The paper shows that Carter's coefficient gives permeance values almost identical with the true ones for
narrow teeth, as established by Carter's later work, unless the tooth width is less than the radial air gap.
A small increase in the coefficient is given for tooth widths down to zero. The flux-density distribution
obtained by the superposition of waveforms for successive slots as in isolation, is also seen to be
indistinguishable in all cases from the true waveforms for near slots, as recently established by Freeman.
It is shown that the harmonic content of the ripple is conveniently expressed by a wavelength spectrum

© All Rights Reserved

Просмотров: 45

The paper shows that Carter's coefficient gives permeance values almost identical with the true ones for
narrow teeth, as established by Carter's later work, unless the tooth width is less than the radial air gap.
A small increase in the coefficient is given for tooth widths down to zero. The flux-density distribution
obtained by the superposition of waveforms for successive slots as in isolation, is also seen to be
indistinguishable in all cases from the true waveforms for near slots, as recently established by Freeman.
It is shown that the harmonic content of the ripple is conveniently expressed by a wavelength spectrum

© All Rights Reserved

- 13584Spectral Analysis Om
- SSC 347 NonLinear Analysis of Marine Structures
- Code
- 10013T2ADECMooringStudyFinal.pdf
- FILE NO. IIT- 260-27 CPT_25_(12 STD) NEW_MPC_17-11-08_Q.P.
- PaperEVM Newer
- 08r059210403 Signals and Systems
- Nr210402 -Probability & Random Variables
- Parametric Time Domain Modelling
- FAULT DIAGNOSIS ON BEARINGS IN SYNCHRONOUS MACHINE BY PROCESSING VIBRO-ACOUSTIC SIGNALS USING HIGHER ORDER SPECTRAL.
- RM E Prestressing Basic Part2 AASHTO Imp
- Spectral Kurtosis
- Paper_Spectral_Leoben_2012.pdf
- PS-5.docx
- Spectra Lo Dev 2
- 12.714_S2L05
- 2.pdf
- Sample Prob 2
- Zhan Gji2013
- Lecture Handout 19Jan09

Вы находитесь на странице: 1из 6

Volume 114

Power

S. Neville, B.Sc.(Eng.), D.I.C., C.Eng., F.I.E.E.

Synopsis

The paper shows that Carter's coefficient gives permeance values almost identical with the true ones for

narrow teeth, as established by Carter's later work, unless the tooth width is less than the radial air gap.

A small increase in the coefficient is given for tooth widths down to zero. Theflux-densitydistribution

obtained by the superposition of waveforms for successive slots as in isolation, is also seen to be

indistinguishable in all cases from the true waveforms for near slots, as recently established by Freeman.

It is shown that the harmonic content of the ripple is conveniently expressed by a wavelength spectrum.

1

Introduction

The analysis of the magnetic field over an open slot,

given by Carter in 1901,1 and his famous coefficient a

expressing the loss of permeance as a fraction of s\g, were

based on the assumption that the slot considered was remote

from all others. In electrical machines, however, we have a

uniform succession of slots, separated by teeth of similar

width, and engineers have been uncertain about the possible

error involved in using Carter's coefficient in these practical

circumstances.

The reduction in flux density on the opposite face of the

gap obviously extends considerably beyond the width of the

slot itself, and a view that has been widely held suggests that,

if the intervening teeth are so narrow that the loss of density

due to a slot is still appreciable at the centreline of an adjacent

tooth, Carter's coefficient does not properly apply. It is said

that to use a then is to ignore the loss of flux occurring,

in Carter's analysis, at distances greater than the halftooth

width. This view has been supported in important mathematical expositions of the subject. Gibbs,2 for example, in

Fig. 104 of Reference 2, contrasts the curtailed flux distribution thus attributed to the use of a, with the true distribution

calculated (from elliptic functions) for the particular width of

tooth, which shows a distinctly greater loss of permeance.

But this view of the situation is not correct. Whenever

Carter's coefficient is used, allowance is inevitably being made

for all the loss of permeance occasioned by an isolated slot,

and when slots are near together this full loss of permeance

is still being attributed to each of them.

The use of a really assumes that (at least as to mean height)

the flux distribution over a succession of slots is that which is

obtained by superposing the distribution curves for all the

slots that are near enough to have any appreciable effect.

When this is done, the resultant curve agrees very closely

with the true distribution, for all slot widths and all practical

tooth widths. For example, if only the combined effects of

adjacent slots need to be considered, the loss of flux density

opposite the middle of a tooth, implied when a is used, is

twice the loss of density at this distance from an isolated slot;

and in Gibbs's Fig. 104 this will be seen to accord very well

with the true density at that point. Gibbs makes it quite

clear (his pages 213 and 214) that no differences could be

found in numerical comparisons of Carter's a with the

Paper 5345 P, first received 17th March and in revised form 18th

May 1967

Mr. Neville was formerly with Associated Electrical Industries Ltd.,

TrafTord Park, Manchester 17, England

64 P23,

with the diagram referred to.

density (on the opposite side of the gap) for single isolated

slots of various widths, expressed in terms of Bo, the uniform

density which would occur in the absence of the slot, or at a

large distance from it.* For distances greater than one gap

width from the side of the slot, the curves are shown separately

to an enlarged scale. The centre of the slot is marked on each

curve, together with the minima for intermediate slot widths

to facilitate interpolation. Since Carter's coefficient expresses

the loss of permeance as a fraction of a\g, there is marked

against each curve the height of a rectangle, of width equal

to the slot, having an area equal to all the area above that

curve. The height of the rectangle is 1 a.

It will be noticed that for wide slots most of the loss of

density occurs over the width of the slot, but for narrow slots

by far the greater part occurs beyond the width of the slot

itself. For example, with s/g = 3, 90% of the loss occurs over

the slot and only 5% on each side of it; whereas, with

sjg = 0-3, only 23% of the loss occurs over the width of the

slot and 38-5% on each side of it. Bearing in mind that tooth

widths are in general comparable with slot widths, this

characteristic obviously profoundly affects the results of

superposing the effects of a sequence of slots.

3

Process of superposition

Fig. 2 shows diagrammatically the variety of results

that can thus occur in different circumstances. If either the

slot or the tooth is wide, as in Fig. 2a, the effect of each slot

may hardly reach the middle of the adjacent slots, and only

two curves have to be superposed to obtain thefluxdistribution (on the opposite face of the gap) implied in the application

of a to such a case. If both tooth and slot are narrow, as in

Fig, 2b, several curves (each as for an isolated slot) overlap.

The combination of such a series of curves, in any half slot

pitch, will be seen to amount to adding together the ordinates

of successive portions of one curve, each of length equal to

half a slot pitch, alternately reversed. The four such components in Fig. 2b are marked by full lines. Owing to the

reversals of slope, their resultant shows a much smaller

variation of density than would occur with wide teeth.

Calculation of these curves involves only logarithmic and trigonometric functions. The procedure is fully explained in Reference 2, pp. 103-116; it presents

no difficulty, but accurate numerical work is necessary to obtain satisfactory

curves. The author can provide quarto-sized copies of the illustrations on a very

fine grid, to enable close readings to be taken.

1245

slot widths given in Fig. 1, combined in each case with three

small widths of tooth, namely tig = 0-5, 0-3, and zero. For

t/g = 0, and s/g = 0-1,2, and 5, the successive stages of the

synthesis are shown by dotted lines. For s/g = 5, the dotted

curve marked 5' shows the distribution from the side of the

slot to its centre line, taken from Fig. 1. Added below this is

the distribution beyond the side of the slot, again from Fig. 1,

2 -X/9

1-5

+o

constructed in the same way.

For comparison with these synthetised curves, the

true field distributions are shown as full lines. The analysis

of the field over near slots was given by Carter in his later

2-0

23

X/9

O-9S

O*9O

O-8S

O'ftO

Fig. 1

Flux distribution opposite isolated slots

before the slot centre is reached (an instance of Fig. 2a),

giving the resultant shown by a dashed line. At s/g = 2, the

curves 2' and 2" are obtained in the same way, but a further

portion of the curve of Fig. 1 remains to be added to 2",

now in the original direction, and this component becomes

negligible at the tooth centreline (compare Fig. 26). For a

still narrower slot, sig = 0 7, the curves 0 7', 0 7" and 0 1'"

are obtained as before, but two further components remain

of significant magnitude, giving 0-7"" and the final dashed

1246

evaluated, with regard to mean value (or permeance), by

Coe and Taylor,4 and recently, with regard to harmonic

content, by Freeman.5 The calculation of these distributions

generally involves elliptic functions. The process then gives

the ordinates of a series of curves corresponding to two

arbitrary parameters; and to obtain curves for specified values

of sfg and t/g requires either graphical crossplotting or an

elaborate process of trial and error. Ordinates for the true

curves shown here have been specially computed by Freeman,

PROC. IEE, Vol. 114, No. 9, SEPTEMBER 1967

paper.5 For slots of infinite width, the distribution can be

calculated directly for specified values of tjg, although still

involving elliptic functions. This case has been investigated

by Davy,6 but, to obtain satisfactory results, more accurate

Bm/B0, i.e. in the estimated value of gap permeance, is considerable, but even in this extreme case its numerical value

does not exceed 3% at slg = 1-4. This difference rapidly

diminishes for wider or narrower slots; so that if the permeance, as estimated from a, is reduced by a small allowance

related to the slot width, or even by an arbitrary 1 %, no

error of practical importance is likely to be made.

4.1

B/Bo

1-0

of course, be completely represented by appropriate small

increases in a when dealing with very narrow teeth, to give

exactly the correct gap permeance. In Fig. 4 the well known

curve of a for isolated slots is plotted to a logarithmic base

of s/g covering the whole range of practice. Just above it,

the curve marked a0 shows the slightly increased value of a

required to give the true permeance with teeth of zero width;

i.e. to bring the synthetised curves for t/g = 0 in Fig. 3 down

to the true curves.* The numerical increase is small, having

its maximum in the region of slg = 1-4-2, but the percentage increase at small values of slg is, of course, quite

appreciable. It is perhaps remarkable that the complete data

defining exactly the permeance of singly slotted air gaps, for

all ratios of s, t, and g, lie between these two curves.

Moreover, the greater part of even this small difference is

concerned with tooth widths smaller than occur in practical

design. In the sauie diagfam the values of(a0 a) are shown

separately to an enlarged scale, togethej with the smaller

increases of a required to yield exactly correct permeances

with tig 0 3 or 0-5, in accordance with the curves of

Fig. 3. For tig = 1, the correction can only just be shown

even on this scale. A dotted line shows approximately the

curve for tjg = 0 1, derived by interpolation.

5

resultant

Fig. 2

Synthesis of fields for a succession of slots

a tlg= \,slg = 2

b tig = h slg = i

The case of teeth of zero width (septa), with any finite slot

width, is readily calculated manually, involving only the

complete elliptic integrals. This case becomes simply algebraic

for infinite slot width.

It will be noticed at once that, for a tooth of only half the

width of the gap, there is little difference between the true

distribution and the synthetised curves which are obtained

by the use of a. Numerically, the difference hardly reaches

0-3% of BQ, showing that a overestimates the permeance in

such cases to only that extent. This difference occurs between

sfg = I -4 and 2, and diminishes for both wider and narrower

slots. It has been verified that, for teeth equal in width to

the gap, the differences are only one fifth of the above values.

For still narrower teeth the differences become a little

greater, but a conspicuous feature is that in all cases (including

even tig = 0) the range of variation of density on the opposite

face of the gap, as derived from the superposition of the fields

of isolated slots, is almost identical with the true value; and

the harmonic content of the two curves is for all practical

purposes identical. Only in the mean height is there any

appreciable difference between the two curves.

For tig = 0-3, a condition that may occur in some very

large generators, the difference in BmlB0 reaches a maximum

of 0-65% at s/g I - 4 - 2, diminishing for larger or smaller

slots. If, therefore, it may be assumed that in the class of

machine in which t/g may be as small as 0-3, the slot ratio

s/g would be similarly small, it is clear that the use of

Carter's coefficient in such cases would be unlikely to affect

practical calculations by more than a small fraction of I %.

PROC. IEE, Vol. 114, No. 9, SEPTEMBER 1967

Harmonic content

It will be seen in Fig. 3 that the shape of the synthetised

curves is even nearer to the truth than is the mean height.

If, therefore, the process were carried out for a wider range

of tooth widths, and the fundamental of each of the resulting

periodic waveforms were determined (as a fraction of Bo),

these values could be plotted to a base of s/g + t/g to form

a curve from which the fundamental for any given slot pitch

could be read off. (The chosen values of tig could even be

made negative, representing overlapping slots, and the curve

thus extended down to a theoretical zero pitch.)

Moreover it is obvious that, in such a curve, the ordinates

at one half (or one third) of any given slot pitch would

represent the fundamentals of waveforms obtained by superposing two (or three) of the waveforms for that slot pitch,

equally spaced by one half (or one third) of that pitch, and

this is simply twice the second harmonic (or three times the

third harmonic) of the original waveform. The one curve

would thus give the amplitudes (and signs) of all the

harmonics.

Fig. 5 gives a set of such curves covering all values of t/s

up to 3, and for a series of values of s/g up to 10, with indication of the end points of intermediate curves to facilitate

interpolation. The sign indicated is for cosine waves having

an origin at the middle of a slot. For an origin at the middle

of a tooth, as in Fig. 3, the sign of even terms is unchanged,

but that of odd terms is reversed. To take, for example,

the three cases of Fig. 3, for, say, s/g = 3, we have on that

curve in Fig. 5, at t/s = 0, 0 1 and 0 167, the fundamental

amplitudes 0 081, 0117, and 0139, which agree

well with Fig. 3. For the corresponding second harmonics

we have at t/s + 1 = 0-50, 0-55, and 0-583, on the same

curve, the values 0012, 0024, and 0031 (all positive), to

be divided by two.

5.1

Wavelength spectra

or 'wavelength spectra' of the flux-distribution curves for

isolated slots of various widths. Formally the spectrum is

* The true value of BmIBo, for teeth of zero width, or (I oo),

2(gjs)K'jK, to modulus k = tanh rcg/s

given by

1247

(Fig. 1) by cos 6\n, where 0 = 0 to n corresponds with the

half slot width, and n takes a series of values representing

ratios of slot-pitch/slot-width; then integrating the alternate

positive and negative areas under the derived curve from

0 = 0 until they become inappreciable. The process involves

a very large amount of graphical work, which must be done

with high accuracy if consistent points on the spectrum are

to be obtained.

It was therefore thought that the process of synthesis

illustrated in Fig. 3, but performed by numerical tabulation,

1 0 0p

each case by a 48-ordinate method, would probably give

more reliable results. The curves of Fig. 5 were actually

obtained by this method, and it may be mentioned that all the

calculated points lie exactly on the smooth curves shown.

Finally it must be repeated that these spectra really represent

the results of synthesis from isolated slots, and are not directly

related to the true distribution over near slots. But the

exactly computed harmonics of the true fields presented by

Freeman have shown that over the whole range any difference is remarkably small.

x/9

0-93

0-90

0-85 \

080

O-75

0-70

0 65

O-6O

osa

O-4O

0-30

o-ao

Fig. 3

Comparison

1248

of synthetised

0-8

and

0-7

/

O-6

0-5

04

0-3

O3

ON

0-2

s.

02

X,

-~^0-1 opprox.

O-1

O1

,

i

0-2

Si.

0-5

2-O

as 10

5-O

10

s/g

20

Fig. 4

Modification of a for very narrow teeth

Fig. 5

Wavelength spectra for s/g = 0- 7-10

References

pp.884-888

2 CIBBS, w. j . : 'Conformal transformations in electrical engineering'

(Chapman & Hall, 1958)

3 CARTER, F. w.: 'The magnetic field of the dynamo-electric machine'

J. IEE, 1926,64, pp. 1115-1138

4 COE, R. T., and TAYLOR, H. w.: 'Some problems in electrical machine

design involving elliptic functions', Phil. Mag., 1928, 6, pp. 100-145

5 FREEMAN, E.M.: 'Calculation of harmonics, due to slotting, in the

flux-density wave-form of a dynamo-electric machine', Proc. IEE,

1962, 109 C, pp. 581-588

6 DAVY, N.: T h e field between semi-infinite rectangular electrodes or

magnetic pole-pieces', Phil. Mag., 1944, 35, pp. 819-840

PROC. IEE, Vol. 114, No. 9, SEPTEMBER 1967

Appendix

Note on the superposition of fields

erroneous conclusion that the method has more general

application.

(a) Although the fields due to various distributions of

potential over the same boundaries are, of course, directly

additive, the superposition of fields appropriate to different

boundaries is never analytically valid. Without attempting

1249

of it over a wide range of circumstances can be readily seen.

The field between boundaries containing more than two rightangle corners can be expressed only in terms of elliptic (or

even more complex) functions, whereas expression of the

fields between the corners individually requires only logarithmic or trigonometric functions. If such separate components

could be validly superposed, it would amount to expressing

the higher functions in terms of the more elementary.

(b) Even where this argument does not directly apply, such

superposition almost invariably gives results that bear no

useful resemblance to reality. The well known field over an

infinite plane surface, from one corner of a rectangular pole

parallel to it, provides a typical instance. If two such corners

are brought near together, with either the same or opposite

polarities, the sum or difference of their separate fields gives

a grossly inaccurate representation of the flux distribution

either over a deep slot or between N and S poles.

(c) Only in very few circumstances does such superposition

give results of useful numerical accuracy. Some explanation

1250

be seen in the physics of the case. On the smooth face opposite

to an isolated slot, the great fall in flux density corresponding

to the sudden increase of gap width is offset by displacement

of flux from the air gaps over the adjoining teeth to the region

over the slot. Flux is drawn from a distance on either side

roughly equal to the width of the gap. If, therefore, slots are

separated by teeth of width comparable to the gap, it is

evident that the reduction of density over a tooth, by flux

spreading into a slot on each side of it, must be the sum of

what it would be for two isolated slots. When, however, the

teeth are distinctly narrower than the gap, the flux over each,

even at its maximum density, does not suffice to provide the

amount withdrawn into wide slots. The reduction of density

over the teeth must, therefore, be greater than would be

deduced from superposition, and the density over the slots

will at the same time be slightly curtailed. With very narrow

teeth this deficiency over the tooth is the conspicuous feature

(as at tig = 0 in Fig. 3), although even then the effect over

the slot remains very small.

- 13584Spectral Analysis OmЗагружено:Vikky Gupta
- SSC 347 NonLinear Analysis of Marine StructuresЗагружено:Nils Antoine Freund
- CodeЗагружено:Ishaku Balami
- 10013T2ADECMooringStudyFinal.pdfЗагружено:myusuf_engineer
- FILE NO. IIT- 260-27 CPT_25_(12 STD) NEW_MPC_17-11-08_Q.P.Загружено:Gadde Gopala Krishna
- PaperEVM NewerЗагружено:Donny M. Adhitama
- 08r059210403 Signals and SystemsЗагружено:andhracolleges
- Nr210402 -Probability & Random VariablesЗагружено:SRINIVASA RAO GANTA
- Parametric Time Domain ModellingЗагружено:Hussein Razaq
- FAULT DIAGNOSIS ON BEARINGS IN SYNCHRONOUS MACHINE BY PROCESSING VIBRO-ACOUSTIC SIGNALS USING HIGHER ORDER SPECTRAL.Загружено:IJAR Journal
- RM E Prestressing Basic Part2 AASHTO ImpЗагружено:Orlando Lopes
- Spectral KurtosisЗагружено:Gayatri Ch
- Paper_Spectral_Leoben_2012.pdfЗагружено:ashish38799
- PS-5.docxЗагружено:Fredie Lee
- Spectra Lo Dev 2Загружено:shrine_mi
- 12.714_S2L05Загружено:karim2005
- 2.pdfЗагружено:herdissam1
- Sample Prob 2Загружено:Siddhant Agarwal
- Zhan Gji2013Загружено:serly
- Lecture Handout 19Jan09Загружено:Lem Fei-Bronjul
- ijsrp-p14142Загружено:Joseph Knight
- HSE report rr377Загружено:nanlingjigong
- 12Загружено:joanna
- Chavez Et Al JNeurosciMeth154 2006Загружено:Isaac Aguilar
- IMU Noise and CharacterizationЗагружено:ranchua1905
- Rr210404 Signals and SystemsЗагружено:sivabharathamurthy
- Javid i 1989Загружено:misaq
- Detailed GuidelinesЗагружено:Hanish Sharma
- Unit 2 Part1Загружено:sameerpatel15770
- Mastering In The Box Manual.pdfЗагружено:Franco Salazar

- Maxwell_v16_L03_Static_Magnetic_Solvers.pdfЗагружено:VahidJam
- isbn9789512292776.pdfЗагружено:CasualKilla
- Modeling High-Order Ferromagnetic Hysteresis Using G-PFB TheoryЗагружено:CasualKilla
- High-field Self-starting Permanent-magnet Synchronous Motor - 04643350 - 1981Загружено:CasualKilla
- Maxwell v16 L02 Geometry OperationsЗагружено:VahidJam
- Challenges for Newspapers from going onlineЗагружено:CasualKilla
- MMPA Magnet Design GuideЗагружено:CasualKilla
- Maxwell v16 L01 IntroductionЗагружено:CasualKilla
- m400-50aЗагружено:CasualKilla
- Design Criteria of High Performance Synchronous Reluctance MotorsЗагружено:Yusuf Yaşa
- Vagati1997 - Design of Low-Torque-Ripple Synchronous Reluctance MotorsЗагружено:CasualKilla
- PM412 Chapter 12 - Project Evaluation Communication Implementation and Closeout - Lecture 2Загружено:CasualKilla
- Linear ActuatorsЗагружено:CasualKilla
- Groover Manufacturing 2Загружено:Thale1905
- Groover Manufacturing 1Загружено:Shashank Agrawal

- chapter15-1Загружено:changaboy24
- New Simplified Models and Formulations for Turbulent Flow and ConvectionЗагружено:bagusu_6
- Physics NoteЗагружено:Muhamad Arif
- Biology-Week 14-LAB-12.14-12.15-Plant and Animal Cell Microscope LabЗагружено:Mollie Smith
- TL1 Dr KasikciЗагружено:eaaziz
- Alex Schiffer - Joe Cell - Experimenters Guide to the Joe CellЗагружено:Anonymous UwXe23x
- Dynamics Lecture2 General Curvilinear Motion - Rectangular Components and Projectile MotionЗагружено:JJ Enzon
- ButЗагружено:Joel Wasserman
- Effect of Antenna Size on Gain BandwidthЗагружено:Ngoc Lan Nguyen
- 13-14_SLL_Multiband PIFA Antenna With Ground Plane SlotsЗагружено:Mahesh Patil
- AP AB Manual(3) - Mastermathmentor.comЗагружено:Stanley
- AOFST_7-140Загружено:Ramyaa Lakshmi
- Link 180Загружено:Omar Najm
- Design Aids of Nu I Girder BridgesЗагружено:heryanto555
- 01 Hydrocarbon Phase Behaviour_2Загружено:knaganjaneya
- nikolai tesla finalЗагружено:api-287301687
- Tutorial on Propogation of ErrorsЗагружено:Ram Yadav
- Microsoft Word - Arbitrary Reference Frame TheoryЗагружено:sameerpatel15770
- JTTEЗагружено:Anonymous eCD5ZR
- DIN 7168 ENЗагружено:aleksena
- Biodiesel process intensification in a very simple microchannel device.pdfЗагружено:Emily Tatiana Alvarez Villa
- Kinematics of a PointЗагружено:Estelle Joyce
- Chap2Загружено:Jesuv Cristian Clete
- CE130_Oct 11Загружено:Devin Brar
- 4E5NA Prelim 2013 P2Загружено:Cw Wc Scw
- Review Chapter 1 Physics MotionЗагружено:Hector Velasco
- GC-MSЗагружено:hyutoyr
- Radionics How Does Radionics Work - The Radionics ForumЗагружено:StellaEstel
- FEA 2016 v11 Release NoteЗагружено:KlLee
- Civil-Breadth-Mor-Question-1-sample.pdfЗагружено:Sharif