IEEE TRANSACTIONS ON MAGNETICS, VOL. 32, NO.

5, SEPTEMBER 1996 4281

Simplified Magnetic Fie1.d Modeling and

Calculation of Large Air-Core Reactor Coils
&in Yu and Stephen ,4.Sebo
Department of Electrical Engineering, The Ohio State University
Columbus, Ohio 43210, U.S.A.

Abstract-A simplified magnetic field model for large mericd methods. When field points are inside or in the
air-core reactor coils is proposed. It uses three flexi- very close vicinity of reactor coils, some accurate modeling
ble magnetic dipoles with their numbers of turns and method should be employed which takes into account the
positions flexible. The magnetic flux density was cal- shape of the coils and/or conductors. Beyond that range,
culated along 5 typical traverses for three different it is necessary to develop a simplified modeling method
types of coils. At designated field points, close to the instead of using the accurate ones in order to speed up
coil, the calculation error of the proposed model is less the computation, and to save computation time.
than 5%. The three flexible dipole model proposed is In this paper, a model with three flexible magnetic
not only simple and easy to be implemented but also dipoles is proposed and the accuracy of its calculation
gives satisfactory accuracy in the regions of practical of the three dimensional power frequency magnetic flux
interest. It can be applied easily to the calculation of density is analyzed.
the three dimensional power frequency magnetic field Based on typical length-to-diameter ratios ( H I D m ) ,the
distribution of substations. following three different types of coils are used in this
study:
I. INTRODUCTION 0 short coil with HID, = 0.5,
Public concern related to electromagnetic field issues 0 medium length coil with H I D , = 0.9,
necessitates the investigation of 60 Hz magnetic field dis- 0 hng coil with H / D m = 2.0.
tribution in the vicinity of large air-core power reactor Five tjpical traverses are used for the calculations, repre-
coils. It is a practical and important problem due to the sented by abbreviations ATC, ATS, LTT, LTH and LTC
fact that the normal operating currents of current-limiting (refer to Fig. l(a)),respectively, which mean:
reactors and line trap reactors are the same as the line
currents which can be up to several thousand amperes. 0 ATC: axial traverse, center,
Also, large coils are employed in the filter circuits of high 0 ATS: axial traverse, side,
voltage DC converter stations, though their currents are 0 LIT: lateral traverse, top,
smaller than the line currents. 0 LTC: lateral traverse, center,
These 60 Hz air-core reactor coils are important com-
0 LTH: lateral traverse, halfway between LTT and
ponents of the substations of electric power transmission
LTC.
systems. The mean diameter of a typical reactor coil is
about 1 m, the length of the coil is usually in the range of The model developed not only simplifies the calculation
0.5 to 2.0 m. The coil is mounted either in the horizon- greatly but also gives accurate results even as close to
tal or in the vertical position. The clearance between a the reactor coil as at a minimum distance of ( O . . . 1.3)H
reactor coil and the ground plane can be as low as 0.5 m. axially or (0.2. 1.2)0, laterally from the outer surface
Therefore, the contributions of these large coils to the of the coil, pending on the location of the traverse and
electromagnetic environment of a substation at or close the type of the coil (see Table I).
to ground level are significant and essential to be known.
Many accurate calculation methods for evaluating the 11. T H R E E FLEXIBLE DIPOLE MODEL
magnetic flux density of air-core coils have been devel-
oped. They include analytical, semi-analytical arid nu- The simplest (but inadequate) model of a reactor coil is
a single coaxial ,magnetic dipode placed at the center of the
Manuscript received March l , 1996. coil, as $shownin Fig. l ( b ) (referred to as SINGLE model).
Q.Yu, e-mail yuqCQer4.eng.ohio-state.edu,phone 614-764-4628;S. It is assumed that the dipole current is equal to the actual
A. Sebo, e-mail sebo@ee.eng.ohio-state.edu, phone 614-292-7410. coil current I times the total number of actual tarns of
This work was sponsored by the Transmission Substations Pro-
gram of the Electrical System Division of Electric Power Research the reactor coil, N . The diameter of the dipole is D,.
Institute and was part of EPRI RP2943-3. Large el-rorsare generated when field points are relatively

0018-9464/96$05.00 0 1996 IEEE

close to the coil and the coil represented is in the long9
category. This is due to the fact that a single current
loop can not and does not reflect the geometry changes of
the coil. Multidipole models must be used to reflect the
geometry shape of the coil. H
The use of three coaxial magnetic dipoles of fixed turns Point
and positions to model a reactor coil can enforce the shape
of the coil. As shown in Fig. l(c), where this three fixed
(a) Reactor coil (b) SINGLE model
magnetzc dapole model is referred to as 3FIXED model,
there are always two dipoles with N/12 turns each at the
-i n1 = N112
top and bottom of the coil to model the longitudinal shape
of the coil, and a third dipole always with 5 N / 6 turns HI2 n, = N x h , W

placed at the center of the coil. The 3FIXED model can H n2 = Nx5I6
nz = N x h /H
=0.8n Dm
improve the calculation accuracy slightly for the long HI^ e Field Point
coil, but the results are still not satisfactory, especially Field Point n3 r N x h l
1 3 = N112
for the field points along the lateral traverses.
Studying the calculation accuracy of the single dipole (c) 3FIXED model (d) FLEXIBLE model
model it was found that its calculation accuracy can be
Fig. 1. Reactor coil and its models.
greatly improved if the coil being modeled is a medium
length coil, as opposed to a long coil. That suggests B W h e n the length of the coal 2s shorter t h a n O.SD,,
that if a long coil is divided into several sections which the three dipole model becomes a one dipole model
are of medium length and each section is modeled by automatically.
a dipole, the overall accuracy will be improved. Three
e W h e n the field poant as zn the axaal areu, i.e., above
dipoles usually are sufficient due to the fact that the prac- the top or below the bottom of the coil, the coil is
tical H / D , values are limited. More dipoles can be used
divided into three sections of equal length. Hence,
if an extremely long coil is being studied. Thus, a three each dipole has N / 3 turns.
flexible dapole model is proposed as shown in Fig. l(d)
(referred to as FLEXIBLE model), where a coil is divided The magnetic moment vector of a magnetic dipole is
into three sectzons and a magnetic dipole is located at the r7i = N I n D m 2 i / 4Am2, where I is in A, the normal di-
center of each section, that is: rection of the coil is assumed to be pointing in the 2 direc-
tion given by the right-hand-rule, D, = [(Df DlD2 + +
The center section has a length of 0.80,. It repre- D,)/3]/2is in m and D1 and Dz are the inside and out-
sents the part of the actual coil that is closest to the side diameters of the coil, also in m.
field point. It is as symmetric about the position of In rectangular coordinates, the magnetic flux density
the field point in the axial direction as possible. Its components at a far zone field point ( x ,y, z ) produced by
number of turns is N0.80, f H . a magnetic dipole whose center is located at (xc,yc, zC)
B The top section is the section between the top of can be calculated by the following equations [l]:
the center section and the top of the actual coil. Its
length is h l , its number of turns is N h l f H .
B The bottom section is the section between the bot-
tom of the center section and the bottom of the actual
coil. Its length is h2, its number of turns is NhZIH.

Comments related to the use of the three flexible dipole

partial list of rules):
model are as follows (this is only a
where R = [(x - zc) $. (y + -
- yC) ( z z , ) ~ ] / ~is the
distance between ( I L , y, y,, zc);the distances
z) and (zC,
and coordinates are in m; and the magnetic flux density
e In general, there exists the relationship 0 . 8 0 , + hl + components B x , Y , are
Z in teslas.
hz = H . The above equations are based on the far zone field
W h e n the field poznt as a t the lateral sade and ats uxaal assumption. In other words, the above equations are ac-
dzstance f r o m the top or bottom 0.f the actual coal as curate only if the radius of the coil is much smaller than
less t h a n 0.80,, the three dipoles are reduced to two the distance between the field point and the center of the
because the length of the top or bottom section ( h l , dipole. However, for the FLEXIBLE model, this is not
or h2) is reduced to zero. Only two sections of the the case when a field point is close to the center of the
length of 0.80, and H - 0.80,, respectively, are top or bottom surface of the coil and the length of the
defined. coil is lcss than the radius of the coil.
 4283

In order to improve the accuracy of the magnetic 0ux

50
density calculations in the areas above or below the end
surfaces, a correction factor, F, for the B, expression in SINGLE * H / D , ==0 05 5
HID.,, -
(1) is introduced if a vertical coil is considered. F takes HID, = 0 9 -
H I/ D ,, ==220 0
into account the difference between the exact solution and
the far-zone solution:
35 3FIXED + H / D , ==0 05 5 -
HID,
o HID, = 0 9
30
A x HID, = 2 0

F is to be used for field points which are on the axial side

E s3 FLEXIBLE 53 H I D , = 0 5
M3 HID, = 0 9
L3 HID, = 2 0
-

15
and close to the surface of the coil, where B, dominates
other components when (Dm/2)> R. For field points on 10

the lateral side, F can be ignored due t o the fact that the 5
distance between a field point and the center of the dipole
is always larger than the radius of the dipole. 0 I 2 3 4 5 6 7 8 9 10

Distance from outer surface of coil, in multiples of coil mean radius, DJ2.

111. RESULTS AND CONCLUSIONS

Fig. 3. Errors of three simplified models along LTH traverse.
The magnetic 0ux density has been calculated along
5 traverses, ATC, ATS, LTT, LTH and LTC for three
different types of coils (short, medium length, long) by
using the three simplified models SINGLE ,3FIXED and
FLEXIBLE outlined above, respectively, and compared Lateral Traverses
Axial Traverses
with that of a code, called COILB which uses a cur-
rent loop stack model. In the current loop stack model,
a coil is considered as a stack of series coaxial current-
carrying loops. The magnetic 0ux density of a coil is
calculated by the integration of the magnetic flux density
of each current-carrying loop. The results along different
traverses have been compared with those of a proprietary
The discrepancies of the resultant magnetic flux density
computer program of a major coil manufacturer for a sin- of the three simplified models vs. the normalized distances
gle coil, then two stacked coils, and a set of three-phase
away from the outer surface of the coil are calculated and
stacked coils. Very good agreement has been found be-
plotted. Two typical ones of them are shown in Figs. 2
tween COILBs and the manufacturers results.
and 3, respectively. The minimum distances away from
the surface of the coil t o limit the discrepancies to 5% for
the FLEXIBLE model are summarized in Table I. It can
be seen that at field points whose distance from the coil
is at least ( 0 . . . 1.3)H axially or (0.2 1.2)0, laterally
from the outer surface of the coil, pending on the location
of the traverse and the type of the coil, the calculation
error of the FLEXIBLE model is less than 5%. That gives
very satisfactory calculation accuracy.
The three flexible dipole model proposed is useful for
the fast evaluation of the three dimensional power fre-
quency magnetic field distribution of air-core coils. The
model can be applied easily t o the calculation of the three
dimensional power frequency magnetic field strength or
magnetic flux density distribution of substations. The
model developed is very efficient in terms of CPU time.
0 I 2 3 4 5 6 7 8
Distance from top surface of coil, in multiple of coil height, H.
REFERENCES

[l] R. Plonsey and R. E. Collins, P r i n c i p l e s and A p p l i c a t i o n s of

Fig. 2. Errors of three simplified models along ATC traverse. E l e c t r o m a g n e t i c Fields , McGraw-Hill, Inc., 1961.