ISSN: 1942-0730 (Print), 1942-0749 (Online) NO.2-01-02... · 2011. 9. 7. · Yu Fu Wing Li Peisheng...

J. Electromagnetic Analysis & Applications, 2009 Published Online June 2009 in SciRes (www.SciRP.org/journal/jemaa)

Copyright © 2009 SciRes JEMAA

CONTENTS

Volume 1 Number 2 June 2009 Design of Axial-Flux Motor for Traction Application

N. Chaker, I. B. Salah, S. Tounsi & R. Neji……………………………………………………………73

Distance Measure Based Rules for Voltage Regulation with Loss Reduction

Y. R. Hernandez & T. Hiyama……………………………………………………………………………85

Partial Discharge Source Classification and De-Noising in Rotating Machines Using Discrete Wavelet Transform and Directional Coupling Capacitor

M. A. Kashiha, D. Z. Tootaghaj & D. Jamshidi…………………………………………………………92

Analysis of Flashover Characteristics under Nanosecond Pulsed Coaxial Electric Field

W. L. Huang, J. F. Cui & G. S. Sun…………………………………………………………………97

Dust Effect on the Performance of Wind Turbine Airfoils

N. X. Ren & J. P. Ou……………………………………………………………………………………102

A Novel Half-Bridge Power Supply for High Speed Drilling Electrical Discharge Machining

H. Huang, J. C. Bai, Z. S. Lu & Y. F. Guo……………………………………………………………108

Air Compressor Control System for Energy Saving in Locomotive Service Plant

W. Y. Mo………………………………………………………………………………………………114

ICI Performance Analysis for All Phase OFDM Systems

R. H. Ge & S. L. Sun…………………………………………………………………………………118

Design and Implementation of a Novel Digital Frequency Superposition Testing Power Supply for Induction Motor

X. J. Dong, S. X. Zhuang & W. Yan……………………………………………………………124

Journal of Electromagnetic Analysis and Applications (JEMAA)

Journal Information

SUBSCRIPTIONS

The Journal of Electromagnetic Analysis and Applications (Online at Scientific Research Publishing, www.SciRP.org)

is published quarterly by Scientific Research Publishing, Inc. 5005 Paseo Segovia, Irvine, CA 92603-3334, USA.

E-mail: [email protected]

Subscription rates: Volume 1 2009 Print: $50 per copy.

Electronic: free, available on www.SciRP.org.

To subscribe, please contact Journals Subscriptions Department, E-mail: [email protected]

Sample copies: If you are interested in subscribing, you may obtain a free sample copy by contacting Scientific

Research Publishing, Inc at the above address.

SERVICES

Advertisements

Advertisement Sales Department, E-mail: [email protected]

Reprints (minimum quantity 100 copies)

Reprints Co-ordinator, Scientific Research Publishing, Inc. 5005 Paseo Segovia, Irvine, CA 92603-3334, USA.


COPYRIGHT

Copyright©2009 Scientific Research Publishing, Inc.

All Rights Reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in

any form or by any means, electronic, mechanical, photocopying, recording, scanning or otherwise, except as

described below, without the permission in writing of the Publisher.

Copying of articles is not permitted except for personal and internal use, to the extent permitted by national

copyright law, or under the terms of a license issued by the national Reproduction Rights Organization.

Requests for permission for other kinds of copying, such as copying for general distribution, for advertising or

promotional purposes, for creating new collective works or for resale, and other enquiries should be addressed to

the Publisher.

Statements and opinions expressed in the articles and communications are those of the individual contributors and

not the statements and opinion of Scientific Research Publishing, Inc. We assume no responsibility or liability for

any damage or injury to persons or property arising out of the use of any materials, instructions, methods or ideas

contained herein. We expressly disclaim any implied warranties of merchantability or fitness for a particular

purpose. If expert assistance is required, the services of a competent professional person should be sought.

PRODUCTION INFORMATION

For manuscripts that have been accepted for publication, please contact:


J. Electromagnetic Analysis & Applications, 2009, 2: 73-83 doi:10.4236/jemaa.2009.12012 Published Online June 2009 (www.SciRP.org/journal/jemaa)


1

Design of Axial-Flux Motor for Traction Application

Nadia Chaker, Ibrahim Ben Salah, Souhir Tounsi, Rafik Neji

Ecole Nationale d’Ingénieurs de Sfax (ENIS), BP. 1173, 3038 Sfax Tunisie, Laboratoire d’Electronique et des Technologies de l’Information (LETI), Equipe Véhicule Electrique et Electronique de Puissance (VEEP), Tunisia. Email: [email protected], [email protected]

Received February 9th, 2009; revised March 26th, 2009; accepted April 2nd, 2009.

ABSTRACT

This paper deals with the design of high power – low dimensions axial-flux permanent-magnet motor intended for trac-tion application. First, two motor configurations are analytically designed and compared using finite element calcula-tion. Then, the configuration yielding the best performances is integrated and modelled with the whole traction chain under MATLAB/SIMULINK environment in order to demonstrate the motor operation on a large speed band.

Keywords: Axial-Flux Permanent-Magnet Motor, Design Criteria, Finite Elements, Traction Chain, Circulation Mission

1. Introduction

Nowadays, the use of internal combustion engines in vehicles is one of the principal causes of several pollution problems as air and sound ones. Therefore, the electrical vehicles constitute an excellent candidate to avoid these problems. However, since their appearance, the major problems of this type of vehicles remain in high cost, weak autonomy and over speed problems. For that, it becomes essential to give a particular care when choosing the principal element of the electric traction chain which is the electric motor.

For electric traction applications, synchronous or asyn-chronous motors [1] with radial or axial fluxes [2,3], can be used. In order to increase the torque generation capabil-ity, these motors can be modulated. Moreover, the conse-quent progress of the permanent-magnet technology makes permanent magnets synchronous motors more and more utilized for variable speed and high performance systems.

In [4], an effectiveness and mass comparison study between radial and axial structures of a perma-nent-magnet synchronous motor was presented. For a constant power, it was demonstrated that the axial con-figuration with 4 pole pairs in rotor and 6 teeth in stator has the best compromise effectiveness-mass. Thus, this motor appears particularly interesting for electric vehicle applications. In fact, the axial-flux permanent-magnet motor has many advantages [5] as: 1): high effectiveness and high power factor, 2) high specific power, 3) no ring-brushes and 4) possibility of modularity.

As an industrial application we can mention that JEUMONT industry used the technology of axial flux structures to develop high power machines intended to

boats and Aeolian alternators driving [6]. This paper presents the design of a high power – low

dimensions axial-flux permanent-magnet motor for a traction application. First, the design criteria of electrical parameters are highlighted for trapezoidal and sinusoidal motor configurations. Then, considering the vehicle specification, the motor geometric parameters are ana-lytically determined for a comparison based on finite element calculation between both configurations per-formances. Finally, a traction chain integrating best con-figuration is modelled under MATLAB/SIMULINK en-vironment in order to demonstrate the motor operation on a large speed band without weakening flux method.

2. Generalities about Axial-Flux Permanent- Magnet Motor

Several axial-flux machine configurations exist depending on the stator(s) position(s) with respect to the rotor(s) ones, as shown in Figure 1. We can find: A structure with one rotor and one stator, Figure 1 (a). A structure, in which the rotor is located between the

stators, Figure 1(b). A structure, in which the stator is located between the

rotors, Figure 1(c). A multistage structure including several rotors and

stators Figure 1(d). In traction applications, more the motor has higher

torque generation capabilities more it is interesting. As

Design of Axial-Flux Motor for Traction Application 74

(b) (c) (d) (a)

Figure 1. Permanent-magnet axial-flux machines configurations. Legend: (a): Single-rotor - single-stator structure; (b): Sin-gle-rotor - two-stators structure; (c): Two-rotors - single-stator structure, called hereafter also as AFIPM machine (Ax-ial-Flux Interior rotor Permanent-Magnet machine); (d): Multistage structure including two stator blocks and three rotor blocks [2]

radial-flux motors, the axial-flux ones can be modulated which leads to the increase of their torque generation capabilities [1,2,4]. In fact, the four configurations shown in Figure 1 are used for traction applications. The torque generated with the fourth configuration, composed of four modules (Figure 1(d)), is twice times greater than the torque of the third and the second configurations which contain two modules, Figure 1(c) and Figure 1(b), and four times than the first configuration developed torque (one module), Figure 1(a).

It is to be signalled that configurations illustrated in Figure 1(b) and Figure 1(c) have the same torque genera-tion capabilities and the choice between both depends if the application needs an outer or inner rotor.

In the present paper, we have been interested in the in-tegration of the axial-flux technology for automotive traction application as shown in Figure 2.

3. Analytical Design of the Unit Motor – Inverter

In the present section, the single-rotor – single-stator structure which is the simplest axial-flux permanent- magnet motor configuration [7] is considered, Figure 1(a). At the beginning and in order to satisfy the design criteria

of the motor associated to its inverter, the electrical pa-rameters are calculated for two configurations: the three-phase motor with trapezoidal back e.m.f wave form and the three-phase motor with sinusoidal back e.m.f wave form. Then, in order to define the structure of the considered motor configurations, the geometrical pa-rameters are analytically calculated, using the vehicle specifications recapitulated in Table 1 (Appendix).

During the design process it is required that: when the vehicle reaches the maximum specified speed, the motor is controlled with full wave form and develops the needed torque. At this operation point:

The electromagnetic torque Tem, via the mechanical transmission system (reducer and differential), is ex-pressed as the following:

maxem dm bT T V V (1)

where Tdm is the starting torque expressed as:

sin( )wdm v b d v

d

RT M V t M g

r .

The maximum value of the motor back e.m.f is:

2 2max max int 4C sph sph C sph ext eE N d dt N d d E N D D B (2)

Taking in account the vehicle specification mentioned in Table 1, the motor must develop a torque of Tem = 40.625Nm with a maximum value of back e.m.f equal to Ec=138.462V.

It is to be signaled that, in coming EF study the last men-tioned values of Tem and Ec will be used to verify the calculated geometrical parameters of the obtained con-cept.

(e) (a) (b) (c) (d)

Figure 2. Integration of axial-flux motor in automotive traction chain. Legend: (a): a battery providing the input direct volt-age of the inverter; (b): a conventional six-switch three-phase inverter insuring the generation of the three-phase voltage sup-plying the motor armature; (c): a permanent-magnet axial-flux motor with sinusoidal back e.m.f used for the vehicle driving; (d): a reducing system insuring the transmission of the motor mechanical energy to the vehicle wheels; (e): vehicle wheel



A conventional six-switch three-phase inverter is used

to supply the motor armature. In order to recuperate en-ergy during the deceleration phases, the inverter has re-versible structure.

In what follows, we are involved in an analytic calcu-lation of the input direct voltage of the inverter Udc, and the maximum current intensity Iph feeding the motor phases.

3.1 Design Criteria of Electrical Parameters for a Trapezoidal Wave Form Motor

To supply this motor, currents in 120° electric crenels shape are considered. The motor power supply appears as a succession of 60° electric sequences during which two phases are simultaneously crossed by two opposite con-stant currents as shown in Figure 4.

Analysing Figure 4, one can notice that: 1): the motor back e.m.f is trapezoidal and 2): the resulting torque is as a simple juxtaposition of the three phase’s constant torques. However, in order to limit the torque ripple, it is required to guarantee the right duration for each operation sequence and also an excellent form of the back e.m.f [8].

At the considered operation point, the electromagnetic torque of the motor is related to the phase current inten-sity as follows [9]:

Figure 3. Six-switch three-phase inverter

Figure 4. Power supplying of a trapezoidal wave form mo-tor. Legend: T1, T2, T3, T4, T5 and T6: inverter switches; i1, i2, i3: phases currents; motor back e.m.fs; Tem: electro-magnetic torque

Current Current level to be reached

Current

Figure 5. Shape of the phase current of the motor

At the considered operation point, the electromagnetic torque of the motor is related to the phase current inten-sity as follows [9]:

max _2em C ph e tra phT E I k I (3)

with 2 2_ max int2 2e tra C sph ext ek E N D D B is the elec-

tric constant of the motor. Consequently, the phase current intensity is:

_ph ems e traI T K (4)

Figure 5 illustrates the crenel shape of the current feeding motor phases.

The current wave form shows two important parame-ters [10]:

The current maintaining time tp:

max1 3 2pt p (5)

where max is the maximum angular velocity of the mo-tor.

The current rising time tm:

_ maxln 1 2m ph dc et L R RI U K tra (6)

with R and L are the phase resistance and the phase in-ductance, respectively.

The torque ripple factor r, is defined as [10]:

m pr t t (7)

For a fixed value of the ripple factor, the input direct voltage Udc can be obtained as follows:

max _ max2 1 exp 2 3dc ph e traU RI r p L R k

(8)

3.2 Design Criteria of Electrical Parameters for a Sinusoidal Wave Form Motor

The output voltages of the inverter applied to the armature of the sinusoidal back e.m.f configuration of the perma-nent-magnet axial-flux motor are illustrated in Figure 6.

The fundamental of the back e.m.f of the first phase is

Tem

0 60 120 180 240 300 360

i1

i2

i3

Iph

B. e

.m.f

s (V

)

Electric angle

T1 T3 T5 T2 T4 T6 T6

tm

tp

Times



Figure 6. Phases voltages shapes

noted Uph11. It is expressed as the following [11]:

11( ) 2 sin 2ph dcU t U T t (9)

To guarantee operation regime under maximum torque, the motor piloting angle between the back e.m.f EC and the phase current Iph is fixed equal to zero in the control system. At the maximum speed, the maximum value of the fundamental can be simply obtained from the Fresnel diagram as:

2

11 maxph ph C phU RI E L I 2

(10)

where max is the maximum pulsation of motor voltage. The input direct voltage Udc can so be calculated using

the following expression:

2

max2dc ph C phU RI E L 2

I (11)

At the considered operation point, the electromagnetic torque developed by the considered configuration of the motor is related to the phase current intensity as follows [9]:

max _ sin3 2em C ph e phT E I k I (12)

with 2 2_ sin int3 2 3 8e sph extk E N D D B e is the elec-

tric constant of the motor. Consequently, the phase current intensity is:

_ sinph ems eI T K (13)

Furthermore and for both configurations of the motor: the permanent-magnet axial-flux motor with trapezoidal back e.m.f wave form and the permanent-magnet axial-flux motor with sinusoidal back e.m.f wave form, the armature parameters R and L are expressed as the following [12]:

Phase resistance:

6ph b tc spR T N L I

with bT is copper resistivity at the temperature Tb,

Ntc is the total number of conductors, is the current density and Lsp is the mean length of one turn.

Uph

2Udc/3 Udc/3 t

Phase inductance: T/6

ph (14)

ph ep fL L L p (15)

with is the air-gap inductance and is the phase leakage inductance through one slot.

epL fpL

3.3 Geometrical Parameters of the Motor

The considered permanent-magnet axial-flux motor is composed of only one module containing two parts: one stator and one rotor, as illustrated in Figure 1(a).

The stator yoke is laminated and made up of iron-silicium, Figure 1(a). It contains 12 identical slots where the three phase winding is inserted. Totally, 6 coils are used and each phase is obtained by putting in series two appropriate coils, as shown in Figure 8. The stator teeth are of two kinds: 1) 6 large teeth called principal teeth, around which coils are winded, and 2) 6 small teeth located in between adjacent principal teeth, called in-serted teeth. The width of the inserted tooth is variable depending on wished back e.m.f form.

The rotor is an iron massive disc where eight samar-ium-cobalt permanent magnets, with a remanent polariza-tion of 1.175T, are mounted on its surface and four pole pairs are so obtained as shown in Figure 7(a) and (b).

Figure 9 shows the geometrical parameters necessary to define the structures of both configurations of the sur-face-mounted axial-flux permanent-magnet motor. Be-fore calculating these geometrical parameters, specific coefficients have to be defined:

pL p : pitch in between poles.

a PL L : rotor occupancy rate by permanent mag-

nets. This coefficient is equal to 1 when the mounted permanent magnets cover the whole surface of the rotor, and is equal to 0.5 when permanent magnet surface is equal to the air one in rotor.

ldla ptm aR A L : report between angular width of a

principal tooth Aptm by the angular width of a magnet La, Figure 7.

ndnp teR N p : report of the teeth number (Nte) by the

number of pole pairs.

did itm ptmR A A : report between the angular width of

an inserted tooth Aitm by the angular width of a principal tooth Aptm, Figure 7.

In coming finite elements study, the last defined coeffi-cients are taken same as ones used in [9] and [10]. These coefficients are resumed in Table 2 (Appendix). In the mentioned works, [9,10], it has been demonstrated that the considered values guarantee the best wave forms of gene-

0 T/3 T/2

t

t



(a) Trapezoidal back e.m.f configuration

(b) Sinusoidal back e.m.f configuration

Figure 7. Structures of both configurations of the perma-nent-magnet axial-flux motor

Figure 8. Cylindrical cut plan of the stator of the axial-flux motor

Figure 9. Geometrical parameters of the permanent-magnet axial-flux motor. Legend: Hr: rotor disc thickness; Hpm: permanent magnet height; Hth: slot height; Hy: stator yoke thickness; e: air-gap thickness; Ls: slot width

rated fluxes and discard leakage fluxes between perma-nent magnets, for trapezoidal and sinusoidal motors.

Referring to [12], the motor geometrical parameters are calculated by integrating the mentioned coefficient in trigonometric formulas. The obtained geometrical pa-rameters for both motor configurations are recapitulated in Table 2 (Appendix).

4. Finite Elements Study

In axial-flux motor, the magnetic phenomena are sym-metrical according to the motor radial direction. Thus, the finite elements study of the two motor configurations can be simplified from 3D to 2D finite elements study which is simpler and more speed from the point of view of cal-culation time. The used software is MAXWELL 2D [13].

Figure 10 illustrates the finite elements study domain of the trapezoidal configuration of the axial-flux motor.

In order to validate the analytical calculated parameters, this study is intended to the computation of the generated fluxes in the motor air-gap which yields the back e.m.fs and the developed torque at load operation point. In the first step, the geometrical parameters analytically calcu-lated in Subsection 2.3 are used to define the geometry used in the finite elements program. Then, in a second step and for different values of the rotor position, two different finite elements calculations are processed: An investigation of the effect of only the permanent

magnets: the machine is working as a generator at no-load operation regime. The fluxes wave forms due to the permanent magnets effect are so carried out. Conse-quently, motor back e.m.fs wave forms and amplitudes are deduced which characterise the electric/mechanic power transfer.

1’

A computation under load operation point: the cur-rent feeding the motor armature is in phase with the back e.m.f obtained through the previous study. The so gener-ated flux is used to find out the machine torque.

Figure 11 shows the flux lines through the magnetic circuit of the sinusoidal wave form configuration of the axial-flux motor due to the effect of only the permanent magnets and Figure 12 shows the flux lines through the magnetic circuit of the sinusoidal wave form configura-tion of the axial-flux motor under load operation.

Analyzing these figures, one can notice that leakage fluxes between permanent magnets do not exist and all generated fluxes are useful.

4.1 Finite Elements Study at Generation Mode

For both configurations of the axial-flux motor, the con-sidered operation point corresponds to the maximum speed of the vehicle (80km/h) which means an angular velocity equal to 341.88rad/s. For trapezoidal configura-

2 2 3’ 3’ 1 1 2’ 2’ 3 3 1’

North magnet

Stator yoke

Rotor disc

South magnet

Principal tooth

Hr

Inserted tooth

Slot

Hy

Hth

e

Hpm

Ls


Design of Axial-Flux Motor for Traction Application


78

Figure 10. Cylindrical cut plan of the trapezoidal wave form motor

Figure 11. Flux lines through the magnetic circuit of the sinusoidal wave form configuration of the axial-flux motor due to the effect of only the permanent magnets

Figure 12. Flux lines through the magnetic circuit of the sinusoidal wave form configuration of the axial-flux motor under load operation

-0,010

-0,008

-0,006

-0,004

-0,002

0,000

0,002

0,004

0,006

0,008

0,010

0 10 20 30 40 50 60 70 80 9

Angle (°)

Flu

x (

Wb

)

0

Flux1_no-load Flux2_no-load Flux3_no-load

-0,008

-0,006

-0,004

-0,002

0,000

0,002

0,004

0,006

0,008

0 10 20 30 40 50 60 70 80 9

Angle (°)

Flu

x (W

b)

0

Flux1_no-load Flux2_no-load Flux3_no-load

(a) Trapeze case (b) Sine case

Figure 13. Fluxes generated in the air-gap of the machine for a generation regime at no-load operation and for a vehicle speed of 80km/h

tion and sinusoidal one, the wave form of generated fluxes in the air-gap are illustrated respectively in Figure

For each motor phase, the back e.m.f can be obtained considering a perfect magnetic circuit and using the fol-lowing expression: 13(a) and (b) where rotor position is varied from 0° to 90°.

C sph ph sph ph sph phe t N d dt N d dt d d N d d (16)

with is the rotor position and ph is the flux of the

considered phase. The differential phd d can be obtained by a lin-

earization between two consecutive positions and the back e.m.f is so expressed as:

1 1 2( ) ( )C sph ph phe N 1 2 (17)

Figure 14 shows the wave form of the obtained back e.m.fs for both axial-flux motor configurations.

Analyzing those figures one can remark that in the case of the trapezoidal configuration of the motor the gener-

ated back e.m.fs are perfectly trapezoidal and in the case of the sinusoidal motor configuration the generated back e.m.fs are also perfectly sinusoidal.

4.2 Finite Elements Study under Load Operation

In this section, the motor is considered under load opera-tion regime. The motor armature is supplied by three currents in phase with the back e.m.fs obtained at the no-load operation of the generation regime. As illustrated in Figure 15(a) and (b), crenel shape currents with a maximum intensity Iph_trapeze = 50.154A are used for the trape-zoidal configuration, and sinusoidal shape currents with a


-200

-150

-100

-50

0

50

100

150

200

0 10 20 30 40 50 60 70 80 9

Angle (°)

Ba

ck e

.m.f

(V

)

0

Back e.m.f 1_no-load Back e.m.f 2_no-load Back e.m.f 3_no-load

-200

-150

-100

-50

0

50

100

150

200

0 10 20 30 40 50 60 70 80 9

Angle (°)

Ba

ck e

.m.f

(V

)

0

Back e.m.f 1_no-load Back e.m.f 2_no-load Back e.m.f 3_no-load

(a) Trapeze case (b) Sine case

Figure 14. Wave form of the generated back e.m.f for a generation regime at no-load operation and for a vehicle speed of 80km/h maximum intensity Iph_sine = 66.872A are used for the sinusoidal configuration.

Figure 15(c) and Figure 15(d) illustrate the wave form of obtained fluxes in the motor air-gap for the maximum vehicle speed (80 km/h), and different rotor positions varying from 0° to 90°, respectively for trapezoidal and sinusoidal motors. Analysing these figures, one can notice the appearance of flux distortion at the load operation re-gime in respect with the no-load operation one. This dis-tortion is essentially due to the magnetic armature reaction.

Considering Equation (16) and for a perfect magnet circuit, the back e.m.f of the motor can be calculated and illustrated as shown in Figure 15(e) and (f) for trapezoi-dal and sinusoidal configurations, respectively. Referring to the aforementioned figures, one can remark the fol-lowing: 1): for both configurations the analytical maxi-mum value of the back e.m.f Ec = 138.462V is reached, 2): the sinusoidal configuration generates a sinusoidal back e.m.f without peaks which yields a torque wave form with no peaks, Figure 15(h). However, the trape-zoidal configuration generates a trapezoidal distorted back e.m.f containing several peaks giving a torque wave form with several peaks too, Figure 15(g).

Referring to Figure 15(g) and (h), both motors are able to develop the requested torque. The ripple figuring in the torques wave forms is due essentially to the motors cog-ged structures which cause the appearance of a cogging torque. Considering the obtained torques wave forms, vibration problems due to the torque ripple are sharper in the case of trapezoidal motor than in the case of the si-nusoidal one. For that, in coming study, only the features of the permanent-magnet axial-flux sinusoidal motor are investigated.

5. Traction Chain Modelling

The present section is devoted to the modelling of an axial-flux motor with sinusoidal back e.m.f wave form associated to a six-switch three-phase inverter for traction

application. The whole traction chain is modelled in order to investigate the motor behaviour vis-à-vis of a desired speed sequence and of the circulation mission of the Na-tional Institute of Research on Transports and their Secu-rity (INRETS).

Figure 16 shows the block diagram of the adopted control vector strategy of the motor implemented under MATLAB/SIMULINK environment.

The synchronous permanent magnet machine can be described in the d-q referential as follows [14]:

d d d d e q q

q q q q e d d e

V Ri L di dt L i

V Ri L di dt L i K

m

(18)

with e is the electric pulsation, Ld and Lq are respec-tively direct in squaring inductances.

The direct and in squaring components of the current can be deduced using:

d d e q q d

q q e d d e m q

I V L i L S R

I V L i K L S

R (19)

with S is Laplace operator. The developed electromagnetic torque is so expressed

[15]:

3 2 1 2em e q d q d qT K I p L L I I (20)

Considering the fundamental dynamic law describing the vehicle motion, the torque needed on wheels is:

w br aero g v w r v wT T T T M R dV dt T M R dV dt

(21)

with Tbr is the torque due to bearing resistance force, Taero

is the torque due to aerodynamic load, Tg is the torque due to gravity forces, a coefficient related to the inertia



Trapezoidal Sinusoidal

-60

-40

-20

0

20

40

60

0 10 20 30 40 50 60 70 80 9

Angle (°)

Cu

rre

nt

(A)

0

I1 I2 I3

(a) Three phase current

-80

-60

-40

-20

0

20

40

60

80

0 10 20 30 40 50 60 70 80 9

Angle (°)

Cu

rren

t (A

)

0

I1 I2 I3

(b) Three phase current

-0,010

-0,008

-0,006

-0,004

-0,002

0,000

0,002

0,004

0,006

0,008

0,010

0 10 20 30 40 50 60 70 80 9

Angle (°)

Flu

x (

Wb

)

0

Flux1_load Flux2_load Flux3_load

(c) Air-gap fluxes

-0,008

-0,006

-0,004

-0,002

0,000

0,002

0,004

0,006

0,008

0 10 20 30 40 50 60 70 80 9

Angle (°)

Flu

x (

Wb

)

0

Flux1_load Flux2_load Flux3_load

(d) Air-gap fluxes

-250

-200

-150

-100

-50

0

50

100

150

200

250

0 10 20 30 40 50 60 70 80 9

Angle (°)

Bac

k e.

m.f

(V

)

0

Back e.m.f 1_load Back e.m.f 2_load Back e.m.f 3_load

(e) Back e.m.fs

-200

-150

-100

-50

0

50

100

150

200

0 10 20 30 40 50 60 70 80 9

Angle (°)

Bac

k e

.m.f

(V

)

0

Back e.m.f 1_load Back e.m.f 2_load Back e.m.f 3_load

(f) Back e.m.fs

0

10

20

30

40

50

60

70

0 10 20 30 40 50 60 70 80 9

Angle (°)

To

rqu

e (

Nm

)

0

T_analytic T_load

(g) Electromagnetic torque

0

10

20

30

40

50

0 20 40 60 80

Angle (°)

To

rqu

e (

Nm

)

100

T_analytic T_load

(h) Electromagnetic torque

Figure 15. Load operation of trapezoidal and sinusoidal motor configurations



3

C3

2

C2

1

C1

triangular signal

Sine Wave2

Sine Wave1

Sine Wave

Relay2

Relay1

Relay

C1

C2

C3

Figure 16. Block diagram of the vector control strategy of the axial-flux motor Figure 16. Block diagram of the vector control strategy of the axial-flux motor

of turning parts (wheels, driving shaft and gearing system) and Tr is the resistive torque. of turning parts (wheels, driving shaft and gearing system) and Tr is the resistive torque.

The control strategy block diagram, shown in Figure 16, presents two regulation loops: the first loop is used for speed regulation and the second one for current regu-lation. The control vector strategy operates with only the in squaring component of the current. For that the direct current is cancelled and its reference value is fixed to zero. Consequently, referring to Equation (20), the de-veloped electromagnetic torque is expressed as:

The control strategy block diagram, shown in Figure 16, presents two regulation loops: the first loop is used for speed regulation and the second one for current regu-lation. The control vector strategy operates with only the in squaring component of the current. For that the direct current is cancelled and its reference value is fixed to zero. Consequently, referring to Equation (20), the de-veloped electromagnetic torque is expressed as:

3 2em e qT K I (22)

The Inverter switches are driven using PWM control signals and the three voltages provided to supply the mo-tor armature are:

1 1 2

2 2 1

3 3 2

3 2

3 2

3 2

ph dc

ph dc

ph dc

U U C C C

U U C C C

U U C C C

3

3

1

(23)

with Uph1, Uph2, Uph3 are the three phase voltage provided by the inverter, and C1, C2, C3 are the command constants for the inverter high transistors T1, T3 and T5, respectively.

The constants C1, C2 and C3 are generated using a PWM generation block, Figure 17, based on the com-parison between a triangular signal which frequency is equal to the desired commutation frequency of the in-verter switches and a sinusoidal one which pulsation gives the desired motor speed. A transistor switched on corresponds to C = 1, and a transistor switched off corre-sponds to C = 0.

At vehicle maximum speed (80km) and for the geo-metrical parameters analytically calculated and validated by the finite elements study, the direct voltage applied to the inverter Udc, the motor resistance R, the direct induc-tance of the motor Ld, the in squaring inductance of the motor Lq and the electric constant of the motor Ke are analytically determined, Table 3 (Appendix). For the ob-tained values and for a switching frequency fc = 3.3kHz, the inverter provides three chopped and equilibrated voltages to supply motor armature, Figure 18.

5.1 Simulation Results

5.1.1 Speed Sequence

The desired speed sequence, Figure 19, requires three speed levels. Considering the parameters given by Table 3,

Figure 17. Generation of the inverter command constants

PI

0

PI

PI

dq

2 a

bc

PW

M g

ener

atio

n

Inve

rter

Iqref C1Vq Va+

PMSM

ab

c 2

dq

- + -

+

-

-

-

m

Id Vd

C2Vb

C3Vc



0 0.001 0.002 0.003 0.004 0.005 0.006 0.007 0.008 0.009 0.01-200

-150

-100

-50

0

50

100

150

200

Time (s)

Ch

op

ped

vo

ltag

e (

V)

Figure 18. Motor phase chopped voltage provided at the output of the inverter

the motor speed carried out using the traction elaborated traction chain model is illustrated in Figure 19. Analyzing this figure, one can notice that the vehicle speed reaches the wanted value in a relatively weak time.

5.1.2 Circulation Mission

To validate the use of a motor for traction application, the INRETS tests the behaviour of such motor using the speed instruction illustrated in Figure 20(a). Such speed instruction is called circulation mission. It consists of a normalised trial for vehicle motor tested for long dis-tances and variable speed under hard constraints.

In the last section, the designed motor was able to fol-

low a desired speed instruction. For the present study, let us consider the INRETS circulation mission and investi-gate the motor behaviour.

Figure 20(b) shows that the electric vehicle speed fol-lows to the required circulation mission with a little delay as found in Subsection 5.1.1.

6. Conclusions

The present paper was devoted to the design of high power – low dimensions axial-flux permanent-magnet motor for electric vehicles. In a first step, the electrical and the geometrical parameters of the motor integrated in the whole traction chain was analytically calculated con-sidering the vehicle specifications. Then in a second step, we have been interested in a finite elements study in or-der to validate and complete the analytical obtained re-sults. It has been found, that the built analytical model provides accurate values of electrical and geometrical motor parameters. Furthermore, a comparison between the electromagnetic torque obtained by the trapezoidal configuration and one developed by the sinusoidal con-figuration was made using the finite elements study. The high ripple noticed in the wave form of trapezoidal mo-tor’s torque leads to discard this configuration and em-phasize the choice of sinusoidal configuration for traction applications. Finally, the designed sinusoidal motor be-haviour was investigated considering firstly a desired

0 2 4 6 8 10 12 14 16 18 20

0

10

20

30

40

50

60

70

80

Time (s)

Veh

icle

sp

eed

(K

m/h

)

Desired speed sequence

Electric vehicle response

Figure 19. Desired speed sequence and motor speed carried out using the developed traction chain model



0 100 200 300 400 500 600 700 800 900 10000

10

20

30

40

50

60

70

80

Time (s)

Veh

icle

Spee

d (K

m/h

)

0 100 200 300 400 500 600 700 800 900 1000

0

10

20

30

40

50

60

70

80

Time (s)

Veh

icle

spee

d (K

m/h

)

(a) INRETS circulation mission (b) Electric vehicle response

Figure 20. INRETS circulation mission and electric vehicle response

speed instruction and next the INRETS circulation mis-sion. For that, we have been involved in the modelling and implementation under matlab/simulink environment of such motor associated to six-switch three-phase in-verter and integrated in the whole traction chain. For both

tests, the motor was able to follow and to provide the requested features which make the sinusoidal axial-flux permanent-magnet motors serious competitors of con-ventional radial-flux permanent-magnet motors for automotive traction applications.

Appendix

Table 1. Motor specification

Parameter symbol value unit Vehicle mass Mv 800 kg

Wheel ray Rw 0.26 m

Basic velocity Vb 30 km/h

Maximum speed of the vehicle Vmax 80 km/h

Pole pair number p 4

Stating time td 4 s

Coefficient related to the inertia if the turning parts 1

Switched frequency fc 3.33 kHz

Reduction report rd 4

Gravity g 9.81 N/kg

External diameter Dext 350 mm

Internal diameter Dint 150 mm

Table 2. Motor dimensioning

Designation Symbol Trapeze Sine Report of the teeth number by the number of pole pairs Rndnp 1.5 1.5

Rotor occupancy rate by permanent magnets β 1 2/3

Report between angular width of a principal tooth by the angular width of a magnet Rldla 1 1

Report between the angular widths of an inserted tooth by the angular width of a principal toot. Rdid 0.2 0.2

Rotor disc thickness Hr 83.357 mm 55.611 mm

Permanent magnets height Hpm 6.873 mm 6.873 mm

Slots height Hth 69.671 mm 17.214 mm

Stator yoke thickness Hy 98.176 mm 65.498 mm

Slots mean angular width Asm 3° 12°

Slots width Ls 6.545 mm 26.132mm

Magnet mean angular width La 45° 30°

Principal tooth mean angular width Aptm 45° 30°

Inserted tooth mean angular width Aitm 9° 6°

Air-gap thickness e 2 mm 2 mm



Table 3. Electric parameters used for the simulation of the traction chain model

Parameter symbol value unit Direct voltage Udc 291 V

Drag coefficient Cx 0.55

Frontal surface Sf 1.8 m²

Coefficient to bearing pneumatic resistance fr 0.01

Electric constant Ke 0.3

Phase resistance Rph 0.007 Ω

Direct inductance Ld 0.157 mH

In squaring inductance Lq 0.157 mH

REFERENCES

[1] Y. Amara, “Contribution à la conception et à la commande des machines synchrones à double excitation, Application au véhicule hybride,” Thèse de doctorat, université, PARIS VI, 2001.

[2] A. Parviainen, “Design of axial-flux permanent-magnet low-speed machines and performance comparison be-tween radial-flux and axial-flux machines,” thesis of doc-torate, Lappeeranta University of Technology, Lappeer-anta, Finland, 2005.

[3] A. Parviainen, M. Niemelä, J. Pyrhönen, J. Mantere, “Performance comparison between low-speed axial-flux and radial flux permanent-magnet machines including mechanical constraints,” Electric Machines and Drives, IEEE International Conference, pp. 1695–1702, 2005.

[4] S. Tounsi, “Modélisation et optimisation de la motori- sation et de l’autonomie d’un véhicule électrique,” Thèse de doctorat en Génie Electrique, Ecole Nationale d’Ingénieurs de Sfax-Tunisie, 2006.

[5] B. Multon and J. Bonal, “Les entraînements électro- magnétiques directs: Diversité, contraintes et solutions, La conversion électromécanique directe,” - ENS Cachan – SEE, 1999.

[6] P. Kurronen, “Torque vibration model of axial-flux sur-face-mounted permanent magnet synchronous machine,” Dissertation, Lappeenranta University of Technology, Finland, 2003.

[7] B. Tounsi, “Etude comparative de groupes électrogènes embarqués à large gamme de vitesse variable associant machines à aimants permanents et conversion statique,” Thèse de Doctorat, INP Toulouse, 2006.

[8] S. Tounsi, R. Neji, and F. Sellami, “Contribution à la

conception d’un Actionneur à Aimants Permanents pour Véhicules Electriques en vue d’Optimiser l’Autonomie,” Revue Internationale de Génie Electrique, RIGE, Vol. 9/6, pp. 693-718, Edition Lavoisier, 2006.

[9] C. Cavallaro, A. O. Di Tommaso, R. Miceli, A. Raciti, G. R. Galluzzo, and M. Trapanese, “Efficiency enhancement of permanent-magnet synchronous motor drives by online loss minimization approaches,” IEEE Transactions on Industrial Electronics, Vol. 52, No. 4, pp. 1153-1160, 2005.

[10] R. Neji, S. Tounsi, and F. Sellami, “Contribution to the definition of a permanent magnet motor with reduced production cost for the electrical vehicle propulsion,” European Transactions on Electrical Power, ETEP, Vol. 16, No. 4, pp. 437-460, 2006.

[11] R. Neji, S. Tounsi, and F. Sellami, “Optimization and design for a radial flux permanent magnet motor for elec-tric vehicle,” Journal of Electrical Systems, JES, Vol. 1, No. 4, 2005.

[12] N. Chaker, “Conception et commande d’un moteur à aimant permanent à flux axial pour véhicule électrique,” Mémoire de Mastère en génie électrique, Ecole Nationale d’Ingénieurs de Sfax, Tunisie, 2006.

[13] MAXWELL® 2D (2002) Student Version; A 2D Magne-tostatic Problem.

[14] F. Khatounian, S. Moreau, J. P. Louis, E. Monmasson, F. Louveau, and J. M. Alexandre, “Modeling and simulation of a hybrid dynamic system used in haptic interfaces,” Electrimacs, Hammamet, Tunisie, 2005.

[15] G. Kayhan, A. A. Ahmed, and P. Halit, “Improving the performance of hysteresis direct torque control of IPMSM using active filter topology,” S¯adhan¯a, Vol. 31, Part 3, pp. 245–258, 2006.




1


Y. Rosales Hernandez, T. Hiyama

Department of Computer Science and Electrical Engineering, Kumamoto University, Kumamoto, Japan. Email: [email protected], [email protected]

Received March 19th, 2009; revised May 25th, 2009; accepted May 28th, 2009.

ABSTRACT

This paper presents a rule-based technique to control the voltage in a power transmission network. Transformers with a tap changer installed in the system are selected by the proposed technique as control devices. For each bus under volt-age violation, the most effective control device is selected by using the minimum electric distance criteria. In order to demonstrate the efficiency of the method, several simulations were performed using an IEEE 30-bus network as a model system. The distance measure technique is compared with classic voltage regulation approach and a genetic algorithm based. The results obtained show the robustness of the proposed method.

Keywords: Knowledge Based Systems, Losses, Voltage Control

1. Introduction

Current approaches to the operation of a modern distribu-tion network demand high operational performance of the system and consequently require highly effective control strategies. Although voltage deviation control is one of the problems that has been extensively investigated, it still remains as an important topic to deal with. Voltage control algorithms may be classified into two categories: rule-based and network model-based. Rule-based algo-rithms use rules that control switched capacitors and transformer tap changers based on real-time measure-ments and past experience. Network model-based sys-tems use network topology, impedance, real-time meas-urements and statistical information to establish the cur-rent state of the system. It then applies optimization tech-niques to get the best possible solution. Within the net-work models-based systems there are many different ap-proaches. A simulated annealing technique for global optimal solution is presented in [1]. The authors propose a knowledge-based expert system which detects buses with maximum voltage deviations and operates the near-est available transformer control unit to correct the prob-lem. Then, a simulated annealing algorithm is utilized to solve the problem of capacitors manipulation. The paper shows a very good result in terms of power loss reduc-tions but does not guarantee an economical use of trans-formers operations. Restriction in the number of switch-ing operation is the focus in [2]. Here, dynamic pro-gramming and fuzzy logic algorithms are combined to control voltages and reduce power losses. The problem is

decomposed into two sub-problems: first, the control of the load tap changers (LTC) and capacitor banks at sub-station level and second, the control of the capacitor banks installed at the feeder level. Dynamic program-ming is used in sub-problem 1 and fuzzy logic is adopted for the second sub-problem. Simulation results show the excellent performance of the proposed approach. The use of genetic algorithms is another approach to the control of voltage and reactive power in the system. The ap-proach in [3] combines the benefits of a linearized system model and genetic algorithms (GA). Whenever a voltage correction is demanded, an initial calculation of the sen-sitivity matrix is done in order to identify an initial popu-lation for the GA. Then the GA finds a proper set of con-trol actions to execute. The method offers good solutions to the voltage/reactive power problem and also reduces the number of control actions. Authors in [4] use a method based on an artificial neural network to find the suitable capacitor switching regime for every load state. The main objective is to reduce power losses and the only constraint considered is bus voltage. The advantage of this method is the short calculation time. However, in real applications, it might be difficult to use because the sys-tem requires training sessions every time any small change is made to the network topology.

The approach presented in this paper is rule-based and is a new decision-making tool for centralized control of voltage. When the system lacks automatic function con-trol the task has to be performed manually by the super-

Distance Measure Based Rules for Voltage Regulation with Loss Reduction 86

visor in the dispatch center. Due to the complexity of a modern power system and the severe consequences to the economy of power failures, reliable algorithms have to be part of the daily support tools in the dispatch center. This research was motivated by the necessity to design a sim-ple and effective support algorithm for the voltage con-trol process. The algorithm is based on the identification of a bus having the worst voltage violations and the nearest bus where a voltage control device is installed. A control device setting is changed in order to improve the voltage situation of the bus in violation. A 30 bus net-work was used as a case study. Some classic control ele-ments such as transformers with tap changers, shunt ca-pacitor banks, synchronous condensers, and generators were modeled. Although the control strategy reported here is focused on tap changer, is possible to use all in-stalled devices as controllable elements. The important features of the case study system, the proposed method, and the search algorithm are explained in Sections 2-4. Simulation results of the 30-bus system under different load conditions are discussed in Sections 5 and 6.

2. Case Study System

The modeled system is an IEEE 30-bus scheme. The sys-tem bus data is given in Table 1, and with Figure 1 show-ing the single line diagram. Shunt capacitor banks are located at bus 10 and 24. The capacitor bank found at node 10 contains up to 10 units with a reactive power capacity of 1.9 Mvar for each unit. In the case of bus 24, banks have been installed containing up to 3 units of 0.8 Mvar each

Table 1. Bus data

Figure 1. 30-bus IEEE scheme

one. The tap changer settings ranges are modeled at set-tings from 0.9 to 1.1 with a step of 0.01 per unit. Four synchronous condensers are also considered at buses 5, 8, 11 and 13.

3. Proposed Method

Usually, the system is exposed to overload and un-der-load conditions in 24 hour intervals. When the system is in the overload condition, transferred power trough lines and transformers might causes excessive voltage drops and consequently appear bus voltages below the minimum limit. In the case of an under-load condition, shunt capacitance of the lines inject an excessive reactive power into the network and the voltage in some buses might be above the maximum limit. A safe voltage op-eration range is considered to be from 0.95 to 1.05 per unit. A rule-based approach is proposed to bring the sys-tem to a normal point of operation, with rules being pre-sented in Table 2. The ranking list order is based on the electrical distance criteria between every voltage control device and the target bus. Once the nearest voltage con-trol device is selected, the device settings have to be modified using a minimum number of steps in order to avoid unnecessary control actions.

Table 2. Voltage control procedure



4. Distance Measure Algorithm

The shortest route from the bus under worst voltage con-dition to a corresponding control device location is cal-culated using Dijkstra’s algorithm [5]. The basic opera-tion of this algorithm uses edge relaxation. In this case, the edges are the electrical distance Lij of the transmis-sion line between buses i and j. The electrical distance is defined in (1).

2 2

ij ij ijL R X (1)

where R: is the resistance of brach i-j X: is the reactance of brach i-j Once the minimum paths are found, a ranking of dis-

tance measures is established in order to develop a deci-sion strategy to solve the problem of voltage violation.

5. Simulation Results

For the controllable devices to have a long operating life it is vital to avoid unnecessary control actions. Therefore, only strictly necessary actions are allowed. A control effort index, CEI, is defined to count the number of con-trol actions used in every simulation. The CEI definition is presented below.

1

n refsCEI tap tapi i

i

(2)

where i: is the i-th controllable device s: actual tap position of the i-th controllable device ref : is the reference tap position of the i-th controllable

device In the initial state of the system, the voltage violations

are under the minimum voltage limit. It was for this rea-son that the shunt capacitors were not adjusted in these simulations. Also, it is important to note that the mini-mum tap modification is 0.01 in per unit so that if the CEI value is 0.36, it means that 36 operations of the tap were made.

The simulation results are shown in three parts. The first part is a comparison between a local control strategy, an evolutionary search based on a genetic algorithm and the distance based method. The second part illustrates the performance of the proposed method under a load varia-tion during a period of 24 hours. In the last section, there is also a power losses analysis.

5.1 A Comparison of Voltage Local Control, Ge-netic Algorithm Based Correction, and the Pro-posed Voltage Control Algorithm

Voltage local control is a classic method based on the local monitoring and operation of each control device. It

means that at every node where a control device is in-stalled, a local and independent control strategy is fol-lowed, and control actions are executed exclusively where voltage problems appear. Figure 2 illustrates an initial voltage profile of the network under a hypothetical load scenario which is assumed to be the maximum load scenario. The voltage profile shows several nodes violat-ing the minimum voltage limit. Buses which are under violation and where a control device is installed are marked with a circle. In this initial condition only trans-former operations are available because all capacitor banks are already connected.

Table 3 lists the positions of transformer taps in the initial state, during two partial solutions and for the final solution. The final solution is reached when the four buses highlighted in Figure 2 are out of the violation zone. The final solution, shown in Figure 3, does not solve the problem of voltage at nodes other than those where the control devices are installed.

The second reference point for this comparison is a genetic algorithm (GA). GAs are considered more flexi-ble and robust than most of deterministic search methods because it requires only information concerning the qual-ity of the solution produced by each parameter set. This is unlike many traditional methods that require derivative information or worse yet, completed knowledge of the problem structure and parameters [6]. For this GA, deci-sion variables are expressed as integers. Each gene represents the tap position of a transformer. Integer vari-ables are used in order to avoid unnecessary coding and recoding. By using this non-binary coding, which is a closer representation of real system parameters, it is ex-pected that there should be an increment in the velocity of convergence [7]. The representation of one individual is shown in Figure 4. The initial population is generated randomly.

Figure 2. Voltage profile of the network obtained for initial conditions



Table 3. Operation of the controllable devices using the con-trol method of local voltage

Figure 3. Voltage profile of the network obtained after exe-cution of voltage local control

Figure 4. Integer representation of one individual

Then, chromosomes are evaluated through a fitness

function (see Equations 3 and 4) where the objective function is the minimum number of adjustments to the tap changers. The voltage deviation at each bus, the reac-tive limit violation at each generator and maximum line current limit are considered as constraints. The evaluation is based on Newton-Raphson power flow calculations, provided by the MatPower package [8]. The genetic op-erators are tournament selection, one-point crossover, and uniform mutation. The stopping criterion is the number of generation being 60 with the probability of mutation be-ing 15%.

min refstap tap Ri i (3)

and

* * *R a vd b ql c cl (4)

where s: actual tap position i: ith-tap transformer ref : reference of tap position vd: violation of voltage deviation ql: violation of generated reactive power cl: violation of current in lines a: weight for violation of limits of voltages b: weight for violation of limits of generated Var c: weight for violation of limits of current flowing

through lines In order to get a clear solution with the GA, a total of

45 independent simulations were executed with a com-mon initial condition, (the conditions being as shown in Figure 2). Figure 5 shows the mean value of voltage de-viation factor at each generation. The mean value of number of control actions are shown in Figure 6, where most of the simulations reach a common solution with fewer than 43 operations. The mean value of fitness for each generation are illustrated in Figure 7. In these three Figures each curve represent one of the 45 simulations. The superposition of the curves demonstrates the similar-ity of the solutions for each simulation. The best solution at each simulation are shown in Figure 8. With 41 control operations being the minimum value that can be reached by the GA. The best solution for each simulation has no constraint violations. For example, Figure 9 shows the best solution for the voltage profile simulation number 45.

Figure 5. Mean value of voltage deviation factor for the 45 simulations

Figure 6. Mean value of the number of operations for the 45 simulations



Figure 7. Mean value of fitness for the 45 simulations

Figure 8. Number of control action for the best solution at each simulation

Figure 9. Voltage profile for the best solution at simulation number 45

Table 4. Operation of the controllable devices using the

distance measure method

In the case of the proposed rule-based method, the ini-tial state is the same as that showed in Figure 2. The worst voltage is located at node 30 and transformer T7 is

the best control device to solve the problem. The tap po-sition in transformer 7 was moved from 0.96 to 1.00 and the voltage problem in node 30 was solved. Then bus 19 appeared as the worst bus and the most effective control device was transformer 2. The process was repeated sev-eral times until a final solution was reached. Table 4 shows the initial conditions of tap positions, two partial solutions and the final solution. Values in boldface font

Figure 10. Voltage profile of the network obtained in sub- solution 1

Figure 11. Voltage profile of the network obtained in sub- solution 2

Figure 12. Voltage profile of the network obtained in the final solution



represent a new modification of the tap position. In Fig-ure 10, 11 and 12 the voltage profile for the two partial solutions and the final result are presented respectively. Final voltage profile shows the capacity of the rule-based method to find a suitable solution, and the total CEI=0.42, means that the number of control actions is 42, which is very close to the optimal solution of the GA-based method.

5.2 Performance of the New Method Applied for a Load Variation over a 24 Hour Interval

It is well known that power demand in a real system is changing continually during the day, and consequently state variables are varying as well. Thus, it is necessary to study the effectiveness of the proposed method for this typical behavior. Load variation was modeled as coinci-dent in time. Appendix A shows the percentage of the rate load at every bus and the voltage at 6 buses after application of the distance measure method. Other than the buses shown in Appendix A, the rest are kept within the non-violating voltage zone. The variations in the transformer taps are illustrated in Appendix B. In the case of capacitor bank adjustment, none were executed be-cause all the banks were connected in the initial state and the voltage violations that appeared were of the un-der-voltage type.

6. Analysis of Power Loss Reduction

In addition to voltage correction, power losses were also monitored and analyzed. This new control method yields a very flat voltage scenario which is very important in order to reduce power loss. Appendix C illustrates how the power losses are reduced gradually in each partial solution obtained by the proposed method. The final so-lution gives a 0.67 % power loss reduction.

7. Conclusions

In this paper a Rule-based method was presented for regulating voltage deviations and to reduce power losses of a transmission system. The control method is based on simple rules. Which allow to operate only the most effec-tive devices to solve voltage violations. Thus, control actions were executed under the principle of imposing the fewest number of operations of control devices. Several simulations were done to compare a local voltage control strategy and GA-based method with the new method. The results proved that:

1) The new method achieves the goal where the local voltage control strategy fails. The most important issue, which is voltage correction, is not successfully accom-plished with the approach based on local control.

2) The rule-based method was compared with several simulations of a GA-based method and the results are

very similar. The number of control actions from GA-approach is 41 while for the rule-based method is 42. There are no constraint violations in the solutions pro-vided by both methods.

3) The rule-based method significantly reduces power losses of the system under maximum load condition and under a load variation period of 24 hours. Voltages at all buses were maintained out of the voltage violation zone.

4) Although the proposed method is based on very simple rules, where significant approximations are used to determine a ranking list of effective controllable de-vices, this approach can be used as a useful, simple and fast tool for dispatcher engineers in a situation requiring correction of voltages.

Appendix A




91

Appendix C

Appendix B

REFERENCES

[1] T. Ananthapadmanabha, “Knowledge-based expert sys-tem for optimal reactive power control in distribution system,” International Journal of Electrical Power and Energy Systems, Vol. 18, pp. 27-31, January 1996.

[2] Y. T. Liu, Z. G. Peng, and X. Z. Qiu, “Optimal volt/var control in distribution systems,” International Journal of Electrical Power and Energy Systems, Vol. 24, pp. 271-276, May 2002.

[3] Y. Malachi and S. Singer, “A genetic algorithm for the corrective control of voltage and reactive power,” IEEE Transactions on Power Systems, Vol. 21, pp. 295-300, February 2006.

[4] B. Das and P. K. Verma, “Artificial neural network-based optimal capacitor switching in a distribution system,” Electric Power System Research, Vol. 60, pp. 55-62, June 2001.

[5] H. I. Hagenaars, J. Imura, and H. Nijneijer, “Approximate continuous-time optimal control in obstacle avoidance by time/space descrifization of non-convex state constraints,” Proceedings of 2004 IEEE Conference on Control Appli-cations, pp. 878-883, 2004.

[6] K. Y. Lee and M. A. El-Sharkawi, “Modern heuristic op-timization techniques: Theory and applications to power systems,” New Jersey: Wiley-IEEE Press, pp. 173, 2008.

[7] Y. Rosales and T. Hiyama, “A review of genetic algo-rithms implemented for voltage/var optimization prob-lems in electric network systems,” submitted for publish-ing.

[8] MATPOWER, http://www.pserc.cornell.edu/matpower/.

http://www.pserc.cornell.edu/matpower/



1

Partial Discharge Source Classification and De-Noising in Rotating Machines Using Discrete Wavelet Transform and Directional Coupling Capacitor

Mohammad Amin Kashiha, Diman Zad Tootaghaj, Dolat Jamshidi

Communications Department, Niroo Research Institute (NRI), Tehran, Iran. Email: [email protected], [email protected], [email protected]

Received March 28th, 2009; revised May 22nd, 2009; accepted May 28th, 2009.

ABSTRACT

This paper introduces a new method to separate PD1 from other disturbing signals present on the high voltage genera-tors and motors. The method is based on combination of a pattern classifier, the Discrete Wavelet Transform (DWT), to de-noise PD and Time-Of-Arrival method to separate PD sources. Furthermore, it will be shown that it can recognize PD sources including rotating machine’s internal and external discharge pulses (e.g. on the bus bar).

Keywords: Partial Discharge, Discrete Wavelet Transform, Time-Of-Arrival, Rotating Machines, De-Noising, Cou-pling Capacitor

1. Introduction

As a result of deterioration of insulating systems in high voltage equipments, small electrical spikes occur within the insulation [1]. This could cause further degradation of insulation and finally failure of the equipment. So, insu-lation assessment of these equipments is necessary in order to avoid catastrophic consequences. During last decades a huge number of studies have been done on recognizing Partial Discharge (PD) pulses in high voltage equipments including rotating machines. Although there have been good achievements in this field [1], still there is a long way to introduce a method to separate PD from noise and interferences in a perfect way. There are sev-eral approaches to extract PD and there are a handful of papers on each of them. Reference [2] discusses a method based on fuzzy classification of PD. Although it is par-tially successful in recognizing cross-coupling resulted from adjacent phases, its authors admit that certainty of their method is not high. There are also some methods based on artificial intelligence [2] but they suffer from low generality and high calculation needs. Nowadays it is proved that time-frequency transforms such as Wavelet [3,4] and Hilbert–Huang transform [5] have the best per-formance in PD de-noising. As authors investigated al-

most all of published methods and most of survey papers approve it too [6,7], the best and most successful method to extract PD from noise and interferences is DWT. Hence, the approach was developed based on DWT de-noising but the method suffers from disability of separating PD originating from rotating machine and the bus bar. Thus, a complementary technique (Time-Of-Arri- val) which is based on the method introduced in [8] was used. The idea of the latter is based on directional cou-pling capacitors as conventional sensors used to attach measurement instruments to high voltage windings.

The paper is organized as follows: Section 2 clarifies problems dealt with in PD de-noising. Section 3 discusses DWT and its application in PD de-noising. Section 4 ex-plains the Time-Of-Arrival technique. Section 5 discusses the results obtained and Section 6 concludes the paper.

2. Problem Definition

PD measurement is a main concern of operators dealing with generators and motors as they want to avoid ma-chine failure. On the other hand, they prefer to do this while the machine is operating (i.e. on-line measurement) because detaching a generator from the network is costly and time-consuming. Meanwhile in on-line operation of 1Partial Discharge

mailto:[email protected]


Partial Discharge Source Classification and De-noising in Rotating Machines 93 Using Discrete Wavelet Transform and Directional Coupling Capacitor


these machines, there are different kinds of noise and interference signals that make measurements unreliable. Therefore, a method is needed to separate PD from these signals. The method this paper follows is based on DWT which is a time-frequency transform. As known, PD is a non-stationary signal [9]. So, conventional transforms such as Fourier Transform (FT) may not be used to ana-lyze spectral specifications of PD as they do not distin-guish short-term and long-term frequency components. But DWT considers time events of the signal. Thus, it is capable of interpreting short-time PD pulses. Next sec-tion discusses PD de-noising using DWT.

3. PD De-Noising Using DWT

Wavelets have very attractive features which cause them to be used in miscellaneous applications [10]. One of the methods which works based on these features is decom-posing and reconstructing signals using QMF2 filters. Reference [2] is a good context to understand how DWT decomposition using QMF filters works. But here the focus is on applying this technique to develop the method. In general, DWT decomposes a signal to its basic fre-quency components as shown in Figure 1.

Recorded signal from sensors includes both PD and noise. It is known that noise has a stochastic nature. So, it is expected that its energy3 is divided equally between filter bands. But PD’s energy is mostly concentrated in a few bands [9]. Energy of the coefficients is a reliable cri-terion to separate bands which may contain PD more than noise. Author’s experience showed that using 2nd order Daubechies (db2) mother wavelet [9], more than 80% of PD signal’s energy is gathered in one of the detail coeffi-cients (cDs) of DWT. This is a useful result which could be considered to determine global threshold and finally separate PD from noise. Energy distribution of a sample is shown in Figure 2. As it is seen cD6 (detail coefficient of 6th decomposition level) includes most of PD’s energy.

Figure 1. The tree structure of the DWT [11]; cDn is the detail .coefficient of nth decomposition level

Figure 2. Energy distribution of the signal on each level, using db 2

Using mentioned method, thresholds are calculated (in fact, we train the system.). The method we calculate a threshold is called soft-thresholding and is discussed in [9]. Then decomposing is repeated and calculated thresholds are exerted to make weak samples of the coefficients zero. Weak samples are supposed to be related to noise com-ponents. Figure 3 shows the reconstructed signal using soft-thresholding in comparison with original noise-pol-luted PD signal.

4. A Modification to the PD De-Noising Method

The method introduced so far, suffers from a defect in recognizing PD sources. There are four types of signals that may reach PD-Analyzer (PDA) equipment:

1) Internal PD (i.e. PD pulses originating from rotat-ing machine)

2) External PD (i.e. PD pulses originating from bus bar) 3) Internal noise (i.e. noise originating from other

sources except discharges) 4) External noise (i.e. noise originating from external

agents including interferences caused by communication systems or PDA itself)

The problem of the mentioned method (and generally pattern-based methods) encounter is that it cannot distin-guish external PDs from internal PDs because it works based on pulse shape and is not dependant on the direc-tion PD comes from.

Hence, a technique is needed to separate internal and external PD pulses. The method used to amend the ap-proach works based on utilizing directional couplers [10]. Figure 4 shows the overall system measuring PD by this technique. This method (called Time-Of-Arrival) was initially used in an analogue system to separate internal and external PD [7]. But a major disadvantage of the analogue system is that one should design a system which works in frequencies upper than 50 MHz [12] because it does not utilize any de-noising technique. Therefore, noise eliminating is done with high-pass filtering because there is no noise or interference in this system at those frequencies.

2Quadrature Mirror Filter 3Energy definition and formulation of calculating coefficients’ energy is introduced in [11].

94 Partial Discharge Source Classification and De-noising in Rotating Machines Using Discrete Wavelet Transform and Directional Coupling Capacitor

Figure 3. Noise-polluted PD pulse in comparison with the coefficient with highest energy

Figure 4. Different kinds of signals reaching PDA

Signals shown in Figure 4 are not generally present in

every generator or motor. For example in hydro-generators there is some negligible internal noise that does not affect the measurement. Also external PD may exist or not based upon the characteristics of the bus system.

In previous sections a method to separate noise from PD was discussed. So, the only issue which is remained is to distinguish internal and external PD.

The “Time-Of-Arrival” technique helps us to separate these two signals because cable length of sensors’ outputs is different so that external PD pulses arrive at PDA ports simultaneously but internal PD pulses arrive at different times [7]. By capturing the two ports of PDA in fixed sample steps using appropriate data acquisition systems [13,14], we extract internal PD in two steps:

1) De-noising PD using DWT as discussed in Section 3. 2) Separating internal and external PD pulses using

Time-Of-Arrival technique.

5. Experimental Results

At first step the proposed method was tested successfully on simulated data. Different kinds of noise and interfer-ence (including AM interference, sinusoid harmonies and Gaussian noise) were exerted on simulated DEP4 and DOP5 [4] PD pulses. Then de-noising algorithm intro-duced before was applied to them. Figure 7 shows a sam-ple of simulation results.

For further evaluation of the proposed method, re-cordings from NEKA6 power plant were made, which is equipped with 4*440 MW turbo-generators. Various kinds of noise were observed at the output of coupling capacitors attached to the output of the generators. Main noise types were harmonies and Gaussian noise and the total mean SNR was about 3 dB. Applying the proposed method to recorded signals and comparing results with observations of experts showed that the method recog-nizes PDs with a mean error as much as 3.2% for FAR7 and 2.9% for FRR8 (FAR happens when a noise pulse is recognized as PD and FRR happens when a PD is recog-nized as noise.). A sample of PD de-noising of recorded data is shown is Figure 8(a) and 8(b).

As the result of this work will be used to manufacture a PD Analyzer in NRI, major methods of de-noising PD to measure its level and frequency band had to be consid-ered. Therefore two successful methods i.e. fuzzy logic presented in [2] and the method presented in this paper were implemented on a high technology hardware com-prised of the followings:

Figure 5. A simple block diagram of PD analyzer system used

4Damped Exponentially Pulse 5Damped Oscillatory Pulse 6NEKA is the most important and oldest power plant in north of Iran 7 False Acceptance Rate 8 False Rejection Rate

Figure 6. Pulse height diagram of PD signals measured in NEKA power plant; March 2008 (diagram no.1: fuzzy method, diagram no.2: DWT method); September 2008 (diagram no.3: fuzzy method, diagram no.4: DWT method)


Partial Discharge Source Classification and De-noising in Rotating Machines 95 Using Discrete Wavelet Transform and Directional Coupling Capacitor


1) Data acquisition system from National Instruments (NI PCI-5154) with sampling rate of 2 GS/s.

2) Online storage and processing unit based on Xilinx Virtex-II FPGA.

3) Data presentation on PC in Labview software. Figure 5 shows a simplified block diagram of the system. Figure 6 shows the pulse height diagram of PD

pulses of NEKA power plant in March (diagrams number 1 and 2) and September (diagrams number 3 and 4) 2008. Diagrams number 1 (in black color) and 3 (in red color) are related to fuzzy logic PD de-noising and dia-grams number 2 (in blue color) and 4 (in green color) are the results of the method presented in this paper. Com-parison shows that the method proposed in this paper yields accuracies as well as the fuzzy method, although it benefits from lower complexity.

6. Conclusions

DWT is a powerful method in PD de-noising of rotating machines. But it is very important that appropriate tools be used to yield enough accuracy. It was found that PD de-noising using Daubechies mother wavelets and soft-thresholding based on maximum energy of decom-position coefficients yields best results. This paper also proposed a method to obviate a defect of DWT method to separate internal and external PDs of rotating machines. The method is based on Time-Of-Arrival theory and is digitally implemented. Testing of the method showed that it can recognize PDs with a mean error as much as 3.2% for FAR and 2.9% for FRR. Finally, results of the pro-posed method were compared to a reference method that was using fuzzy algorithms.

Figure 7. PD de-noising simulation results, (a) Simulated PD signal; (b) Noise–polluted PD; (c) De-noised PD using seven levels of decomposition and db2 mother wavelet

Figure 8(a). Noise-polluted signal (recorded from NEKA power plant at 27th March 2008)

Figure 8(b). De-noised PD using proposed method

REFERENCES

[1] N. C. Sahoo, M. M. A. Salama, and R. Bartnikas, “Trends in partial discharge pattern classification: A survey,” IEEE Transactions on Dielectrics and Electrical Insulation, Vol. 12, No. 2, April 2005.

[2] A. Contin, et. al., “Digital detection and fuzzy classification of partial discharge signals,” IEEE Transactions on Dielec-trics and Electrical Insulation, Vol. 9, No. 3, June 2002.

[3] L. Satish and B. Nazneen, “Wavelet-based denoising of partial discharge signals buried in excessive noise and in-terference,” IEEE Transactions on Dielectrics and Elec-trical Insulation, Vol. 10, No. 2, April 2003

[4] H. Zhang, T. R. Blackburn, B. T. Phung, and D. Sen, “A novel wavelet transform technique for on-line partial dis-charge measurements Part 1: WT de-noising algorithm,” IEEE Transactions on Dielectrics and Electrical Insulation, Vol. 14, No. 1; February 2007; I. S. Jacobs and C. P. Bean, “Fine particles, thin films and exchange anisotropy,” in Magnetism, Vol. 3; G. T. Rado and H. Suhl, Editors, New York, Academic, pp. 271–350, 1963.

[5] X. D. Wang, et al., “Analysis of partial discharge signal using the Hilbert–Huang transform,” IEEE Transactions on Power Delivery, Vol. 21, No. 3, July 2006.

96 Partial Discharge Source Classification and De-noising in Rotating Machines Using Discrete Wavelet Transform and Directional Coupling Capacitor


[6] S. Sriram, S. Nitin, K. M. M. Prabhu, and M. J. Bastiaans, “Signal denoising techniques for partial discharge meas-urements,” IEEE Transactions on Dielectrics and Electri-cal Insulation, Vol. 12, No. 6, December 2005.

[7] G. C. Stone and V. Warren, “Objective methods to inter-pret partial-discharge data on rotating-machine stator windings,” IEEE Transactions on Industry Applications, Vol. 42, No. 1, January/February 2006.

[8] E. Hernandez and G. Weiss, “A first course on wavelets,” CRC Press, 1996.

[9] X. Ma, C. Zhou, and I. J. Kemp, “Interpretation of wave-let analysis and its application in partial discharge detec-tion,” IEEE Transactions on Dielectrics and Electrical In-sulation, Vol. 9, No. 3, June 2002; J. Clerk Maxwell, “A treatise on electricity and magnetism,” 3rd Edition, Vol. 2. Oxford, Clarendon, pp.68–73, 1892.

[10] IEC 60270, “High-voltage test techniques - partial dis-charge measurements”, 3rd Edition, December 2000.

[11] X. X. Zhou, C. Zhou, and I. J. Kemp, “An improved methodology for application of wavelet transform to par-tial discharge measurement denoising,” IEEE Transac-tions on Dielectrics and Electrical Insulation, Vol. 12, No. 3, June 2005; K. Elissa, “Title of paper if known,” unpub-lished.

[12] IEEE Standard 1434-2000, IEEE trial guide to the meas-urement of partial discharges in rotating machinery, 2000.

[13] S. H. Wang, F. C. Lu, and Y. P. Liu, “High-speed data acquisition system for partial discharge on-line monitor-ing in transformer,” Proceedings of the 14th International Symposium on High Voltage Engineering, Tsinghua University, August 2005.

[14] J. Borghetto, “Partial discharge inference by an advanced system: Analysis of online measurements performed on hydro-generator,” IEEE Transactions on Energy Conver-sion, Vol. 19, No. 2, June 2004.



1


W. L. Huang1, J. F. Cui1, G. S. Sun2

1School of Mechatronics Engineering, Zhengzhou Institute of Aeronautical Industry Management, Zhengzhou, China; 2Institute of Electrical Engineering, Chinese Academy of Sciences, Beijing, China. Email: [email protected]

Received February 17th, 2009; revised March 24th, 2009; accepted April 2nd, 2009.

ABSTRACT

Under nanosecond pulsed coaxial electric field, surface flashover voltage over the interfaces between nylon 1010 and transformer oil increases almost linearly with gap length, and the steeper rising edge of applied pulse, the higher flash-over voltage. Surface flashover properties are closely related to the electric field at the triple junctions of solid-liquid-electrode and the field gradient along the interfaces. Although the increased difference between inner and outer electrode radii will enhance electric field strength at the triple junctions and nonuniformity degree of potential distribution along interfaces, it reduces simultaneously terribly the surface field strength of coaxial inner electrode, so that flashover voltage doesn’t descend, but ascends almost linearly with gap length. The average flashover strength in coaxial electric field can be estimated by that in uniform electric field for large enough difference between inner and outer electrode radii, which is useful to practical engineering design for coaxial pulsed power apparatuses.

Keywords: Pulsed Power Technology, Coaxial Electrodes, Gap Length, Surface Flashover, Triple Junctions, Surface Potential

1. Introduction

Although coaxial pulsed power apparatuses are wide-spread, researches on discharge characteristic over trans-former oil/solid interfaces under nanosecond pulsed co-axial field, as opposed to discharge characteristic over gas/solid or vacuum/solid interfaces under uniform field, are less, which can’t meet the rapid development of pulsed power technology [1-7]. Surface flashover, whether applied dc, ac or impulse voltages, are closely relevant to potential distribution gradient along interfaces between different dielectrics. Trinh [8] and Menju [9], respectively, pointed out the relationship between flash-over field strength and different parameters of coaxial experimental electrodes might be speculated on through corresponding states of potential distribution along the interfaces and field strength at triple junctions. Liu [10] plotted the varying curve of breakdown voltage vs. non-uniformity degree of electric field in electrode gap in transformer oil, discovered that breakdown voltage in uniform field was about three times higher than that in nonuniform field under applied impulse voltage of pulse-width of 40 µs, and achieved that change of uni-formity degree of electric field in gap worked on break-down voltage more prominent under impulse voltage than under ac voltage. This contribution analyzed flashover

characteristic over transformer oil/nylon 1010 interfaces under coaxial nanosecond pulsed field in the case of changing radius difference between inner and outer elec-trodes from the viewpoint of potential distribution over the interfaces and field strength at triple junctions, and compared flashover properties in coaxial field with ones in uniform field, thus draw conclusions useful to studying flashover over interfaces between different dielectrics and to the developments of pulsed power technology.

2. Experimental Setup and Methods

The entire structures of electric connection of experi-mental setup are shown in Figure 1. The high voltage nanosecond pulsed power supply is SPG200 based on Semiconductor Opening Switch (SOS) in Northwest In-stitute of Nuclear Technology, whose amplitude of output voltage which can reach maximum 350 kV is modulated by adjusting the resistor of cycling salt water. A resistor of 200 Ω is in series with electrode gap in order to avoid excessive current of gap breakdown impairing SPG200 [11,12]. An impulse of pulse-width of about 30 ns, rising edge of about 10 ns, is produced by SPG200. Accord-ingly, the waveforms of breakdown voltage and current


Analysis of Flashover Characteristics under Nanosecond Pulsed Coaxial Electric Field 98

Figure 1. Schematic diagram of experimental setup

Figure 2. Voltage and current waveforms with no gap

Figure 3. Breakdown voltage and current waveforms

for the direct short gap and for flashover, respectively, are shown in Figure 2 and Figure 3. Breakdown voltage and current signals, whose data acquisition and process-ing were executed by an oscilloscope of Tektronic684C combined with a computer, were attained by a capacitor divider and a resistor divider respectively [13].

The coaxial brass testing electrodes shown in Figure 4 should be characteristic of carefully coaxial structures, should be ensured to contact tightly with experimental solid samples and to replace solid samples conveniently, should be long enough in axis to avoid the influence of edge-effect of coaxial field on flashover over solid/liquid interfaces [14]. Solid samples of nylon 1010 with axial length of 6 mm were given shape to round ring. The liq-uid dielectric was transformer oil, marked 45 numbers, with industrial desiccating, filtrating and degassing.

Flashover voltage was described as the voltage applied between inner and outer electrodes while gap breakdown happened at the front edge of impulse voltage whose am-plitude was about 135 kV, waveform of 10/20 ns, and two kinds of steepness of rising edge of 7.8 kV/ns and 8.6 kV/ns. The change of gap length of 1, 2, 3, 4 mm was achieved through the according change of diameter of outer electrode of 8, 10, 12, 14 mm, at the same time re-taining the diameter of inner electrode of 6 mm. The total number of solid samples in same size and shape was five, and every solid sample was tested four times between them an interval of 10 min was introduced [15,16]. It is very necessary to replace solid sample, transformer oil after flashover for four times and experimental electrodes as etched spots have shown on the surface of electrodes.

3. Results and Discussion

3.1 Flashover Voltage and Gap Length

For two different kinds of steepness of rising edge of applied voltage impulse, experimental data for flashover voltage Uf of nylon 1010 vs. gap length D were shown in Figure 5 in which every datum point was the average of 15 times tests and was associated with a standard devia-tion. It is observed from Figure 5 that Uf of nylon 1010 increases almost linearly with D. As a result of surface flashover of rising edge of applied impulse, Uf is hardly

Figure 4. Brass electrodes



Figure 5. Flashover voltage according to variable gap length affected by the amplitude of applied impulse, whereas remarkably by the rising edge of applied impulse. For example, Uf of nylon 1010 is enhanced about 15 kV with the increase of 1 kV/ns in the steepness of rising edge of applied impulse at the gap length of 3 mm.

3.2 Simulation Analyses of Experimental Results

3.2.1 Field Strength at Triple Junctions

The factor of field strength improvement at triple junc-tions is defined by

= Ej / Em, (1)

where Ej is the field strength at triple junctions, Em is the average field strength along solid/transformer oil inter-faces and denoted by

Em = U0 / D, (2)

where U0 is the amplitude of applied static voltage, D is gap length. A i for field strength improvement factor at triple junctions at inner electrode surface as well as a o for that at outer electrode surface varying with gap length D is shown in Figure 6 in whose legend are i and o. From Figure 6, on the one hand the i, far greater than 1, increasing rapidly with D indicates the field strength Ei at triple junctions at inner electrode surface is enhanced to a certain extend and is, along with the increase of D, more and more higher than the average field strength Em; on the other hand the o, at all time less than 1, decreasing nearly linearly with D indicates the field strength Eo at triple junctions at outer electrode surface is always lower than Em. The above opinions imply, under coaxial non-uniform field, because of Ei far greater than Eo surface flashover over solid/transformer oil interfaces should begin from the triple junctions at the inner electrode sur-face whose physical, chemical natures would character flashover properties. As a result for a single reason, the increase of D would enhance Ei, consequently reduce Uf which contradicts the experimental phenomena.

3.2.2 Potential Distribution along the Interfaces be-tween Nylon 1010/Transform Oil

The nonuniformity degree γ of potential distribution over the interfaces between nylon 1010/transformer oil is de-fined by

γ = Pmax /Pmin, (3)

where Pmax, Pmin respectively stand for the maximum, minimum potential gradient in unit length along the in-terfaces. That the γ increases almost linearly with D should be the other reason for declining Uf, which also contradicts the experimental phenomena.

3.2.3 Field Strength at the Surface of Inner Electrode

When flashover occurs, that the field strength Ein at the inner electrode surface varies with gap length D is shown in Figure 8, where Ein reduces nonlinearly rapidly with D. This is the only positive reason for Uf to increase with D. From Figure 6 and 7, Uf should decrease with D, but in-versely from Figure 8. So under the combined effects of the above three reasons Uf behaves to increase almost linearly with D. It is sure that the positive effect of Ein on the surface flashover shown in Figure 8 is intense enough to counteract the negative effects of and γ shown in Figure 6 and 7 re-spectively, so that Uf increased almost linearly with D.

Figure 6. Field strength at triple junctions vs. gap length

Figure 7. Potential distribution along the interfaces vs. gap length



Figure 8. Field strength at the inner electrode surface vs. gap length

Figure 9. Flashover comparison between coaxial and uni-form field

The above statement indicates , in a general way, the higher field strength at solid-liquid-electrode triple junc-tions as well as the more nonuniform potential distribu-tion over the interfaces between different dielectrics should lessen the resultant flashover voltage, however, under the actual circumstances, one should think over all sorts of factors working on flashover, otherwise would receive false results.

3.3 Comparing the Results under Coaxial Elec-tric Field with Those under Uniform Field

The average flashover field strength Eav in the coaxial field was compared with that in the uniform field in Fig-ure 9. Eav equals to flashover voltage Uf divided by gap length D in the coaxial and uniform field respectively, i.e.

Eav = Uf / D. (4)

The Eav in the coaxial field is lower than that in the uniform field for the same solid samples, applied impulse voltage and gap length; For example, the Eav in the coax-ial field is about half of the one in the uniform field at the gap length of 1 mm. With the increase of D, the differ-ence of Eav between two types of electric fields is dimin-ished and the decline of both Eav with D is also stabilized

gradually. That two curves of Eav vs. D in Figure 9 ap-proaches each other after surpassing a relatively greater gap length ( > 4 mm under this experimental conditions , for example) reveals it is sensible to speculate on flash-over voltage under coaxial field by the one under uniform field, which is valuable to the industrial designs of pulsed power equipments of coaxial structure.

4. Conclusions

The flashover voltage along the interfaces between nylon 1010/transformer oil increases nearly linearly with gap length under nanosecond pulsed coaxial electric field.

The composition effects of enhanced field strength at triple junctions, heightened nonuniformity degree of po-tential distribution over the interfaces between nylon 1010/ transformer oil and descended sharply field strength at the inner electrode surface make flashover voltage increase almost linearly with gap length whose increase would lead the coaxial field to tend to the very nonuniform field under point-plane electrodes.

Flashover natures are closely related to the electric field strength at triple junctions of solid-liquid-electrode and the potential distribution along the interfaces between different dielectrics. It is, in a general way, sure that the higher field strength at triple junctions and the more nonuniform potential distribution along the interfaces between different dielectrics would result in the lower flashover voltage, whereas in practice one should take all kinds of factors controlling flashover into account for correct results.

While gap length of coaxial electrodes is large enough, the average flashover field strength in coaxial field can be estimated by means of that in uniform field, which is rather significant to the design of coaxial structural pulsed power apparatus.

5. Acknowledgments

The authors gratefully acknowledge the support and as-sistance in experiment by Room 3 and 6 in Northwest Institute of Nuclear Technology.

REFERENCES

[1] X. S. Liu, “High power pulsed technology,” Beijing: Na-tional Defence Industry Press, 2005.

[2] Z. Z. Zeng, “An introduction to practical pulsed power technology,” Xi’an: Shanxi Science and Technology Pub-lishing House, 2003.

[3] J. C. Martin, “Nanosecond pulse techniques,” Proceedings of the IEEE, Vol. 80, No. 6, pp. 934–945, 1992.

[4] T. H. Martin, A. H. Guenther, and M. Kristiansen, “J. C. Martin on pulsed power,” New York: Plenum Press, 1996.




101

[5] M. Buttram, “Some future directions for repetitive pulsed power,” IEEE Transactions on Plasma Science, Vol. 30, No. 1, pp. 262–266, 2002.

[6] A. H. Sharbaugh, J. C. Devins, and S. J. Rzad, “Progress in the field of electric breakdown in dielectric liquids,” IEEE Transactions on Electronic Insulation, Vol. 13, No. 4, pp. 249–276, 1978.

[7] W. L. Huang, G. S. Sun, and P. Yan, “Overview of flash-over properties over solid-liquid interfaces under nano-second pulses,” High Voltage Engineering, Vol. 31, No. 9, pp. 50–52, 2005.

[8] N. G. Trinh, F. A. M. Rizk, and C. Vincent, “Electro-static-field optimization of the profile of epoxy spaces for compressed SF6-insulated cables,” IEEE Transactions on Power Apparatus and Systems, Vol. 99, No. 6, pp. 2164– 2174, 1980.

[9] S. Menju, Y. Tsuchikawa, and N. Kobayashi, “Electric potential and field of conical insulators for SF6 metal-clad switchgear,” IEEE Summer Meeting and International Symposium on High Power Testing, Portland, USA, pp. 390–398, 1971.

[10] Q. C. Liu, “Electrical insulation structural design principle (the lower volume),” Beijing: China Machine Press, 1987.

[11] J. C. Su, G. Z. Liu, Z. J. Ding, Y. Z. Ding, J. G. Yu, X. X. Song, et al., “Experiment and applications of SOS-based pulsed power,” High Power Laser & Particle Beams, Vol. 17, No. 8, pp. 1195–1200, 2005.

[12] J. C. Su, G. Z. Liu, Y. Z. Ding, Z. J. Ding, S. Qiu, Z. M. Song, et al., “Nanosecond SOS-based pulse generator SPG200,” 3rd International Symposium on Pulsed Power & Plasma Application, Mianyang, China, pp. 258–261, 2002.

[13] W. L. Huang, G. S. Sun, P. Yan, J. Wang, and G. J. Li, “Flashover properties of polymethyl methacrylate and nylon in coaxial electric field under nanosecond pulse voltage,” High Power Laser & Particle Beams, Vol. 18, No. 7, pp. 1229–1232, 2006.

[14] W. L. Huang, G. S. Sun, P. Yan, and J. Q. Ren, “Simula-tion of brim-effects of coaxial electrode varying with the diameter of internal electrode,” Transactions of China Electrotechnical Society, Vol. 21, No. 4, pp. 117–121, 2006.

[15] S. J. Rzad, J. C. Devins, and R. J. Schwabe, “The influ-ence of a DC bias on streamers produced by step voltages in transformer oil and over solid/liquid interfaces,” IEEE Transactions on Electrical Insulation, Vol. 18, No. 1, pp. 1–10, 1983.

[16] L. Kebbabi and A. Beroual, “Influence of the properties of materials and the hydrostatic pressure on creepage dis-charge characteristics over solid/liquid interfaces,” 2003 Annual Report Conference on Electrical Insulation and Dielectric Phenomena, Albuquerque, USA, pp. 293–296, 2003.



1

Dust Effect on the Performance of Wind Turbine Airfoils

Nianxin Ren1,*, Jinping Ou1,2

1School of Civil Engineering, Harbin Institute of Technology, Harbin, 150090, China; 2School of Civil and Hydraulic Engineering, Dalian University of Technology, Dalian, 116024, China. Email: *[email protected]

Received January 9th, 2009; revised February 23rd, 2009; accepted March 3rd, 2009.

ABSTRACT

The full two-dimensional Navier-Stokes algorithm and the SST k- turbulence model were used to investigate incom-pressible viscous flow past the wind turbine two-dimensional airfoil under clean and roughness surface conditions. The NACA 63-430 airfoil is chosen to be the subject, which is widely used in wind turbine airfoil and generally located at mid-span of the blade with thickness to chord length ratio of about 0.3. The numerical simulation of the airfoil under clean surface condition has been done. As a result, the numerical results had a good consistency with the experimental data. The wind turbine blade surface dust accumulation according to the operational periods in natural environment has been taken into consideration. Then, the lift coefficients and the drag coefficients of NACA 63-430 airfoil have been computed under different roughness heights, different roughness areas and different roughness locations. The role that roughness plays in promoting premature transition to turbulence and flow separation has been verified by the numeri-cal results. The trends of the lift coefficients and the drag coefficients with the roughness height and roughness area increasing have been obtained. What’s more, the critical values of roughness height, roughness area, and roughness location have been proposed. Furthermore, the performance of the airfoil under different operational periods has been simulated, and an advice for the period of cleaning wind turbine blades is proposed. As a result, the numerical simula-tion method has been verified to be economically available for investigation of the dust effect on wind turbine airfoils.

Keywords: NACA 63-430 Airfois, Lift Coefficient, Drag Coefficient, Roughness Height, Roughness Area, Roughness Location

1. Introduction

As is known to all, the world energy crisis is more and more serious in our modern society. Wind energy, which is the most mature technology among so many kinds of clean and renewable energy, is developing with an amazing speed now. The crucial to rotor design is the subject of airfoils. One of the most critical problem for wind turbine rotors is degradation of the performance, and the unpredictability of stall due to dust accumulation on blade surface area. The objective of this work is to provide a better understanding for the effect of blade surface roughness on the performance of the wind turbine thick airfoil. It is very useful to enable wind turbine designers to predict loads and energy losses during wind turbines operating under dust conditions, it is necessary to qualitatively and quantitatively know the change in the aerodynamic properties due to the dust accumulation on the surface of the blade. So, the analysis of airfoil surface roughness has practical application interest in addition to academic interest. Now, the importance of the dust effect on the performance of the wind turbine airfoil is well realized.

Generally, roughness has a large effect on the flow dynamic processes. Therefore, the stall-regulation phe-nomena in wind turbines are affected by a high degree due to the increase of the blade surface roughness. De-spite some previous experimental and numerical work where surface roughness is involved, the information that has been obtained on this subject still remains far from complete [1-3], and the processes of boundary-layer separation and stall phenomena, which occur on the wind turbines blade in the presence of surface roughness, are not fully understood. To obtain more detailed information about the effect of the surface roughness on the lift coef-ficients and the drag coefficients of mid-span thick air-foils, the full two-dimensional Navier-Stokes algorithm and the SST k- turbulence model are used to investigate incompressible viscous flow past the wind turbine two- dimensional airfoil with surface roughness. The NACA 63-430 airfoil is chosen to be the subject, which is widely used in wind turbine airfoil and generally located at mid-span of the blade with thickness to chord length ratio


Dust Effect on the Performance of Wind Turbine Airfoils 103

of about 0.3. The lift coefficients and the drag coeffi-cients of NACA 63-430 airfoil have been computed un-der different roughness heights, roughness areas, and different roughness locations. The role that roughness plays in promoting premature transition to turbulence and flow separation has been verified by the numerical results. Furthermore, the degenerate trend of performance for NACA 63-430 airfoil under different operation periods has been simulated.

2. Numerical Modeling

2.1 Governing Equations

The flow past the airfoil was modeled by the full Na-vier-Stokes equation for two-dimensional, viscous, im-pressible flow. The continuous equation and Momentum equation based on Reynolds averaged N-S equations are as follows:

i

0x

iu

(1)

i j ' 'ii j

j i j j

( u u )( u ) up( u

t x x x xi u )

(2)

where i, j=1, 2; =1.255kg/m3; =1.789410-5 kg/(ms).

2.2 SST k- Turbulence Model

The shear-stress transport (SST) k- model was devel-oped by Menter [4] to effectively blend the robust and accurate formulation of the k- model in the near-wall region with the free-stream independence of the k- model in the far field. The SST k- model is described as follows:

i j j

( ) k( ) ( )

t x x xi k k kk

ku G Y S

k (3)

i j j

( )( ) ( )

t x x xiu G Y D S

(4)

where k represents the turbulence kinetic energy; represents the specific dissipation rate; Gk represents the generation of turbulence kinetic energy due to mean ve-locity gradients; G represents the generation of ; k and represent the effective diffusivity of k and , re-spectively; Yk and Y represent the dissipation of k and due to turbulence; D represents the cross-diffusion term, calculated as described below; Sk and S are user-defined source terms.

The SST k- model is similar to the standard k- model, but includes the following refinements:

·The standard k- model and the transformed k- model are both multiplied by a blending function and both models are added together. The blending func-tion is designed to be one in the near-wall region, which activates the standard k- model, and zero away from the surface, which activates the trans-formed k- model.

·The SST model incorporates a damped cross-diffusion derivative term in the equation.

·The definition of the turbulent viscosity is modified to account for the transport of the turbulent shear stress.

·The modeling constants are different. These features make the SST k- model more accurate

and reliable for a wider class of flows (e.g., adverse pressure gradient flows, airfoils, transonic shock waves) than the standard k- model. Other modifications include the addition of a cross-diffusion term in the equation and a blending function to ensure that the model equa-tions behave appropriately in both the near-wall and far-field zones.

2.3 Modeling

The whole computational zone consists of a semicircle with the radius of 10m and a rectangle with the length of 25m (Figure 1). The length of numerical airfoil which locates near the center of the semicircle is 1m. The height of the grid near the airfoil surface is 210-5m (Figure 2). Reynolds number is 1.6106 [5].

3. Numerical Results

3.1 Numerical Simulation under Clean Surface Condition

First of all, to ensure that the numerical model is avail-able for the free-stream flow past the airfoil, the numeri-cal simulation under clean airfoil surface condition was made to compare with the wind tunnel experimental data [5]. The simulative condition was according to the wind

Figure 1. The whole computational zone


104 Dust Effect on the Performance of Wind Turbine Airfoils

Figure 2. Local amplified grids for the airfoil

tunnel set-ups, for example, Reynolds number 1.6106; free stream turbulence level of 1%. The lift coefficients and the drag coefficients of the NACA 63-430 airfoil were computed under the angle of attack between 0 and 25 degree, based on the full two-dimensional Navier- Stokes algorithm and the SST k- turbulence model. The comparison between numerical results and experimental data shows in Figures 3-4.

In the above two figures, it was obvious that the nu-merical results had a good consistency with the experi-mental data, despite small discrepancy in comparison of drag coefficient curves. Therefore, the numerical model was confirmed to be available for the free-stream flow past the NACA 63-430 airfoil. Then, in the following of the paper, the numerical model would be used to investi-gate the airfoil surface roughness effect.

3.2 Numerical Simulation under Different Sur-face Roughness Heights

It is supposed that all the surface of the airfoil is rough-ness area and the angle of attack is 10.6 degree. That’s because at this angle of attack the numerical results have a perfect consistency with the experimental data. Then, numerical simulations of the NACA 63-430 airfoil under different surface roughness heights has been done and the results are shown in Figure 5.

Figure 3. Comparison of lift coefficient curves

NACA 63-430 Airfoil

0.00

0 5

Angl e)

Dra

Figure 4. Comparison of drag coefficient curves

Figure 5. Numerical results of Cl and Cd under different roughness heights

In Figure 5, it is evident that the lift coefficient curve decreased very rapidly when the roughness height is less than 0.3mm. However, the lift coefficient curve decreases very slowly when the roughness height is more than 0.3mm. The same trend of the change also happened in the drag coefficient curve, which increases very rapidly when the roughness height is less than 0.3mm, but in-creases very slowly when the roughness height is more than 0.3mm. That’s because the surface roughness plays a role in promoting premature transition to turbulence and flow separation. When the roughness height is less than 0.3mm, the effect of roughness is very obvious. But, when the roughness height is more than 0.3mm, the whole airfoil has already fully become turbulent bound-ary and at the same time the flow separation is very serious, so the effect of roughness on the performance of the airfoil seems no longer evident. Therefore, the roughness height of 0.3mm could be viewed as a roughness critical height.

Besides lift coefficients information and drag coeffi-cients information of the airfoil, the numerical results contain more information of the flow past the airfoil un-

NACA 63-430 Airfoil

0

0

1

0Ro

Cue

.0

.2

0

0

0

1l and Cd val

.4

.6

.8

.0

.2

.0 0.5 1.0ughness height(mm)

Lift coefficient

Drag coefficient

NACA 63-430 Airfoil

Lift coefficient Drag coefficient

Roughness height (mm)

0.0 0.5 1.0

C1

and

Cd

valu

e

1.2

1.0

0.8

0.6

0.4

0.2

0.0

NACA 63-430 Airfoil

0.0

0.20.4

0.6

0.8

1.0

0

Angl )

Lt

1.2

1.4

5 10 15 20 25

e of attack(degree

ift coefficien

Numerical results

Experimental data

0 5 10 15 20 25 Angle of attack(degree)

Numerical results

Experimental data

NACA 63-430 Airfoil

Lif

t coe

ffic

ient

1.4

1.2

0.8

0.6

0.4

0.2

0.0

0.05

0.10

0.15

0.20

0.25

5 10 15 20 2

e of attack(degre

g coefficient

Numerical results

Experimental data

NACA 63-430 Airfoil

Numerical results

Experimental data

0 5 10 15 20 25 Angle of attack (degree)

Dra

g co

effi

cien

t

0.25

0.20

0.15

0.10

0.05

0.00



der different roughness conditions, which are also very useful to clarify the physical mechanism of the surface roughness effect. For example, the turbulent intensity distribution and pressure coefficient distribution of the airfoil surface are shown in Figure 6 and Figure 7, re-spectively.

In Figure 6, it is clear that the turbulent intensity of the airfoil under clean condition is very small, however, with the roughness height increasing from 0 to 0.3mm, turbu-lent intensity increases very obviously, especially for the leading edge of the airfoil. In Figure 7, the area enclosed by the pressure coefficient curve of the airfoil obviously decreases with the roughness height increasing from 0 to 0.3mm, especially for the location at the front 50% of the chord length.

Furthermore, the two pictures could be better to dem-onstrate and clarify that surface roughness plays a role in promoting premature transition to turbulence and flow separation. By the way, it also could be found that the increase of turbulent intensity and the change of the pressure coefficient are no longer evident when the roughness height is more than 0.3mm.

Figure 6. Turbulent intensity curves under different rough-ness heights

Figure 7. Pressure coefficient curves under different rough-ness heights

3.3 Numerical Simulation under Different Roughness Areas

To clarify the effect of different roughness areas on per-formance of the airfoil, simulations under the angle of attack 10.6 degree and roughness height 0.3mm were done. The roughness area is denoted by the ratio of the chord length covered with roughness, which is calculated from the leading edge. The simulation results are shown in Figure 8. It is obvious that the lift coefficient curve decreases rapidly when the roughness area is less than 0.5. However, the lift coefficient curve decreases very slowly when the roughness height is more than 0.5. It can be concluded that the front 50% of the chord length more easily promotes premature transition to turbulence and flow separation. In the flowing section, further study and explanation will be given.

3.4 Numerical Simulation under Different Roughness Locations

Furthermore, to see the effect of different roughness lo-cations on performance of the airfoil more clearly, simu-lations under the angle of attack 10.6 degree and rough-ness height 0.3mm have been done. The whole NACA 63-430 airfoil is averagely divided into 10 parts, and each part is with a length of 0.1m. The simulation results are shown in Figure 9. The decrease percentage of the lift coefficient is defined by:

NACA 6

% 100%clean roughnessdecrease

clean

Cl ClCl

Cl

(5)

From the Figure 9, it could be seen that the influence of surface roughness located at the front 50% of the chord length is more obvious than that located at the back 50% of the chord length. On one hand, it is because the roughness located at the front 50% of the chord length more easily promotes premature transition to turbulence and flow separation. On the other hand, it is also because the airfoil at font 50% of the chord length is generally thicker than that at the back 50% of the chord length. Therefore, the roughness location at 25% of the chord length had the most obvious effect, where is the location of the largest thickness.

It is worth to be noticed that, when the roughness loca-tion is at the trailing edge 10% of the chord length, the lift coefficient under surface roughness condition is about 3% larger than that under clean surface condition. So, it could be concluded that properly increasing the surface roughness of trailing edge could be benefit to promote the airfoil’s lift coefficient to some degree, which has a good agreement with experimental data [6]. The conclusion would be useful for practical projects.

3-430 Airfoil clean

0

0.0

0.1

Turbulent intensity

5

0.1

5

0.2

0 0.2 0.4 0.6 0.8 1

x/c

0.1mm roughness0.3mm roughness1mm roughness

0 0.2 0.4 0.6 0.8 1x/c

NACA 63-430 Airfoil

Tur

bule

nt in

tens

ity 0.2

0.15

0.1

0.05

clean 0.1mm roughness0.3mm roughness1mm roughness

0

NACA 63-430 Airfoil

-2.5

0 0.2 0.4 0.6 0.8 1

-2

-

-

Pressure

coefficient

1.5

-1

0.5

0

0.5

1

1.5

x/c

clean0.1mm roughness0.3mm roughness1mm roughness

NACA 63-430 Airfoil 1.5

1

Pre

ssur

e co

effi

cien

t

0.50

-0.5-1

-1.5 clean 0.1mm roughness-20.3mm roughness1mm roughness -2.5

x/c


106 Dust Effect on the Performance of Wind Turbine Airfoils

Figure 8. Lift coefficient under different roughness areas

Figure 9. Lift coefficient under different roughness locations

3.5 Numerical Simulation Under Different Operational Periods

In this section, considering the effect of the blade surface dust accumulation, the performance of the wind turbine NACA 63-430 airfoil under different operational periods has been studied. The relationship between surface roughness and operational period is determined by field investigation, which was done by Khalfallah [2]. Then, the roughness heights and the roughness areas under different operational periods can be calculated by the following linearization function proposed by Khalfallah, which is not taken the rain washing effect into consid-eration. As a result, the specific four operation periods are list in Table 1.

0.08 0.02dD T (6)

/ 25aD T (7)

where, Dd is the dust size (diameter in mm); T is length of operational period in months; Da is the dust area in per-centage of the chord length.

Subsequently, the former available numerical model has been used again to investigate the performance of the

NACA 63-430 airfoil under the four operation periods and the numerical results are shown in Figure 10 and Figure 11.

It can be seen that, the dust effect on the performance of wind turbine airfoil is obvious when the operation pe-riod is 5months. The decrease of the lift coefficient and the increase of the drag coefficient are remarkable to a large degree, especially for the angle of attack around 13 degrees. That’s because the roughness height of the 5months operational period is 042mm, which is larger than the critical value 0.3mm. As the operation period increasing from 5months to 12.5months, the roughness area is increasing from 0.2 to 0.5, which just reaches the critical roughness area value (0.5). Therefore, the lift co-efficient further decreases to a large degree at the angle of attack less than 13 degrees. But, when the operational period is longer than 12.5 months, the change of the lift coefficient is rather small. That’s because that, when the operational period is longer than 12.5 months, the rough-ness height and roughness area both larger than the criti-cal values proposed in the previous section. As a result, the effect of the surface roughness on promoting prema-ture transition to turbulence and flow separation tends to be neglectable.

NACA 63-430

Table 1. Roughness heights and roughness areas according to different wind turbine operational periods

T(months) Dd (mm) Da

2.5 0.22 0.1

5 0.42 0.2

12.5 1.02 0.5

20 1.62 0.8

Figure 10. The comparison of lift coefficient curves during different operation periods

-1

x/c

0%

0%

10%

20%

30%

5% 15% 25% 35% 45% 55% 65% 75% 85% 95%

Cl decrease

percentage

NACA 3-430 6

5% 15% 25% 35% 45% 55% 65% 75% 85% 95% x/c

30%

20%

10%

0%

-10%

C1

decr

ease

pe

rcen

tage




·The performance of NACA 63-430 airfoil is more sen-sitive to roughness location at the front 50% of the chord length than roughness location at the back 50% of the chord length. In particular, proper roughness height at the trailing edge can be benefit to promote the lift coefficient to some degree. This conclusion is very useful for the wind turbine blade designer to further improve the performance of the blade.

·Considering the dust accumulation effect, the per-formance degenerate trend of NACA 63-430 airfoil under different operation periods has been studied. As a result, the period of 3 months without any rain is proposed as the proper period of cleaning the blade surface. As a result, the numerical simulation method has been

verified to be economically available for the investigation of the dust effect on the performance of the wind turbine airfoil.

Figure 11. The comparison of drag coefficient curves during different operation periods

In practical project, it is of great economic interest to keep the good performance of the wind turbine blade, so it is very meaningful to periodically clean the dust of the blade surface when there is no rain washing for a long time. Considering the performance degenerate trend of the wind turbine airfoil according to the operational pe-riod, the period of 3 months is proposed for the proper period of cleaning the blade surface.

5. Acknowledgments

The support of the National Science Foundation of China under project No. 50538020 and the National Science and Technology Planning under project No. 2006BAJ03B00 is gratefully acknowledged.

4. Conclusions REFERENCES

In this paper, the full two-dimensional Navier–Stokes algorithm and the SST k- turbulence model were used to investigate incompressible viscous flow past the wind turbine NACA 63-430 airfoil under clean and roughness surface conditions. The key findings can be summarized as follows:

[1] R. Van and W. A. Timmer, “Roughness sensitivity considerations for thick rotor blade airfoils,” in 41st Aerospace Sciences Meeting, Reno, USA, pp. 472-480, 2003.

[2] G. K. Mohammed and M. K. Aboelyazied, “Effect of dust on the performance of wind turbines,” Desalination, Vol. 209, No. 1-3, pp. 209-220, April 30, 2007.

·The full two-dimensional Navier-Stokes algorithm and the SST k- turbulence model have been verified to be available for predicting the performance of the wind turbine airfoil under clean and roughness surface condi-tions. The numerical results under clean surface condi-tion have a good consistency with the experimental data, despite small discrepancy in comparison of drag coeffi-cient curves.

[3] W. H. Wade and P. R. Alric, “Numerical prediction of unsteady vortex shedding for large leading-edge roughness,” Computers & Fluids, Vol. 33, No. 3, pp. 405- 434, March 2004.

[4] F. R. Menter, “Zonal two-equation model k–w models for aerodynamic flows,” in 24th Fluid Dynamics Conference, Orlando, Florida, 1993.

[5] P. Fuglsang, I. Antoniou, and S. D. Kristian, “Wind tunnel tests of the FFA-W3-241, FFA-W3-301 and NACA 63-430 airfoils,” http://www.risoe.dk/rispubl/VEA/veapdf/ ris-r-1041.pdf.

·The lift coefficient and the drag coefficient of NACA 63-430 airfoil is influenced more obviously by the roughness height less than 0.3mm than by the roughness height more than 0.3mm. In other words, the performance of airfoils is more sensitive to small roughness height. As a result, the roughness height of 0.3mm is proposed to be roughness critical height.

[6] N. S. Bao, F. E. Huo, and Z. Q. Ye, “Aerodynamic performance with roughness on wind turbine airfoil surface (in Chinese),” Acat Energiae Solars Sinca, Vol. 4, No. 8, pp. 458-462, 2005.



1

A Novel Half-Bridge Power Supply for High Speed Drilling Electrical Discharge Machining

He Huang, Jicheng Bai, Zesheng Lu, Yongfeng Guo

Department of Manufacturing and Automation Engineering, Harbin Institute of Technology, Harbin, Heilongjiang Province, China. Email: [email protected], [email protected], [email protected], [email protected]

Received February 10th, 2009; revised March 12th, 2009; accepted March 20th, 2009.

ABSTRACT

High Speed Drilling Electrical Discharge Machining (HSDEDM) uses controlled electric sparks to erode the metal in a work-piece. Through the years, HSDEDM process has widely been used in high speed drilling and in manufacturing large aspect ratio holes for hard-to-machine material. The power supplies of HSDEDM providing high power applica-tions can have different topologies. In this paper, a novel Pulsed-Width-Modulated (PWM) half-bridge HSDEDM power supply that achieves Zero-Voltage-Switching (ZVS) for switches and Zero-Current-Switching (ZCS) for the dis-charge gap has been developed. This power supply has excellent features that include minimal component count and inherent protection under short circuit conditions. This topology has an energy conservation feature and removes the need for output bulk capacitors and resistances. Energy used in the erosion process will be controlled by the switched IGBTs in the half-bridge network and be transferred to the gap between the tool and work-piece. The relative tool wear and machining speed of our proposed topology have been compared with that of a normal power supply with current limiting resistances.

Keywords: High Speed Drilling Electrical Discharge Machining, Half-Bridge Power Supply, Zero Current Switching, Zero Voltage Switching

1. Introduction

Electrical Discharge Machining (EDM), also known as spark erosion machining, is becoming increasingly popular. EDM sees the removal of matter from high hardness con-ductive materials by means of a series of repeated electri-cal discharges between electrode and work-piece, which are separated by a discharge gap. Dielectric fluid is forced into the discharge gap where electrical discharge erosion occurs. When a voltage large enough is applied, the dielec-tric fluid breaks down, the gap is ionized and electrons are emitted from the tool (cathode). When more electrons gather in the gap, the resistance drops, which causes elec-tric spark to jump between the work-piece surface and the tool. The whole sequence of operation occurs within a few microseconds and is accompanied by a shock wave in the dielectric. The impact of the wave on the electrode causes high transient pressure. The current density in the dis-charge channel is of the order 10,000 A/cm2. The tem-perature of the central point of the channel is of the order of tens of thousands of °C. The forces of the electric and magnetic fields caused by the spark produce a tensile force and tear-off particles of molten and softened metal from this spot on the work-piece [1,2].

Within the scope of EDM, High Speed Drilling Elec-trical Discharge Machining (HSDEDM) is an important

technology. The basic components of a HSDEDM system are as follows and the schematic is shown in Figure 1.

Like other EDM methods, the HSDEDM process is especially suitable for machining high strength and hard metal materials. Generally, a rotating thin copper or brass tube electrode is used as the drilling tool. High speed and pressure dielectric (water) is pumped through this hollow

Figure 1. Diagram of HSDEDM process

A Novel Half-Bridge Power Supply for High Speed Drilling Electrical Discharge Machining 109


electrode and injected into the discharge gap to flush out eroded debris from working area to ensure the process remains stable. The dielectric fluid also acts as a coolant. Rotation of the electrode helps reduce deviation and un-even wear, which maybe otherwise be caused by the high pressure dielectric, on the face end of the electrode. The servo system acts to adjust the discharge gap and feed throughout the process. The most outstanding feature of HSDEDM is high speed of manufacture which can be up to 60mm/min. This is hundreds of times the speed of the normal EDM and is exceeding the level of the traditional drilling. The processed apertures range from 0.3mm to 3mm and the aspect ratio can be up to 100 [3].

In HSDEDM, pulse power supply plays an important role in providing the thermal action of the electrical dis-charges between the electrode tool and the work-piece in order to achieve material removal. The Metal Removal Rate (MRR), the surface finish and Electrode Wear (EW) mainly depend on the magnitude and duration of the pulse discharge [4]. As the current increases, so do the MMR and the EW, although the surface finish decreases. As the discharge frequency is increased, the surface fin-ish improves but the EW increases.

In order to improve the whole efficiency of the elec-tronic equipment, the search for a more efficient switching technique has been developed. For a switch such as Insu-lated Gate Bipolar Transistor (IGBT), the zero current switching (ZCS) technology has been frequently used, because this technology can reduce the current rapidly when IGBT is turned off, therefore reducing the energy loss caused by the tail current [5,6]. It is also important, in miniaturizing power supply, to balance switching fre-quency and conversion efficiency. Zero Voltage Switching (ZVS) is a powerful means for increasing the switching frequency with small switching loss. On the other hand, switching noise is a critical problem at high frequency switching. ZVS is also effective in decreasing the spike current and surge voltage when switches are turned off [6].

This research has produced a pulse power supply that is used in HSDEDM system without an output capacitor for charging and discharging spark energy and bulky re-sistor for limiting current. A Pulsed Width Modulated (PWM) half-bridge topology with ZCS and ZVS is de-veloped in this power supply for suppressing the noise current in discharge gap and surge voltage of switches.

2. Developments in Power Configurations

The shapes of voltage and current pulses in the discharge gap depend on the chosen power supply. There are three types of power supplies that have received most interest amongst the scientific community. They are Resistance Capacitance (RC) power supply (A), transistor switching circuit (B) and energy-saving power supply (C).

2.1 Resistance-Capacitance (RC) Power Supply

EDM was developed in Russia in the mid-1940-s by

Lazalenko and he used the basic RC power supply which is still used today in many cases when a fine surface fin-ish is required [7]. The common configuration of these systems is shown in Figure 2.

The AC input voltage is normally fed into a variac and isolation transformer and is rectified to produce an uncon-trolled DC. The bulk capacitor C1 is used to filter the rec-tified output and the output capacitor C2, which stores the energy needed for the spark, is connected in parallel to the discharge gap. Every time dielectric breaks down, the energy in C2 will be discharged and in the next cycle it will be charged again. The current is limited by the resis-tor, R1, which is normally a high power resistor. R1 is where most of the losses of an EDM power supply occur.

2.2 Transistor Switching Circuit Power Supply

Figure 3 shows a different technique of implementing the spark generator whilst utilizing the linear power supply. The transistor is Pulse Width Modulated (PWM) to control the total period of the charging time of the capacitor. This circuitry provides a higher MMR than the normal rectan-gular pulse power supply. In current research, more and more linear converters are being used to switch the electric energy. These improvements increase efficiency, power density and decrease the size of the magnetic components [1]. However this type of configuration still uses bulky resistors. The problem of high losses still remains.

2.3 Energy-Saving Power Supply

The power supply with the feature of energy-saving was developed in the end-1980s. Dr. Zhao Wansheng, with his group, developed a pulse generator for EDM with the feature. The schematic is shown in Figure 4. They de-signed the circuitry as a forward converter without the output capacitor. Current limiting is achieved by sensing the load current and feeding it back to the control circuit. Once the current reaches a present current limit, the con-troller shuts down the drive and starts again at next cycle. The forward converter topology is designed for use in low power (50W) to medium power (250W) applications [8]. The second winding of flyback transformer, T1, is used in the discharge circuit and acts as clamp winding. However this high inductive reactance will induce volt-age swing and current tail in the discharge gap.

It can be seen from the developments in Power supply, the switch mode configuration and improvement of effi-ciency are research focuses in EDM.

Figure 2. Traditional EDM power supply

110 A Novel Half-Bridge Power Supply for High Speed Drilling Electrical Discharge Machining


3.1 Configuration

Two sets of IGBTs (TL1 to TLn and TR1 to TRn) are used as switches and connected with each other through the discharge side inductance L, rd, discharge wires and discharge gap. L consists of the inevitable lead induc-tance, self-inductance in the discharge gap and any other stray inductances in the discharge path. The resistance of the discharge wires and the spark gap during discharge constitute rd. D1 and D2 are fly-wheel diodes. D3, D4, R3, R4, C1 and C2 constitute two sets of RCD snubber circuits which are used to protect the switches from volt-age surge caused by the sudden turning off. Fast recovery diodes DL1 to DLn and DR1 to DRn are anti-parallel diodes in the switches which are also used to protect the IGBTs. In this way, ZVS of the switch is achieved even with high switching frequency.

Figure 3. Transistorized RC power circuit

3. Half-Bridge Network Power Supply

The proposed topology is a PWM half-bridge converter, which uses IGBTs a as switch to control the energy. The configuration is shown in Figure 5.

3.2 ZVS of Switches

There are two sets of RCD snubber circuits parallel con-

Figure 4. Energy-saving EDM power supply developed by Harbin Institute of Technology

Figure 5. Circuit diagram of the HSDEDM power generator using a half-bridge network



nected with switches as shown in Figure 5. The snubber circuits achieve ZVS for switches TLn and TRn. If with-out any protection measures for switch, when switch is turned off, the prime current ip decreases linearly, while the collector-emitter voltage (Vce) of the switch increases rapidly, and the surge voltage with extremely high value will be generated. In verify experiments, the surge volt-age usually reaches more than six times of the supply voltage as shown in Figure 6(a). This surge voltage with high value is harm to switches. In this proposed topology, when switch is turned off, surge energy is stored in cap C1 or C2 through the fast recovery diodes D3 or D4. In the next time, when the switch is turned on, C1 or C2 dis-charge though the resistances R3 or R4. The measured waveforms of Vce and prime current of switch with ZVS can be seen in Figure 6(b).

3.3 ZCS of Gap and Switching Methodology

The traditional power supplies of EDM or HSDEDM always use bulky resistors to limit and control the current in the gap. The electrical energy surges also can be con-sumed by these resistors. For energy-saving, there are not current limit resistors in the proposed topology. If no action being taken to limit the power, the current will rise to a very high level to break the switches, especially when short circuit occurs. When the switch is off, the IGBTs also would be damaged by electrical surges caused by the sudden changes in electrical flow. In this PWM half-bridge network, when the drive IGBT is off, the fly-wheel diodes are used to smooth out the electrical surges caused by the sudden changes in electrical flow, therefore the ZCS can be obtained by this topology. This facilitates dielectric deionization quickly, therefore, short circuit and electric arc discharging are reduced and higher utilization of the pulse can be gained.

For this half-bridge network, the following PWM op-erational sequence is taken to generate the proper electri-cal pulses for the security of the switching IGBTs. The current in the discharge is shown in Figure 7.

The left and the right IGBTs are all switched on in time t0. The power streams from source through the left IGBT, discharge wire, gap and the right IGBT to ground. The current in the gap rises rapidly to I1 which is safe for the switches. The left or the right IGBT is switched off in time t1.

If the left IGBT is off, the energy stored in L streams through the circuit which consists of D1, wire, gap and the right IGBT to the ground. If the right IGBT is off, the energy streams though another circuit which consists the left IGBT, wire, gap and D2 and goes back to the source. In that time, the current decreases to I2. The step 1 and 2 are repeated several times. The

on-time is t2 and the off-time is still t1. The current os-cillates between I1 and I2. While the IGBT in left or right side is switched on and off repeatedly, the IGBT in the other side always is turned on.

Figure 6. Waveforms of prime current in gap and Vce of switches

Figure 7. Theoretical current waveform in the discharge gap

The left and the right IGBTs are all switched off. The

energy stored in the wire inductance streams though flay-wheel diodes, D1 and D2 back to source. The current decreases rapidly to zero.

In Figure 7, a trapezoidal current waveform with saw-tooth form on the top can be seen. Compared with normal rectangular current waveform, the average current peaks is (I1+I2)/2 and the width of the pulse is almost t0+n(t1+t2)+t3, n is the repetition of switching on and off of IGBT just in left or right side. The peak current is de-pended on the simultaneous on-time (t0) of switches in

112 A Novel Half-Bridge Power Supply for High Speed Drilling Electrical Discharge Machining

both sides of the bridge in the beginning of the pulse. If just one IGBT is switched on in both sides in that time then t0 can not be set too large in order to avoid exceeding the rated current level of the IGBT. Larger peak current can be achieved by turning on more switches in both sides at the beginning of the pulse to help share the current. The width of the pulse is depended on number of the repetitions of switching. If only one IGBT is switched repeatedly in the middle of the pulse, the frequency is high and close to the rated frequency of the IGBT. This will lead to the in-creased transient thermal impedance of the IGBT, thus more heat generated by the current will damage the switch. In the proposed topology, several IGBTs are combined in parallel connection with collectors and emitters on each side. These IGBTs are switched by turns to decrease the working frequency of every one.

Figure 8. Experimental gap voltage and current waveforms

(a) (b)

(c) (d)

(e) (f)

Figure 9. Relative tool wear and machining speed respond to 3 electrical parameters




Figure 8 shows gap voltage and current waveforms of

normal discharges. During the on-time the output current is around 25 A and the output maintaining voltage is around 20 V.

It can be seen from Figure 8 that the power supply proposed in this research has good performance in con-trolling the energy in the discharge gap, even without current limiting resistor. The process features of the pro-posed power supply will be demonstrated by comparing with the normal power supply which using bulky resistor to limit current.


Two types of HSDEDM power supplies were utilized to get relative tool wear and machining speed data. The first type is a normal power supply with current limiting re-sistances and the second type is the half-bridge network power supply without the resistances as proposed in this paper. Both types of the power supplies are characterized based on Relative Tool Wear (RTW) and EW data using three electrical parameters: gap peak current, pulse width and duty cycle. The comparison is shown in Figure 8. A brass tube electrode with 1 mm in external diameter and 0.3 mm in inner diameter was used in the experiments. The work-piece material was 40Gr steel and water was used as dielectric. The open voltage is 80 V.

As shown in Figure 9, the proposed power supply achieves a higher machining speed in most situations, especially at larger electrical parameters. However, the relative tool wear achieved by using the proposed power supply is little larger than the normal one. In theory, the proposed power supply characters the improvement in electric energy efficiency, because the current-limiting resistances are removed. Detailed analyses of tool wear, improvement of efficiency and the methods to decrease electrode wear are future tasks.

5. Conclusions

A power supply for HSDEDM based on a half-bridge network without bulky current-limiting resistances in the discharge circuit was developed resulting in a great re-duction in weight and size. A PWM operational sequence of switches was utilized to generate trapezoidal pulse waveform with saw-tooth form in the discharge gap. The electrical parameters of peak current, pulse width and duty cycle were achieved by control the switching time of these IGBTs in the circuit, This was also used to limit the current in the discharge circuit to avoid the power being too large to break the IGBTs. ZVS of switches was achieved to decrease the surge voltage when IGBTs were turned off. The “fly-wheel” diodes were developed to smooth the electrical surge and the ZCS technology was used to reduce the current rapidly when switches were turned off, therefore reducing the energy loss caused by the tail current. In comparative tests, it was shown that

the normal power supply with current-limiting resistances has good performance when considering relative tool wear and machining speed at light loading conditions. However the machining speed of the normal one cannot compare to the performance of half-bridge network power supply at lager maximum poser outputs. This de-velopment paves the way for alternative in designing HSDEDM power supplies for high power applications.

6. Acknowledgments

This work was funded by National Science Foundation of China with the grant number: 50875064 and supposed in the frame of 863 research project of China with the grant number: 2007AA04Z345 and supposed in the frame of Heilongjiang province important science and technology fund with grant number: GA06A501. The authors would like to thank Y. L. Liu for help in data acquisition, G. Q. Deng and C. J. Li for preparing the devices, and A. Hird for invaluable advice concerning English writing.

REFERENCES

[1] C. M. F. Odulio, L. G. Sison, and M. T. Escoto Jr, “En-ergy-saving flyback converter for EDM applications,” IEEE Region 10 Annual International Conference Pro-ceedings, pp. 1-6, 2007.

[2] N. L. Kachhara and K. S. Shah, “The electric discharge machining process.” Journal of the Institution of Engi-neers (India) -Mechanical Engineering Division, No. 51, pp. 67-73, 1971.

[3] M. R. Cao, S. C. Yang, S. Q. Yang and Y. T. Qiao, “Ex-perimental research on the process influencing machining velocity to small hole’s EDM,” Modem Manufacture En-gineering (in Chinese), Vol. 4, pp. 82-83, 2005.

[4] R. Casanueva, M. Ochoa, F. J. Azcondo, and S. Bracho, “Current mode controlled LCC resonant converter for electrical discharge machining applications,” Proceedings of the 2000 IEEE International Symposium on Industrial Electronics, Vol. 2, pp. 505-510, 2000.

[5] G. Hua and F. C. Lee, “Soft-switching techniques in PWM converters,” IEEE Transactions on Industrial Elec-tronics, Vol. 42, pp. 595-603, 1995.

[6] K. Yoshida and T. Ninomiya, “A novel current resonant ZVS-PWM half-bridge converter,” Electronics & Com-munications in Japan, Part I: Communications, Vol. 82, pp. 45-55, 1999.

[7] B. Sen, N. Kiyawat, P. K. Singh, S. Mitra, J. H. Ye, and P. Purkait, “Developments in electric power supply con-figurations for electrical-discharge-machining (EDM),” The 5th International Conference on Power Electronics and Drive Systems, Vol. 1, pp. 659-664, 2003.

[8] B. Y. Song, W. S. Zhao, and G. L. Shao, “Current type of electrical discharge machining pulse generator,” China Mechanical Engineering (in Chinese), Vol. 12, pp. 386- 390, 2001.



1

Air Compressor Control System for Energy Saving in Locomotive Service Plant

Wenyu Mo

Department of Control Science of Engineering, Huazhong University of Science and Technology, Wuhan, China. Email: [email protected]

Received January 2nd, 2009; revised March 12th, 2009; accepted March 20th, 2009.

ABSTRACT

The actuality and disadvantages of traditional high power asynchronism motor drive air compressor in locomotive ser-vice plant are discussed. In order to reduce the energy consumption and obtain safe running, a variable frequency con-trol method to the motor is supplied. A PLC with touch screen is used for monitoring the status of the compressor and its control system. It also presents energy consumption analysis caused by the variable frequency control method in a locomotive service plant.

Keywords: Variable Frequency, Air Compressor, Locomotive Service Plant

1. Introduction

Air compressor is a key equipment to provide air power, which is driven by asynchronism motor. Air compressor is operated by adjusting rotation speed of the motor. In general, the output air pressure in pipeline of the air tank is acted as controlling object in the system.

The air pressure in the pipeline is controlled to be fluctuate in a certain range. Its upper limit (0.7MPa) is under the rating pressure of the pipeline, and its lower limit (0.4MPa) is above the rating pressure of the equip-ment using the compressive air. Usually, there are two ways to control the air pressure in the pipeline to content this demand.

The first method is starting up and stopping the motor continually for adjusting air pressure in pipeline. When the air pressure is under the upper limit, the air compres-sor operates until the air pressure goes up to the upper limit, then the motor stops running. But the air pressure will be lowered with air leak or air consuming equipment operating. When the air pressure goes down to the lower limit, the motor begins to work and air compressor oper-ates again. The variable air pressure in the pipeline is shown in Figure 1. This method is simple and low cost, but it is suitable for small power motor because the motor will be started up continually.

The second method is using pressure valve to limit air pressure in the pipeline. When the pressure goes up to the upper limit pressure, the valve will close entrance of the air compressor, then the compressor is in idle state. In this case, the compressor is still driven in operation by the motor, but it does not export compressive air, so the

air pressure will not go up further. When the pressure goes down to the lower limit, the valve will be open again, and then the compressor exports compressive air and the air pressure will be up again. The variable air pressure is same as that of the first method. In this case, motor is running continuously, which can be used for high power motor.

In locomotive serve plant, the second method is widely used for control the air pressure from air compressor be-cause of high power motor being used, which rating power is about 100KW. Although the motor is in running operation, its starting up should be controlled. Tradition-ally, there are two ways to fulfill the starting up, which are linking series resistor in the rotor loop and converting Y- connection of the starter loop [1]. However, there are still some disadvantages in these two ways as follows.

1) The air pressure in pipeline fluctuates greatly be-tween the upper and lower limits.

2) The continual upload and download of air compres-sor causes voltage fluctuation in electrical power supply.

Figure 1. Variable air pressure in pipeline

Air Compressor Control System for Energy Saving in Locomotive Service Plant 115


3) The air compressor is in full speed rotation all the time, which may lead to mechanical failure and tempera-ture rise.

4) The air compressor and the air pressure valve in their operations cause a great noise pollution.

5) The driving motor of air compressor is inefficient and energy consumptive and cause low power factor.

So, it is necessary to change the traditional control method of the air compressor operation for energy saving, less pollution and low failure possibility.

2. Variable Frequency Control of Air Compressor Motor

2.1 System Principle

With the rapid development of power electronics tech-nology, transducer is widely used in adjusting speed of AC asynchrony motor [2]. So, a transducer is designed as an executor in the air compressor control system to adjust output air pressure.

In order to control the air pressure in the pipeline, the motor’s rotation speed should be in control. However, the motor disables to change its rotation speed itself, the only method is to adjust its frequency to change the rotation speed. So, a transducer used to control the motor’s rotation speed, then, the flux from air compressor can be adjusted. The system schematic diagram is shown in Figure 2.

After comparing enactment pressure signal with feed-back pressure signal, a pressure signal error is obtained, which is used for calculation of PID arithmetic and then converted the error signal to a control value to adjust fre-quency of AC power supply. Then the asynchronies mo-tor will drive air compressor to an appropriate rotation speed with variable frequency power supply to eliminate the pressure error and ensure a constant air pressure.

The air pressure adjusted by PID arithmetic in the pipeline is shown in Figure 3. In general, a transducer has an inner PID control unit.

Figure 2. System schematic diagram

( =Enactment pressure – Feedback pressure)

Figure 3. Air Pressure adjusted by PID arithmetic

0: negative effects, when > 0, transducer output fre-

quency raises; < 0, frequency declines. 1: positive effects, when > 0, transducer output fre-

quency declines; , frequency raises. When the pressure detected by pressure sensor is

higher than the enactment pressure, PID regulator output signal declines and the transducer output frequency falls down, then the air compressor rotation speed reduces and the output air pressure declines; when the detected pres-sure is lower than the enactment pressure, PID regulator output signal raises and the transducer output frequency increases, then the air compressor rotation speed in-creases and the output air pressure increases. The system controls the air pressure automatically through the above method.

In Figure 4, a transducer of YASKAMA 616PC5 forms the air pressure feedback control system. On the transducer, FS, FV, FI and FC are ports of pressure en-actment and feedback input signal. The FS provides power supply (+15V). An input voltage, which deter-mines frequency of AC power supply from the transducer, is linked to the FV port from a resistor (4.7K). A feed-back voltage of air pressure in the air tank detected by a remote pressure gauge is linked to FI port. These two signals are compared in the transducer and an error can be calculated, by which PID arithmetic is used for calcu-lating control variable. The transducer has self-educated ability, i.e. PID parameters can be adjusted automatically in terms of actual pressure change characteristic in the transducer.

S1, S3, S4 and SC are ports for several control func-tions. When K1 is closed, the transducer operates nor-mally. If there is a failure outside the transducer, K2 will be closed, and the transducer will stop operation. In this

116 Air Compressor Control System for Energy Saving in Locomotive Service Plant

Figure 4. Connection circuit of the transducer

case, a signal can be detected in S3 port to protect the whole system. However, if there is a failure in the trans-ducer, MA and MC ports will be connected in it, by which control system (PLC) can stop the whole system operation. When the failure is removed, control system makes K3 close, a signal inputs to the S4 port, which can reset the transducer. M1 and M2 ports output running state signal. Three-phrase AC power supply are linked to R, S, and T ports on the transducer. Then, the variable frequency AC power supply can be exported from U, V, and W ports.

A PLC acts as a control unit in the system, from which switches K1, K2 and K3 can output relative signals. Sig-nals from ports M1, M2, MA and MC on the transducer can also be input to the PLC.

2.2 Existing Problems and Solutions

Using the transducer would generate harmonic wave. The external input industrial frequency power supply AC (380V/50Hz) is rectified into DC by three-phrase bridge. Finally, it is inverted into any frequency AC power sup-ply by high power transistors after capacitor filtering. In the rectifier circuit, the input current waveform is irregu-lar rectangular wave. The wave is classified into basic wave and harmonic wave based on the Fourier series. The high-order harmonic would interfere with the power supply system, and damages transforms, motors, capaci-tors, switches and so on.

Solutions for the input side of transducer: 1) setting reactance to increase rectifier impedance for improving the rectifier overlap angle; 2) parallel using AC filter in the power circuit to separate high-order harmonic at all levels respectively from the power supply system.

Solutions for the output side of transducer: using high frequency switch components, adding filter equipments and adopting closed-loop control, using isolation, shield-ing, grounding and reasonable routing to improve the high-order harmonic interference.

Besides, the starting torque required when the air com-pressor starts up is large. However, using the conven-tional method to control will bring damage to other rele-vant equipments. This system adopts vector control tech-nical to raise the starting torque of motor. Besides, more accurate rotation speed can be gained to control the air compressor. So, almost constant pressure of air can be acquired from this system.

3. PLC Monitor and Control

3.1 Status Parameters Monitored

Cooling water and lubricating oil are the necessary sub-stances for air compressor running normally. When the air compressor operates, temperature rise of the com-pressor body can be used to monitor if the compressor runs normally, or not. So, there are three parameters to be monitored in the system, which are pressure of cooling water, pressure of lubricating oil and temperature of the compressor body.

The pressure of cooling water can be used to show cooling system normal operation easily, which includes pump, pipeline and valves. Any problem can cause the pressure abnormal. So, a pressure sensor is mounted at output of cooling water pump for monitoring the whole cooling water system. In the same way, another pressure sensor is used for the lubricating oil monitoring.

As the temperature of compressor body is one of syn-thetic images to show all malfunctions. In order to meas-ure temperature of the compressor body, a temperature sensor is mounted at the output pipeline of cooling water from the compressor. The temperature of cooling water from the output pipeline is almost equal to the compres-sor body temperature.

All these sensors are mounted on pipeline outside the compressor, so it is convenient for maintenance and re-paired.

3.2 Design of PLC Monitor and Control

In order to monitor and control the whole air compressor system, a PLC with touch screen is used [3]. The PLC is composed of power module, CPU module, analog input module, digital input module and digital output module. The configuration of PLC system is shown in Figure 5.

A touch screen is connected to the CPU module, which can display status of the transducer, pressure in the air tank, pressures of cooling water and lubricating oil, and so on. It also can accept many touch instructions instead of mechanical buttons. The CPU module can receive all kinds of data from analog and digital module by interior bus. It also has memory for program and data, and a se-rial bus RS232C for connecting with a computer.

The power module provides power supplies for every module, which includes +5V and +12V voltages.


Air Compressor Control System for Energy Saving in Locomotive Service Plant 117


Table 1. Testing parameters 75KW motor

Items Voltage (V)

Current (A)

Fre-quency

(HZ)

Rotate Speed (r/m)

Input Power (KW)

AC Power Supply

380 117 50 1450 69.5

Trans-ducer Power Supply

320 80 42 1210 39.8

300 working days, the air compressor control system

will save energy in one year as follows. Obviously, the new system can save lots of energy.

Besides, it has other advantages: 1) It reduces greatly noise, realizes soft start and soft

stop of the equipment and avoids the shock current of power grid when the air compressor starts up.

Figure 5. Configuration of the PLC The analog input module (0~+5V input) accepts volt-

age signals from the sensors to measure the compressed air pressure, cooling water pressure, lubricating oil pres-sure and temperature of the cooling water.

2) It has high degree of automation and overcomes the disadvantages of manual adjustment.

5. Conclusions The digital input module (12V input) accepts status signal of the transducer from M1 and M2, failure signal of the transducer from MA and MC, and some operation signals from outside buttons.

High power air compressor is a kind of equipment widely used in locomotive service plant. In order to save electri-cal energy and improve operation condition, it is neces-sary to rebuild traditional control system of high power motor with the transducer and PLC system. In fact the system also has low cost, high reliability and efficiency. It also reduces greatly noise of air compressor operation and possibility of failure. Finally, this system has solved the interference by the high-order harmonic and adopted vector control technique to acquire high starting torque and stabilize motor rotation speed. As the effect of this system is obviously in energy saving, it should be widely used in such a place as locomotive service plant.

The digital output module (relay output) sends out sev-eral control signals, which are K1, K2 and K3, operation status signal, and failure signal.


We use a PLC and a transducer to rebuild one air com-pressor control system in a locomotive service plant, which use a high power motor of 75KW rating power. Its other rating parameters are frequency of 50Hz, voltage of 380VAC and current of 150A. We have measured a set of actual parameters of the motor and its power supply at the same regular load condition before and after the re-building, which are listed in Table 1. On the majority of time, the motor can run normally at 42Hz frequency of power supply to content air supply requirement of the whole plant.

REFERENCES

[1] X. Z. Deng, “Control of electrical and mechanical driv-ing,” Wuhan: Publishing Company of Huazhong Univer-sity of Science and Technology, pp. 153–178, 2002.

From the Table 1, it can be shown that current decrease rate is

[2] O. Ojo, Z. Q. Wu, G. Dong, and S. Asuri, “Variable fre-quency control of an induction motor drive with reduced switching devices,” IEEE International Conference on Electric Machines and Drives, San Antonio, TX, United States, pp. 1385–1391, May 2005.

δ = ( I1 – I2 ) / I1 = (117 – 80) / 117 = 31.6%

Electricity energy saving rate is [3] B. Georges and J. Aubin, “Application of PLC for on-line

monitoring of power transformers,” IEEE Power Engi-neering Society Winter Meeting, Columbus, OH, United States, Vol. 2, pp. 483–486, February 2001.

η = ( P1 – P2 ) / P1

= (69.5 – 39.8) / 69.5 = 42.7%

As everyday has three shift working time and one year has [4] A. D Kurtz, et al., “High accuracy miniature pressure transducer,” International Instrumentation Symposium, Vol. 470, pp. 303–318, 2007.

W = ( P1 – P2 ) × 24 × 300 = (69.5 – 39.8) ×24 × 300 = 2.14×105 (KWh)



1

ICI Performance Analysis for All Phase OFDM Systems

Renhua Ge1, Shanlin Sun1,2

1Guilin College of Aerospace Technology, Guilin, 541004, China; 2Tianjin University, Tianjin, 300072, China. Email: [email protected]

Received January 26th, 2009; revised March 2nd, 2009; accepted March 10th, 2009.

ABSTRACT

Orthogonal frequency division multiplexing (OFDM) is a strong candidate for the next generation wireless communica-tion. But the frequency offset between the local oscillators at the transmitter and receiver causes a single frequency offset in the signal, while a time-varying channel can cause a spread of frequency offsets known as the Doppler spread. Frequency offsets ruin the orthogonal of OFDM sub-carriers and cause inter-carrier interference (ICI), therefore, quickly diminishing the performance of the system. A novel all phase OFDM (AP-OFDM) system is established. APFFT is introduced for the first time to overcome ICI aroused by carrier frequency offset (CFO) in OFDM systems. This scheme makes use of APFFT in time domain and zero inserting in frequency domain to reduce the amount of ICI gener-ated as a result of frequency offset, with little additional computational complexity. At the same time, the proposed sys-tem has zero phase error. It is proved to be correct and effective in mathematics. The simulation results indicate that AP-OFDM system has a better performance than conventional OFDM system.

Keywords: All Phase Orthogonal Frequency-division Multiplexing (APOFDM), Carrier Frequency Offset (CFO), In-ter-carrier Interference (ICI)

1. Introduction

Orthogonal frequency-division multiplexing (OFDM) communication systems require precise frequency syn-chronization [1-4], since otherwise inter-carrier interfer-ence (ICI) will occur. Currently, three different ap-proaches for reducing ICI have been developed in the literature. One is to estimate and remove the frequency offset [3-7]. While many methods exist that can estimate and remove the frequency offset quite accurately, they often have considerable computational complexity. An-other approach is to use signal processing and/or coding to reduce the sensitivity of the OFDM system to the fre-quency offset [8]. These methods can either be used as low complexity alternatives to fine frequency-offset es-timation techniques or they can be used together with a somewhat accurate oscillator. Windowing has been used in [9] and [10] to reduce the ICI created as a result of frequency offset. A simple and effective method known as ICI self-cancellation scheme [11,12] has been pro-posed by Zhao and Haggman. Other frequency-domain coding methods have been proposed in [13,14] that do not reduce the data rate. However, these methods produce less reduction in ICI. A new self-cancellation scheme has been proposed by Alireza, Seyedi and Saulnier [15]. This method has better performance with more computational complexity.

This paper concentrates on the further development of the ICI cancellation method and APFFT is proposed for ICI cancellation. The advantages of APFFT are zero phase error and less ICI. It improves the OFDM system's ability to resist frequency offset with the cost of limited additional computational complexity.

2. The Model of APOFDM System

In conventional FFT system, the spectrum of the receiver signal can expressed as [16].

00 /)()1(

/sin

sin1

jNkNj ee

Nk

k

NkX

(1)

where is the normalization frequency offset. Thus, the amplitude spectrum is:

Nk

k

NkX

/sin

sin1)(

(2)

And the phase spectrum is:

01 )(1

k

N

N (3)

Frequency offset can introduce ICI and phase noise and diminish the performance of the system Figure 1

ICI Performance Analysis for All Phase OFDM Systems 119


shows the block of APOFDM. Figure 2 is the N-all phase FFT circuit [16,17].

Assuming that X = X(k), k = 0, 1, …, N-1 is the complex digital sequence in frequency domain and

N

knjN

Nk

ekXN

kXIFFTnx212/

12/

1

. After zero

interval interpolation, we obtain sequence , 0, 1, 2, ..., 2 1X X k k N

. and

, 0, 2, , 2( 12

0, 1, 3, 5, , 2 1

kX k N

X k

k N

)

.

The number of the point for the new sequence is 2N, so 2N-IFFT should be implemented.

12

02)(

2

1)(

N

k

knNWkX

Nnx

1

0

22)2(

2

1 N

m

mnNWmX

N, n=0,1,…,2N-1 (4)

12

0

)(222

1 N

k

NnkNWkX

NNnx

NN

N

m

mnN WWmX

N2

2

1

0

222

2

1

nxWmXN

N

m

mnN

1

0

222

2

1, n=0,1,…,2N-1

(5)

The purpose of zero interval interpolation in frequency is to obtain the copy of N subcarriers of OFDM systems and keep the number of subcarriers unchangeable. Therefore, the former N points are the copy of the last N points in time domain. These 2N points are indispensable to APFFT. That is to say, in receiver, we only need two times clock sample compared with FFT system to get APFFT sequence in time domain.

To digital sequence with 2N points, the first point x(0) is removed and the remaining sequence can be expressed as ).12(,)2(),1(),(),1(,)2(),1( NxNxNxNxNxxx In this digital sequence, we can obtain N sub-sequences and each of them has N points including x(N). Then, every sub-sequence with a recycling style is moved. Fi-nally, the point x(N) is moved to the first place and then

Figure 1. The block of APOFDM system

Figure 2. The block of N-step APFFT

120 ICI Performance Analysis for All Phase OFDM Systems

get the others N sun-sequences with N points. Making them aligned and adding them, a new N points sequence is obtained, which is called all-phase digital sequence.

For instance, N=4 and 2N-1=7, we can get the se-

quence . This sequence

can be divided into 4 sub-sequences.

)7(),6(),5(),4(),3(),2(),1( xxxxxxx

1st section: , 2nd section: )4(),3(),2(),1( xxxx

)5(),4(),3(),2( xxxx

),5(),4( xx

),2(),1( xx

, 3rd section: ,

4th section: . After cycle moving,

we obtain: , ,

)6(),5(),4(),3( xxxx

)4(),3(),2(),5( xxxx

)7(),6( xx

)4(),3( xx

)4(),3(),6(),5( xxxx and )4(),7(),6(),5( xxxx . We

add them together with parallelism principle and normal-

ize them. The AP digital sequence can be expressed as

)4(4

4

1,)7()3(3

4

1,)6(2)2(2

4

1,)5(3)1(

4

1xxxxxxx .

Now, we analyzes the connection between kX

and

. According to the character of discrete Fourier trans-

form, is the FFT of sequence

kX

kX Nxx ,1 x,...,2

cycle moving in time domain. Thus

N

kij

ekXkX2

, i = 0, 1, …, N - 1 (6)

where (k =0,1,…N-1). The

spectrum of APFFT is the sum of , thus:

kX

1

0

/2N

n

Nknjenx

kX

1

0

/21

0

11 N

i

NkijN

iAP eiX

NiX

NkX (7)

Supposing that )2

( 00

N

nkj

enx , is a single fre-

quency complex signal, thus,

N

kijN

m

N

kmjN

i

jAP eemxe

NkX

21

0

21

02

][1

0

1

0

21

0

/22

0001 N

m

mkkjN

i

Nikkjj eeeN

Nkkj

kkj

Nkkj

kkjj

e

e

e

ee

N /2

2

/2

2

2 0

0

0

00

1

1

1

11

NkkjNkkj

kkjkkj

NkkjNkkj

kkjkkjj

ee

ee

ee

eee

N ////2 00

00

00

000

1

Nkk

kke

Nj

/sin

sin1

02

02

20

(8)

Thus, the normalized spectrum of APFFT is

Nk

ke

NkX j

ap /sin

sin12

2

20

(9)

Having taken into account all the combinations of N points sequence including x(N), the plus phase error and the minus of x(N) offset each other, thus the phase error is zero.

Figure 3 is the spectrum of the same signal with FFT and APFFT respectively. ICI caused by CFO has more difference between them. The ICI of FFT is higher than that of APFFT.

Figure 4 is the phase and amplitude spectrum of OFDM and APOFDM systems with frequency offset 0.05. From Equation (1), (9), Figure 3 and Figure 4, we can indicate that, with APFFT, the phase error of the re-ceiver signal is 0 and the normalized ICI power is the square of that of FFT signal. Therefore, its sensitivity to

Figure 3. Signal spectrum of APFFT and FFT

Figure 4. Phase and amplitude spectrum of signal s=exp (j*(w*t*20/N+20*pi/180)) with APFFT and FFT, frequency offset is 0.05




1

ICI caused by CFO is lower and APFFT is more suitable for OFDM system.

3. Performance Analysis for AP-OFDM Sys-tems with CFO

The change of performance duo to CFO is more inter-ested in the research content. In this section we use a method similar to the one used in [16] to calculate the exact BER of the proposed system in the presence of a frequency shift. We assume that the system operates over an AWGN channel and that the data symbols are quater-nary phase-shift keying (QPSK) modulated. Similar analysis is possible if a 64-QAM modulation is used [16]. We assume an ideal additive white Gaussian noise (AWGN) channel. The CFO does not change during one OFDM symbol. The sampled signal for the kth sub-channel after the receiver fast Fourier transform processing can be written as:

1

00,

, 0, 1, ,N

k k l k l kl l k

y X S S X n k N

(10)

where

1 , 1 3 , 1 5 , 1 7 ,

3 , 3 3 , 3 5 , 3 7 ,

5 , 5 3 , 5 5 , 5 7 ,

7 , 7 3 , 7 5 , 7 7

k

j j j j

j j jX

j

j j j j

j j j

j

denotes

the transmitted symbol for the kth sub-carrier, nk is a complex Gaussian noise sample (with its real and imagi-nary components being independent and identically dis-tributed with variance ), and N is the number of sub-carriers. The second term in (12) is the ICI caused by the CFO. The sequence (ICI coefficients) depends on the CFO and is given by

2

kS

0

2

/sin

sin1

j

APk eNk

k

NAkXS

(11)

If the ICI assumed to be a Gaussian- distributed ran-dom variables (RV) with a zero mean, both the BER and SER can readily be computed for carious modulation formats. The approximate error rates then can be ex-pressed in terms of the function and effective SNR for the kth sub-carrier as

)(Q

2

2

0

1 APICIS

Seff

S

(12)

where 22/ ks XE which is the SNR for the kth

sub-carrier in the absence of a CFO. The variance of the signal constellation

2

kXE will be independent of if all sub-carriers use the same modulation format, which is the normal case. The variance of the ICI on the kth sub-car-rier can be given by

1

,0

22N

kllklAPICI S (13)

Compared with FFT-OFDM systems [16], the variance of the ICI of APFFT-OFDM systems can be expressed by

1

0

1

0

2222 ˆ

var1

ˆN

n

N

n

ICIAPICIAP N

n

N

nN

NnX

N

1

0

223

2

2ˆ N

n

ICI nNnNN

2

12

6

121ˆ 33

2 NNN

NNNN

NICI

222

2

ˆ3

1ˆ

3

12ICIICI

N

N

(14)

Equation (14) shows that, after APFFT, the ICI noise to every sub-carrier is the third of that in FFT system. Using the method in [16], the effect caused by CFO can be expressed with the losing of SNR.

SD 2

3

1

10ln

10 (15)

4. APOFDM Performance Test with Simulation

4.1 The Design Based SIR

According to the method of ICI cancellation in [15], we choose N=64, L=2, M=1 and SIRmin=25dB to design a system and compare its performance with FFT system, Zhao-Haggman (L=2) system and A-Seyedi system. Fig-ure 5 shows that the performance of APFFT is better than FFT and Zhao-Haggman (L=2) systems in low frequency offset, and poorer in high frequency offset. We will only be interested in frequency shifts in the range 5.0 .

4.2 The Design Based BER

As SIR can not represent the system performance, and the result in Figure 6 obtained before frequency com-pensation, thus the APFFT-OFDM system SIR can not necessarily better than the other existing systems. The model method in [12] is used and all phase compensa-tion is adopted, BER performances are compared in Fig-ure 6. The parameters are same to the above. Assuming that channel is pilot and the pilot frequency account for 20% of the symbol. The CFO is 0.5. From Figure 6, when BER=10-4, APFFT is better than Zhao-Haggman system and A. Seyedi system 1.1 dB and 0.8 dB respec-tively. This is because of the zero phase error of APFFT with CFO. Equation (14), Figure 3 and Figure 4 can prove it to be correct. That is to say, the simulation and the theory are consistent.

122 ICI Performance Analysis for All Phase OFDM Systems

Figure 5. The correction between SIR and normalization frequency offset in ideal AWGN

Figure 6. Calculated and simulated BER for the FFT system, Zhao-Haggman system, A. Seyedi system and APFFT sys-tem over an AWGN channel with perfect equalization

Figure 7. BER of FFT, Zhao-Haggman, A. Seyedi and APFFT-OFDM system over a frequency-selective Channel

4.3 BER of Frequency-Selective System

APFFT can not estimate the CFO but can reduce the sen-sitivity to CFO. Thus, it is true that APFFT is more suited to the OFDM system. In a frequency-selective channel, 4 sub-paths are chosen to establish channel model and

every sub-path attenuation coefficient accord with Rayleigh distribution. Channel model and Power spectral density are the same with [18]. The cyclic prefix is TCP = T/8. Figure 7 shows that, when channel is perfect, the lowest BER varies considerably in all the schemes we have mentioned above. The BER of APOFDM is best.

5. Conclusions

The amplitude of FFT has the character of function sinc in frequency domain. Moreover, the amplitude spectrum of APFFT is the square of function sinc. At the same time, the phase of FFT can change with CFO but APFFT can not. The plus phase error and minus phase error balance out each other, thus, the phase error of APFFT is zero. So, APFFT is more suitable for OFDM system. Compared with conventional OFDM system, APFFT reduce the sensitivity of system to CFO. In this paper, AP-OFDM system is established, the character of ICI cancellation in OFDM system is proved in mathematics, and the meth-ods in [15] are used to analyze the performance of APOFDM system. Through Monte Carlo simulations, we have shown that the proposed system has better per-formance in the presence of an oscillator frequency offset or when ICI is created as a result of channel fading. Sig-nificantly larger gains are achieved when the equalization process is imperfect.

REFERENCES

[1] R. W. Chang and R. A. Gibby, “A theoretical study of performance of an orthogonal multiplexing data transmis-sion scheme,” IEEE Transactions on Communications, Vol. COM-16, pp. 529-540, August 1968.

[2] L. J. Cimini, Jr., “Analysis and simulation of a digital mobile channel using orthogonal frequency division mul-tiplexing,” IEEE Transactions on Communications, Vol. COM-33, pp. 665-765, July 1985.

[3] N. A. Dhahi, et al., “Optimum finit-length equalization for multicarrier transceivers,” IEEE Transactions on Communications, Vol. 44, pp. 56-64, January 1996.

[4] J. J. van de Beek, M. Sandell, and P. O. Borjesson, “ML estimation of time and frequency offset in OFDM sys-tems,” IEEE Transactions on Signal Process., Vol. 45, No. 7, pp. 1800-1805, July 1997.

[5] U. Tureli, D. Kivanc, and H. Liu, “Experimental and ana-lytical studies on a high-resolution OFDM carrier fre-quency offset estimator,” IEEE Transactions on Vehicular Technology, Vol. 50, No. 2, pp. 629-643, March 2001.

[6] M. J. F.-G. Garcia, O. Edfors, and J. M. Paez-Borrallo, “Frequency offset correction for coherent OFDM in wire-less systems,” IEEE Transactions on Consumer Electron-ics, Vol. 47, No. 1, pp. 187-193, February 2001.

[7] M. Luise, M. Marselli, and R. Reggiannini, “Low-com-plexity blind carrier frequency recovery for OFDM sig-nals over frequency-selective radio channels,” IEEE Transactions on Communications, Vol. 50, No. 7, pp.




1182-1188, July 2002. [8] J. Armstrong, “Analysis of new and existing methods of

reducing intercarrier interference due to carrier frequency offset in OFDM,” IEEE Transactions on Communications, Vol. 47, No. 3, pp. 365-369, March 1999.

[9] M. Gudmundson and P. O. Anderson, “Adjacent channel interference in an OFDM system,” in Proceedings of 46th IEEE Vehicular Technology Conference, pp. 918-922, 1996.

[10] C. Muschallik, “Improving an OFDM reception using an adaptive Nyquist windowing,” IEEE Transactions on Consumer Electronics, Vol. 42, No. 3, pp. 259-269, Au-gust 1996.

[11] Y. Zhao and S. G. Haggman, “Sensitivity to Doppler shift and carrier frequency errors in OFDM systems - the con-sequences and solutions,” in Proceedings of 46th IEEE Ve-hicular Technology Conference, pp. 1564-1568, April 1996.

[12] Y. Zhao and S. G. Haggman, “Intercarrier interference self-cancellation scheme for OFDM mobile communica-tion systems,” IEEE Transactions on Communications, Vol. 49, pp. 1185-1191, July 2001.

[13] Y. Zhao and S. G. Haggman, “Intercarrier interference compression in OFDM communication systems by using

correlative coding,” IEEE Communications Letters, Vol. 2, No. 8, pp. 214-216, August 1998.

[14] K. Sathananthan and C. Tellambura, “Forward error correc-tion codes to reduce intercarrier interference in OFDM,” in Proceedings of IEEE International Symposium on Cir-cuits and Systems, Vol. 4, pp. 566-569, May 2001.

[15] A. Seyedi and G. J. Saulnier, “General ICI self-cancellation scheme for OFDM systems,” IEEE Transactions on Vehicular Technology, Vol. 54, No. 1, pp. 198-210, January 2005.

[16] K. Sathananthan and C. Terambulla, “Probability of error calculation of OFDM systems with frequency offset,” IEEE Transactions on Communications, Vol. 49, No. 11, pp. 1884-1888, November 2001.

[17] Z. H. Wang, Z. X. Hou, and F. Su, “All phase FFT spec-trum analysis,” Journal of China Institute of Communications, Vol. 24, No.11A. pp. 16-20.

[18] Y. Oda and K. Tsunekawa, “Advanced LOS path loss model in microwave mobile communications,” 10th In-ternational Conference on Antennas and Propagation, Conference Publication No. 436, IEE ’97, April 14-17, 1997.



1

Design and Implementation of a Novel Digital Frequency Superposition Testing Power Supply for Induction Motor

Xingjian Dong1, Shenxian Zhuang1, Wen Yan2

1School of Electric Engineering, Southwest Jiaotong University, Chengdu, China; 2 Hunan YinHe Electric Corporation, Changsha, China. Email: [email protected]

Received April 15th, 2009; revised June 8th, 2009; accepted June 13th, 2009.

ABSTRACT

A novel design and implementation of frequency superposition testing power supply for induction motor is proposed. An equivalent power using dynamic space voltage vector synthesis is generated to replace the two separate powers of the traditional method. The principle of frequency superposition testing is firstly introduced, and then the detailed design and implementation of the digital frequency superposition power are given. The simulation of the power supply system shows the promising results. Finally, experimental results validate the feasibility and reliability of the proposed power.

Keywords: Digital Power, Frequency-Superposition Testing, SVPWM

1. Introduction

Currently, there are two main methods for temperature rising testing which is an important index of induction motors. One is direct loading testing; the other is fre-quency superposition testing [1]. When a direct loading testing is performed, a duplicate motor must be coupled with the tested induction motor mechanically. However, for a large scale induction motor, it’s difficult to add a rated load on the shaft and it’s not safe. Frequency su-perposition testing is an equivalent loading testing with out a duplicate motor coupled. The tested induction mo-tor is fed with two separate voltage sources of two dif-ferent frequencies. When a frequency superposition test-ing is performed, the most sophisticated case is that the two power supplies are supported by a dynamotor. There must be at least 4 to 5 operators to operate it. Another drawback is that the dynamotor runs with a big noise.

This paper designs and implements a digital frequency superposition testing power supply by using open loop constant volts per hertz control based on SVPWM (Space Vector Pulse Width Modulation). The reference voltage is synthesized with the main reference voltage and the auxiliary reference target voltage by using parallelogram rule. The two reference voltages are controlled via a HMI (Human Machine Interface) client.

The concept of frequency superposition testing is sim-ply described in Section 2. The method to generate a

synthesized voltage vector using space voltage vector synthesis and the digital implementation of the frequency superposition testing power supply are described in Sec-tion 3. The simulation and the experiment results of the power supply proposed are given in Section 4, Section 5 respectively. The conclusion is presented in Section 6.

2. Traditional Frequency Superposition Testing Power

In the traditional frequency superposition testing [1,2], the tested motor is fed with two independent powers cou-pled with a transformer. Figure 1 shows the schematic of the traditional frequency superposition power with dynamotor fed. Traditionally, the auxiliary frequency is (80% - 120%) times of main frequency. Under the syn-thesized power supply, the voltage, current, speed, flux of the tested motor run with a beating frequency. When the synthesized stator flux runs faster than the rotor’s, the tested motor runs as a motor, when the synthesized stator flux runs slower than the rotor’s, the tested motor runs as a generator. The RMS (Root Mean Square) values of the stator currents increase while the auxiliary voltage in-creases. Until the RMS values of the electrical parameters are around the rated values. The motor works as a rated load added on its shaft.

Design and Implementation of a Novel Digital Frequency Superposition Testing Power Supply for Induction Motor 125

Figure 1. Schematic of the traditional frequency superposi-tion testing with dynamotor fed for induction motor

3. Design and Implementation of the Digital Frequency Superposition Testing Power Supply

3.1 Design of the Proposed Frequency Super- Position Testing Power Supply

The hardware diagram of the proposed power has been shown in Figure 2. The hardware system consists of 1 power converter unit, 1 digital control board, 1 power analyzers and 1 HMI client.

The drive is based on TMS320LF2812 DSP platform. The main and auxiliary reference voltages are sent to the control board from the HMI client.

3.2 Implementation of the Power Converter Unit

The structure of the proposed three-phase voltage source power converter is shown in Figure 3. Uu, Uv and Uw are the output voltages. S1…S6 are the six IGBTs that shape the output, which are controlled by Su+ and Su-, Sv+ and Sv-, Sw+ and Sw-. When an upper IGBT is switched on, the corresponding lower IGBT is switched

off, i.e., when Su+, Sv+ or Sw+ is 1.the corresponding Su-, Sv- or Sw- is 0. The on and off demonstrates the states of the up transistors S1, S3 and S5.

There are eight possible combinations of on and off patterns for Su+, Sv+, Sw+. The derived output phase and line-to-line voltages in terms of DC supply voltage Udc are shown in Table 1.

The objective of SVPWM technique [3] is to approxi-mate the reference voltage vector Uo by a combination of the eight switching patterns. One combination is shown in Figure 4.

For every PWM period, the desired reference voltage Uo can be approximated by having the power converter in two adjacent switching pattern Ux,Ux+1 of T1 and T2 duration, respectively. The output voltages are decided by the selection of the duration time T1 and T2. The output voltage space vector is shown in Figure 5.

3.3 Synthesis of the Output Reference Voltage

Figure 6 shows a synthesis of the dynamic space voltage vectors.

According to Figure 6, using the parallelogram rule, we have:

2 2

2 12 cos(out main aux main auxU U U U U )

(1)

1

1 2

sin sinarg tan

cos cosmain aux

main aux

U U

U U

2

(2)

where, , mainU 1 are the amplitude and electrical angle

of the main voltage vector. , auxU 2 are the amplitude

and electrical angle of the auxiliary voltage vector. , outU

are the amplitude and electrical angle of the synthe-

sized voltage vector.

Figure 2. Hardware diagram of digital power for frequency superposition testing


126 Design and Implementation of a Novel Digital Frequency Superposition Testing Power Supply for Induction Motor

Figure 3. The diagram of the power converter unit

Table 1. Output phase and line-to-line voltages in terms of DC supply Udc

Su+ Sv+ Sw+ Uu Uv Uw Uuv Uvw Uwu0 0 0 0 0 0 0 0 0

0 0 1 -1/3 -1/3 2/3 0 -1 1

0 1 1 -2/3 1/3 1/3 -1 0 1

0 1 0 -1/3 2/3 -1/3 -1 1 0

1 1 0 1/3 1/3 -2/3 0 1 -1

1 0 0 2/3 -1/3 -1/3 1 0 -1

1 0 1 1/3 -2/3 1/3 1 -1 0

1 1 1 0 0 0 0 0 0

(000) (001) (101) (101) (101) (001) (000)

Su+

Sv+

Sw+1 PWM Period

T2

T1

Figure 4. One combination of the eight switch patterns

According t 1), (2) we can get the target voltage

space vector in every carrier period. The sector,

the target voltage space vectors and the duration time of the two nearest adjacent voltage vectors can be decided by using SVPWM principle [4-7].

o (

outU

4. Simulation Results

In order to verify the validity of the proposed frequency superposition power, some simulation work has been performed. The parameters of the tested induction motor involved in the simulation are given in Table 2. Figure 7 gives the simulation results of the proposed model.

Before 2s, the motor is only fed with the main power; the voltage of the auxiliary power is set to 0, i.e. the mo-tor runs under No-Load Mode.

At 2s, the auxiliary power is added to the motor, the motor runs under a frequency superposition mode.


Figure 8 shows the DC bus voltage and the stator current obtained with the Tektronix Scope. Such experimental results were obtained with a roller induction motor. The parameters of the tested motor the experiment are given in Table 2.

In this case, the main frequency is given to 21.8Hz, the auxiliary one is 17.8Hz (82% of 21.8). The auxiliary voltage is 0.2 times as the main power which is given to the rated voltage.

This experimental figure shows that the stator current is oscillated at a RMS value equals to the rate current.

Figure 5. Diagram of the voltage space vector

12

Figure 6. Synthesis of the dynamic space voltage vectors

Table 2. Rated parameters of the tested induction motor

Parameters Value

Rated voltage (V) 380

Rated current (A) 107

Rated frequency (Hz) 21.8

Rated power (kW) 60


Design and Implementation of a Novel Digital Frequency Superposition Testing Power Supply for Induction Motor 127

Figure 7. Simulation results of the stator phase A current and the DC bus voltage

Figure 8. Experimental results of the proposed frequency superposition power supply 6. Conclusions

In this paper, a design of digital motor drive for fre-quency superposition testing is proposed. It has been im-plemented in an open loop control by using digital signal

processor TMS320LF2812. Simulation results show that the stator current oscillated at a beating frequency after Frequency-Superposition. The tested motor runs as a rated load coupled with it. The experimental results indi-cate that this power supply can realize the equivalent loading test.

REFERENCES

[1] I. Boldea and A. A. Nasar, “The induction machine hand-book,” CRC, Boca Raton, Florida, 2002.

[2] H. R. Schwenk, “Equivalent loading of induction machine for temperature tests,” IEEE Transactions on Power Ap-paratus and Systems, Vol. 96, No. 4, PP. 1126-1131, 1977.

[3] V. Vlatkovic and D. Borojevic, “Digi-tal-signal-processor-based control using the constant V/F principle and a space-vector PWM algorithm,” IEEE Transactions on Industrial Electronics, Vol. 41, No. 3, pp. 326-332, 1997.

[4] H. W. van der Broeck, H. C. Skudelny and G. V. Stanke, “Analysis and realization of a pulse width modulator based on voltage space vectors,” IEEE Transactions on Industrial Application, Vol. 24, No. 1, pp. 142-150, 1988.

[5] Y.-Y. Tzou and H.-J. Hsu, “FPGA realization of space vector PWM control IC for 3 phase PWM inverters,” IEEE Transactions on Power Electronics, Vol. 12, No. 6, pp. 953-963, 1997.

[6] K. L. Zhou and D. W. Wang, “Relationship between space-vector modulation and three-phase carrier-based PWM: A comprehensive analysis,” IEEE Transactions on Industrial Electronics, Vol. 49, No. 1, pp. 186-192, 2002.

[7] Timothy and Skvarenina, “The power electronics hand-book,” CRC, Boca Raton, Florida, 2002.


Call for Papers

Asia-Pacific Power and Energy Engineering Conference

(APPEEC 2010)

March 28-31, 2010 Chengdu, China

http://www.scirp.org/conf/appeec2010

This conference is sponsored by IEEE Power & Energy Society (PES), State Grid of China, Sichuan

University, Chongqing University and Wuhan University. The conference proceedings will be published by

IEEE. All papers accepted will be included in IEEE Xplore and indexed by EI and ISTP. This conference

will be held in Chengdu where “The Land of Abundance” is located.

Sponsors: IEEE Power Engineering Society (PES), State Grid of China, Sichuan University, Chongqing University and Wuhan University.

Topics

Power Generation-Conventional and Renewable Hydropower technologies and applications

Thermal power technologies and applications

Nuclear energy generation and utilization

Wind power generation and utilization

Bio-energy technologies, process and utilization

Geothermal and tidal wave energy

Photovoltaic for solar power application

New technologies and design for energy efficiency

New technologies for minimizing CO2 generation

Environmental-friendly technologies for power generation

Power System Management Power system management technologies

Integrated substation automation technologies

Power system monitoring and mitigation technologies

Online monitoring and fault diagnosis system

Control strategies for modern power system stability

Modeling and simulation of large power systems

Application of wide area measurement system (WAMS)

Power system analysis and optimization

Load modeling, estimation and forecast

Power system planning and operation

Power Transmission and Distribution Ultra High Voltage (UHV) technologies

HVDC and flexible AC transmission system (FACTS)

Over-voltage, lightning protection and grounding

Electromagnetic transient in high voltage power systems

Insulation condition monitoring in power systems

Advanced distribution and SCADA technologies

Electromagnetic compatibility in power systems

Plasma physics and the pulsed power technology

Electromagnetic analysis in power systems

Smart Grid Concept and structure frame Operation and control Distributed generation and renewable energy application Information and smart meter reading Integrated energy and communications

Submission Requirements: The working language of the conference is English. All papers must be submitted in IEEE electronic format. Instructions and full information on the conference are posted on the conference website. Anyone wishing to propose a special session or a tutorial should contact us: [email protected].

Important Dates:

Sept. 30, 2009 Deadline for Full Paper Submission

Nov. 30, 2009 Notification of Acceptance

Contact Information:

Website: http://www.scirp.org/conf/appeec2010


http://www.scirp.org/conf/appeec2010

Date post:	18-Mar-2021
Category:	Documents
Upload:	others
View:	3 times
Download:	0 times

ISSN: 1942-0730 (Print), 1942-0749 (Online) NO.2-01-02... · 2011. 9. 7. · Yu Fu Wing Li Peisheng...

Documents