on
optimizing the generator.
Table 1: Generator rate
Characteristic

Symbol

Value

Unit

Power
Voltage
Frequency
Power Factor
Number of Poles
Phase
Rotor Type
Armature Current

S
U
f
cos
2p
m

17.5
13.8
60
0.8
4
3
Laminated Cylindrical
732.1

MVA
kV
Hz

Besides these parameters, the variation of the

dimensions of the conductors and number of turns
plays a part in determining the dimensions of the
coils. The Figure 4 shows the aspect of the strands
as well of the turns.

GENERATORS CHARACTERISTICS AND

TOPOLOGY

The synchronous generator, driven by steam

turbines and described in this article, has four
poles, with the rotor and the stator cylinder made
of thin sheet metal sub-grouped into packages
separated by cooling channels. The main features
of the generator are described in the Table 1 and
its general aspect shown in Figure 3.
Figure 4: Detail of Coils, Strands and Turns

Figure 3: Round Rotor Synchronous Generator Aspect

Design optimization of a round-rotor synchronous generator for enhancing short-circuit ratio (SCR) - January/2013

TECHNICAL
NOTES
The summary of input parameters is detailed in
table below:

met, the designer obtains a larger set of solutions

allowing him greater control over the final project.

Table 2: Evaluated parameters - Input data

2.3

Parameters
Internal
Stator
Diameter
External
Rotor
Diameter
Parallel
Strands of
Stator in
Height
Number of
Stator Slots
Number of
Turns in
Stator
Winding
Stator
Strand
Width
Stator
Strand
Height

2.2

Symbol

Maximum
Value

Minimum
Value

Step

Unit

Di1

1015

1000

mm

De2

970

958

mm

Ncph1

mm

N1

108

72

12

Z1

bn11

0.20

mm

hn1

3.55

0.15

mm

The short circuit ratio is defined [2] as the ratio

between the necessary field current to obtain the
rated voltage on no-load ( ie0 ) and the field current
necessary to achieve the rated current on shortcircuited ( iesc ):

ie0
iesc

The generator's calculation is yielded through

classical equations [2,8], according steady state
model shown in Figure 5.
Some approaches, such ignoring the low
resistance and considering a fixed value for
saturation factor, are assumed. To compensate
that, the final results will be compared with an
industrial calculation, validated through tests with
round rotor synchronous machines manufactured
over the last 10 years.

Figure 5: Steady-state model of a round rotor synchronous

generator

OBJECTIVE FUNCTION

SCR =

MODEL

(2)

The direct axis reactance is calculated according

to equation [8]:

xdspu = xa + x

(5)

Where:
xa = Magnetization Reactance

x = Leakage Reactance
This parameter can also be defined as the inverse
of the saturated synchronous reactance at the
direct axis ( xds ) in pu:

SCR =

1
xdspu

(3)

Here the SCR will be the objective function, which

will be maximized:

ObjectiveFunction = Max(SCR)

(4)

In this paper, it is selected a single objective

function with a sensible choice of restrictions. For
specific problems with a reduced number of
parameters, as in this case, with the maximum and
minimum limits of the chosen parameters being

2.4

PARAMETERS AND CONSTRAINTS

The evaluated parameters dealt in this paper are

summarized in table 2. The generator geometry,
winding and strands, are modified in order to reach
an optimal value for SCR.
The choice of range of the input parameters is
critical to the success of the optimization process
as well as for its speed, this way, it was
established a procedure that reduces the
probability of divergence of the model and provides
fast and effective information of the previous
values of input data and their impacts on
optimization.
In this paper, it is used the concept of imaginary
machines, described in detail in [9], in establishing
these input values. In summary, this is performed
allowing all parameters to vary continuously and

Design optimization of a round-rotor synchronous generator for enhancing short-circuit ratio (SCR) - January/2013

TECHNICAL
NOTES
using optimization algorithms by gradient methods.
This way both the quality of the model and the
initial parameters are evaluated beforehand,
without much computational effort.
Table 2 shows the values which will be used as
input data on generator's optimization. The
parameters are detailed on Figure 2 and Figure 4.
In addition to the input variables shown in Table 2,
other parameters, some fixed and some
dependent on the input data are used, the main
are listed in Table 3.
According it is stated on the proposition of this
paper, both the outer stators diameter and the
stack of sheets are fixed. The same procedure is
applied to the parallel strands of stator, which are
kept fixed in width.
With respect to the rotor, except for the outer
diameter, all other features are preserved.
Table 3: Evaluated parameters - Main data
Parameters

Symbol

Type

Values

Unit

External
Stator
Diameter

De1

Fixed

1015

mm

Stack Length

Ltp1

Fixed

970

mm

Ncpl1

Fixed

mm

br1
hr1

Variable
Variable

mm
mm

GENETIC ALGORITHM - MOGA-II USED

FOR SOLVING A SINGLE-OBJECTIVE
PROBLEM

It is desirable that an optimization tool is flexible to

properly settle any type of problem, this way is
desirable that the algorithm is able to attend
problems both single and multi-objective
problems, depending on the proposition and
models.
In this paper, a multi-objective genetic algorithm is
used to optimize a single objective problem,
starting from the statement of Poles in [6], that
questions the robustness and efficiency of the
MOGA-II algorithm in a single objective problem
from real world.
MOGA-II is a version of MOGA Poloni [4] with
elitism. As with any genetic algorithm, this one
aims to meet conflicting objectives, guiding the
resolution of the problem towards a set of
individuals arranged in a Pareto frontier.

The constraints and limits are defined according

Table 4. They go towards attending of thermal and
electromagnetic boundaries defined by theoretical
induction limits and iron losses. Such boundaries
are stated based on industrial tests from several
existent generators.

The efficiency of the MOGA-II is ruled by its

operators and the use of elitism [6]. The operators
for reproduction are:

One Point Crossover;

Directional Crossover;

Mutation;

Selection.
At each stage of the reproduction process of the
operators is selected from the probability of
occurrence placed as shown in table below:
The elitism used in a multi-objective GA preserves
the individuals close to the Pareto ranking. In this
evaluation, when it is used in single objective
problems, it restricts to copy to the next generation
the best fitness solution.

Table 4: Constraints and limits

Table 5: Moga-II - Operators probabilities

Parallel Strands of
Stator in Width
Stator Slot Width
Stator Slot Height

Variables
Tooth
Induction
Yoke
Induction
Stator
Joule
Losses
Rotor
Joule
Losses
Stator
Tooth
Width

Maximum
Limit

Minimum
Limit

Tolerances

Bd1

2.1

1.4

0.0

Bc1

1.5

1.2

0.0

Symbol

Parameters

Unit

0.5

Probability of Selection

0.05

Probability of Mutation
Elitism

Pj1

75

kW

Pj2

92

kW

bd1

9.0

mm

Probabilities

Probability of Direction
Crossover

0.1
Enabled

Design optimization of a round-rotor synchronous generator for enhancing short-circuit ratio (SCR) - January/2013

TECHNICAL
NOTES
5

RESULTS

The table 6 presents the results of the optimized

generator, comparing it to the original one. It can
be noticed that some of the parameters from the
existing turbogenerator, which were not being used
on their whole capacity, were extended to their
limits in order to enhance its short-circuit ratio
(SCR).
Once the problem involves electrical machines
which utilize discrete parameters, it was used a
genetic algorithm with the purpose to achieving the
best solution, preventing to finding local solutions
instead of global ones.
Genetic algorithms on the other hand, can take a
long time due to amount of interactions and initial
population, mainly if it is considered some wrong
input data or if some constraints are set
mistakenly. In this way, in order to reduce
unnecessary processing time, it was employed a
procedure of setting parameters utilizing an
imaginary machine concept, which is broadly
explored in [9] for this specific case.

Table 6: Comparative results

Symbol

De1
Di1
De2
Ltv
N1
bn11
hn1
Z1

br1
hr1

Original
Generator

Optimized
Generator
Generator Topology

Description
Stator
External
Diameter
Stator
Internal
Diameter
Rotor
External
Diameter
Package
Lenght
Number of
Slots
Stator
Strand
Width
Stator
Strand
Height
Number of
Turns per
Slot
Airgap
Stator Slot
Width
Stator Slot
Height

Unit

Diff.

1610

1610

mm

1000

1012

mm

1.2%

958

960

mm

0.2%

1130

1130

mm

72

84

16.7%

7.1

5.6

mm

3.35

2.8

mm

16.7%

21

26

mm

24%

21.7

18.7

mm

115

121

mm

302

314

0.40%

2.07

2.07

1.49

1.48

75

71

kW

-5.3%

95

92

kW

-3.2%

0.59

25.5%

Performance Data
If
Bd
Bc
Pj1
Pj2

Field
Current
Tooth
Induction
Yoke
Induction
Stator Joule
Losses
Rotor Joule
Losses

Objective
SCR

Short Circuit
Ratio

0.47

The particular problem dealt in this paper is a real

world problem, however with few variables.
Considering that all constraints were well set and
the model was well settled in a previous step in
which it was considered the generator as an
imaginary machine [9], it is not necessary to
consider it as a multi-objective problem.
However, since in a daily task of design and
optimization, each case can be presented from
different perspectives, it is interesting to make use
of a flexible genetic algorithm which attends to
most part of situations. In order to comply with this
purpose, this paper uses a multi-objective genetic
algorithm: MOGA-II in a single-objective problem.
This application was raised and developed in [6], in
which there was a proposal to apply it in a real
world problem.
It is possible to observe on table 6, that the
improvement of 25.5% on the SCR is achieved by
Design optimization of a round-rotor synchronous generator for enhancing short-circuit ratio (SCR) - January/2013

TECHNICAL
NOTES
handling some parameters such as some
diameters, dimensions of the strands, geometry
and number of slots.
It must be emphasized that these results are
reached using a simple steady-state model of a
round rotor synchronous generator, using some
assumptions, which can hide the true behavior of a
machine. In order to confirm the result, the same
model was run into an industrial calculation used
by WEG, validated by tests on its manufactured
generators.
The Table 7 shows this comparison in which it is
verified the maintaining of SCR.
Table 7: Comparative results
Symbol

If
Bd
Bc
Pj1
Pj2

Description
Field
Current
Tooth
Induction
Yoke
Induction
Stator Joule
Losses
Rotor Joule
Losses

Optimized
Generator
Generator
Industrial
Calculation Software*
Performance Data

Unit

Diff.

314

329.4

4.7%

2.07

2.09

1.48

1.48

71

72.2

kW

92

102

kW

0.61

1.7%
9.8%

Objective
Short Circuit
Ratio
*WEG Courtesy

SCR

0.59

3.3%

REFERERENCES

[1] Srgio Granville, Priscila Lino, Recent

Advances of Sugarcane Biomass Cogeneration in
Brazil, IEEE PES '09 Power & Energy Society
General Meeting, 2009
[2] M. Kostenko and L. Piotrovsky, Electrical
Machines, vol. II; Foreign Languages Publishing
House, pp.233-234.
[3] Jacek F. Gieras, Michell Wing, Design and
Applications Permanent Magnet Motor
Technology, Crc Press, 2002, pp354-355
[4] Silvia Poles MOGAII An Improved MultiObjective Genetic Algorithm,STECO Technical
Report 2003-006 , December 4,2003
[5] C.-E. Estephan, Z. Baba, Specifying a
Turbogenerator's Electrical Parameters Guided by
Standards and Guide Codes ,2001
[6] Silvia Poles, Enrico Rigoni, Tea Robic MOGAII Performance on Noisy Optimization Problems in
Proc. Int. Conf. Bioinspired Optimization
MethodsApplicat. (BIOMA 04), Ljubljana,
Slovenia, Oct. 2004, pp. 51-62.
[7]Tadahiko Murata, Hisao IshiBuchi, MOGA:
Multi-Objective Genetic Algorithms, IEEE
International Conference on Evolutionary
Computation,pp289
[8] Boldea, I. The Electric Generators Handbook Synchronous Generators Timisoara: CRC Taylor&
Francis Group, LLC, 2006
[9] Wurtz, F., Kuo-Peng, P, S. De Carvalho,E. The
Imaginary Pareto Front: a helpful Tool for setting
Optimisation Problem for Design of
Electromagnetic Devices. Accepted for
Publication on COMPUMAG2011.

Design optimization of a round-rotor synchronous generator for enhancing short-circuit ratio (SCR) - January/2013