ABSTRACT
1
- TU)'
[ h2 -
12):-s(
~
1
+ - ( s + w ) 4 ~ n ( s + w ) - -s41ns
- -In +
[(s - ? u ) ~ h 2 ] Equation (6) can therefore be described as follows: The
term v m c ~ . . . Dam!) in the numerator of
equation (6) represents the GMD of all the round con-
12 6
ductors constituting conductor 2, to conductor a , which
+-1 (S - w ) In(s
12
~ - w)
4
+
-h(s
3
+ h
tan-' -
S+W
is part of conductor 1, and this term therefore represents
4 h the total effect of conductor 2 on the round conductor a.
3
+
+ - h 3 (s w) tan-' -+ - 8hs3 tan-' -
h 3 S There are n round conductors constituting conductor 1,
s 4 h and there are therefore n of these distance. The geometric
-zh3stan-1-+-h(s+w)3tan-1 __
3 h 3 s-w mean of these distances (the nth root of the product of
4 s - 1u
- + -hA (s - U ) tan-' - - L5w2h2 (4) these distances) represents the GMD between conductors
3 h
1 and 2.
w = 2 cm, h = 10 cm, s = 10 cm
+
Dll = 0, %235(w h ) = 26,8 mm
Ll = 276,9 nH.m-'
Each rectangular conductor in Figure 2 has been replaced
by a number of identical round conductors arranged in
such a way that they resemble the shape of the original A program written in MathCad and implementing the dis-
conductor. Each of the thin round conductors constituting cretization algorithm yielded the results shown in Table
conductor 1 is denoted by a letter Q to n and each of those 1.
constituting conductor 2 is denoted by a letter U' to m'. A finite element analysis package was also used to cal-
The GMD between two round conductors is simply the culate the inductances L1 and La, and values of 272,2
distance between their center points (denoted by Dab', for nH.m-' and 272,3 H.m-l were obtained.
Am?
2/’
_..._.. ._._._ ..... .-._-..
_............_...... .--.....-.-__-
~
ment analysis
it is as flexible as more expensive methods in terms
of conductor shapes and arrangements that can be
dealt with.
The main limitation of the use of the GILZD is that it is re-
stricted to direct-current conditions where the current is
distributed uniformly across the conductor cross-section.
or cases where the current distribution is known exactly
(for example, at very high frequencies where current only
flows on the conductor surfaces, or when the conductors
are so thin that the skin effect, does not influence the cur-
rent distribution noticeably).
Fig. 6. Iterative Process to find Optimal Foil Shape by Simulation
CONCLUSIONS
Lines of constant magnetic vector potential are, in fact, The Geometric Mean Distance as a tool for analysis and
flux lines, and these lines should not cross the foil cross- design has long been discarded in favor of Finite-Element
sections if all filaments are to link the same amount of Methods (FEM), but in certain cases it can still be used
flux, and hence have equal inductances. to great advantage to yield approximate results more
Initially the two flat foils separated by a certain distance, quickly and cheaply than finite-element methods.
would be simulated and the calculated direct-current flux
lines plotted. REFERENCES
The user would then have to alter the geometry in such [I] T. J. Higgins. “Theory and Application of Com-
a way that the foils do not cross any direct-current flux plex Logarithms and Geometrical Mean Distances”,
lines, and the process would be repeated. Trans. AIEE, vol. 66, pp. 12- 16, 1947.
Once the foil shapes need not be changed any more to PI Copper Development Association,
prevent flux lines from crossing the foil surfaces, the op- Copper for Busbars. Potters Bar, Hertfordshire:
timal shapes have been determined. Copper Development Association, ch. 9, pp. 33-
36, 1984.
The first two steps in the process are shown in Figure 6. 131 P. J. H. Rata and A. G. Thomas,
Application to Busbars Aluminium Busbar. London: Northern Aluminium
CO. (Ltd.), ch. 5, pp. 37- 40, 1960.
The foil shapes arrived at by means of the above method, [4] W. D. Stephenson, Jr.. Elements
are usually not easily manufacturable, but there are some of Power System Analysis New York: McGraw-Hill
practical busbar shapes that closely resemble the calcu- Book CO.. ch. 3, pp. 52- 55, 1982.
lated shapes for various inter-conductor distances. Figure [5] A. H. M. Arnold, “The Inductance of Linear Con-
7 [6] shows some theoretical foil shapes and the closest ductors of Rectangular Section”, JIEE, vol. TO, pp.
corresponding practical busbar shapes. ,579- 586. 1932.
[6] A. H. M. Arnold, “The Transmission of Alternating-
ADVANTAGES AND LIMITATIONS OF THE Current Power with small Eddy-Current Losses”,
GMD JIEE. vol. 80. pp. 395- 400, 1937.
The advantages of using the G I I D for inductance calcu-
lations are that in some cases: