Professional Documents
Culture Documents
Hill 1993
Hill 1993
Abstmct-Frequency-dependent distributed transmission line track transfer function for various excitation conditions. For
self and mutual impedances and admittances for a single-track, a low number of rails, the mathematical computation can be
power-rail electrified railway are obtained by experiment and
modeling. Impedance results obtained by numerical calculation carried out in the time domain, but for larger systems with
from an electromagnetic minimum energy field solution using parallel tracks, transformation into the modal domain, utilizing
the finite-elementmethod are compared with analytic impedance track modal propagation constants and characteristic imped-
models based on the Carson-Pollaczek equations for a stratified, ances, may be more appropriate. This approach is convenient
weakly conducting ground and with practical experimental mea- for complex system analysis and can be used, for example,
surements made on a short length of rail track. The running
rail self conductances and the power rail to running rail mutual together with knowledge of the nonlinear dependencies in
capacitances are also evaluated by the finite element method and the matrix elements as functions of current and frequency
compared with experimental measurements. A physical interpre- to model electromagnetic compatibility phenomena such as
tation of the results is given in terms of conductive and induced intermodulation distortion arising from the mixing of power
ground currents caused by finite ground conductivity and relative frequency harmonics with signaling currents [11.
permittivity.
The accurate determination of track impedance and ad-
mittance has received little attention in the literature, most
I. INTRODUCTION traction simulation studies making simplifying assumptions
A realistic model of the electromagnetic properties of rail about rail and ground material properties. Knowledge of the
track is necessary for the analysis and design of electric impedance and admittance is important in several respects.
railway power and signaling systems. Representing the track The impedance is necessary for track circuit design, which
as a multiconductor coupled transmission line (MTL) is a usually considers only the running rail loop [2], and in
feasible approach provided accurate values for the distributed traction current power flow simulation, where simplified linear
self and mutual track impedance and admittance are known. data has been adopted for a variety of applications such as
The numerical values of these equivalent components are, modeling harmonic propagation along the overhead catenary
however, difficult to measure under practical conditions, since [3], predicting ground current distribution [4], and assessing
they are affected by the shape, position, permeability, and telecommunications interference [5]. The self admittances
conductivity of the rails, and the permittivity and conductivity determine the rail potential distribution in high-voltage AC
of the ground and track substructure. The problem addressed in railways which must be kept below specific safety limits [6]
this paper is to model and measure the equivalent MTL circuit and the magnitude of stray traction currents on dc railways
impedances and admittances for a single power-rail electrified [7], [8], where the variability in track self admittance can
rail track, with special reference to the frequency dependency be controlled by supplementary grounding arrangements or
of the components. The task is particularly difficult because by the addition of thyristor-controlled ground-leakage devices
the running rails and power distribution cables are located on [9]. Knowledge of the track mutual admittance is important
or above the ground which acts as a weak and diffuse, but for the evaluation of the transmission characteristics of signals
significant, conductor of electric current, and ground currents along the track, for example in audio frequency track signaling
are generated which affect the values of the track MTL systems where modulated signals travel along the running rails
equivalent components. at minimum attenuation [lo].
Determination of the equivalent components in the track In this paper, existing analytic models for self and mutual
model is important since the track self and mutual impedance impedance are reviewed [11]-[13]. The models are based on
and admittance matrices can then be used to obtain the the theory of the behavior of electrical conductors laid on and
in contact with a weakly conducting ground. To overcome
This work was supported by the U.K. Science and Engineering Research
Council under Research Grant GR/E/58816.
shortcomings in this analytical theory due to variable material
R. J. Hill is with the School of Electrical Engineering, University of Bath, properties, the self and mutual impedances and admittances
Claverton Down, Bath BA2 7AY,U.K. have been modeled using a minimum energy electromagnetic
D. C. Carpenter was with the School of Electrical Engineering, University
of Bath, Claverton Down, Bath BA2 7AY,U.K. He is now with Vector Fields
field solution implemented by a finite element method (FEM)
Ltd., Kidlington, Oxford OX5 lJE, U.K. software package. This was done because of the difficulty of
IEEE Log Number 9206415. experimental verification due to the unavailability of an abso-
0018-9598/93$03.00 0 1993 IEEE
226 IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY,VOL. 42, NO. 2, MAY 1993
lute ground reference point, and the impossibility of making Impedance model Admittance model
differential measurements between points at opposite track
-
211
0
ends because of the creation of substantial detection circuit
loops. Experiments in partial support of the modeling are, how- Rail 2
ever, described which exploit independence of impedance and
Rail 3 0
admittance elements for short lengths of rail track, where the
signal propagation wavelength is much greater than the track
Y
measurement length. The impedance results are interpreted
in terms of physical phenomena governing electromagnetic
fields and currents in the rails, track substructure, and ground, Fig. 1. Coupled transmission line ground return model for track impedance
and admittance.
and the interpretation is supported by a FEM-derived ground
current contour plot. The admittance results validate the FEM
conductance model, and are used to estimate the effective between lines 1and j . Incorporating similar equations for lines
permittivity of the ground as a function of frequency. 2 and 3, and rearranging, the matrix equation
+80
I I
The technique of conductivity measurement as a function of where J is the current density, I is the total current, and A is
depth is well established from geological prospecting [151, the vector potential. The relationships between B (magnetic
and the depth variation may be treated as constant, in uniform flux density), H (magnetic field strength) and A (vector
layers or as a continuous function. In low frequency analyses, potential) are
a two-layer conductivity function has been shown to give B=VxA (11)
good results [16]. More generally, models of self and mutual
and
conductance may be derived from experimentally determined
stratified ground conductivity-depth-frequency relationships B=pH (12)
228 IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY,VOL. 42, NO. 2, MAY 1993
-30 where h is the height of the wire above the conducting plane.
E- - ; ,
- layer 1
Equation (16) contains no length dependency as did (14). If the
where p is the permeability. The internal self inductance is due exp - 2hX
to the field energy within the conductor. For a long, isolated dX. (18)
cylindrical conductor it is [18]
P H/m. where p and u are the equivalent homogeneous ground per-
lii,int = - (13) meability and conductivity. Evaluation of (17) and (18) yields
8lr
The external self inductance is due to the field energy outside a complex impedance whose real part represents additional
the conductor, and for a long isolated conductor is power dissipation in the lossy ground, and whose imaginary
part represents the energy stored in the magnetic field in the
air and ground. The equations have been applied extensively
to calculate the ground-return impedance of power lines. The
where y is the length of the wire and a is the radius (y >> a). assumption of ground homogeneity, however, is dubious, and
The inductance per unit length is thus itself a function of attention has been focused on devising an analytic expression
length. This potential source of difficulty in deciding on correct for external self impedance for the multilayer ground case
data values in a model is tempered by the fact that the [211, [221.
dependency is weak since it is contained within a natural The result even with a two-layer approximation for the
logarithm function. ground model provides a more accurate representation since
The conductor resistance is obtained from the dissipated the bottom layer can be assigned a lower conductivity, ac-
energy within the conductor according to centuating the current density near the surface. A convenient
expression for the ground-return self impedance for a two-
V V
r=-=- (15) layer ground has been given by Bickford et al. [23] as
I J, J . d A
PlPO
where V is the longitudinal voltage causing the current I to
flow and J is the current density over area d A within the
conductor cross section S. In (13) and (15), both internal where
inductance .and resistance are frequency dependent from the
skin effect [19].
When the conductor is brought parallel to a conducting
semi-infinite ground, eddy currents are generated by induction
in the ground whose associated magnetic field modifies the
field surrounding the wire. Return current is constrained to
flow in the ground near the outward path beneath the wire, and
so the circuit impedance is changed. The circuit can be
represented as a conductor with its image. If the ground is 72 = x2 + jWPnP0.n. (22)
a perfect conductor, the image will be located beneath the In (19)-(22), the subscripts 1, 2, 3 refer to air, upper ground
ground at a depth equal to the conductor height. For this layer and lower ground layer respectively, and d is the depth
condition, the external self impedance is that of a parallel of the interface between ground layers.
wire circuit and is given by The model for external self impedance taking account of the
Po 2h ground conductivity must lie within extreme impedance limits
z, = j w l , = j w - In - film (16)
2lr a corresponding to the cases of perfectly insulating and perfectly
HILL AND CARPENTER: RAIL TRACK TRANSMISSION LINE MODELING AND EXPERIMENTAL RESULTS 229
be significant if the line heights above the ground vary, for produce shunt losses along the track, the magnitude of which is
example in the calculation of mutual impedance between a related to the values of the self and mutual conductances. The
rail and an overhead catenary. actual conductive connection of the rails to the ground arises
The application of Carson’s equations to the calculation of from the resistance of rail fastenings to track ties, which in turn
self and mutual impedances in overhead power and communi- are in contact with the underlying ballast and hence the soil.
cations lines has produced much published work. Rosen [26], The ground also has permittivity which determines the
for example, has evaluated tables of mutual inductance for magnitude of quadrature or displacement currents flowing be-
power transmission systems assuming an equivalent single- tween the conductors or to ground. Capacitive currents flowing
layer ground conductivity. For the same conditions, Deri et between the running rails at power and audio frequencies are
al. [27] have introduced the concept of an ideal current normally small compared with conductive currents because
return plane below the surface with an equivalent depth the fastenings to the ties and ground substructure ensure the
determined by the complex wave penetration. Their method establishment of a continuous, dominating ground conduc-
is also applicable to a multilayer ground. tance. However, the power rail self and mutual capacitances
To evaluate the Carson integrals accurately, some assump- may be significant and if so will influence the transmission
tions concerned with the physical nature of the system are line characteristics by contributing to the modal propagation
necessary. Wise [28] has neglected ground relative permeabil- constants and characteristic impedances. For power rails in
ity, displacement currents, and wave attenuation. Olsen and close proximity to the ground, but isolated from it, the large
Pankaskie [29] have compared the exact and Carson theories, bottom physical section of the rail produces significant capac-
examining in detail the associated approximations. Their work, itance effects to ground compared with the direct capacitance
applied to the case of rail track, confirms the validity of the from the power rail to the running rails (Fig. 2). In contrast
Carson equations for the assumptions: to the ground conductivity variation, the determination of
the distance of the observer from the track is much permittivity as a function of depth and frequency is difficult.
less than the freespace wavelength at right angles to the A model of an effective constant ground permittivity does,
wire (true even at higher harmonic frequencies where the however, provide a useful initial approach for the mutual
wavelength is still several hundred meters); capacitance model [30].
the relative magnitude of ground phase constant is much
less than that of the freespace phase constant, i.e.,
Iv. FINITE ELEMENTMODELING
(35)
where A. General Considerations
2a
ko = W ! / ( P O € O ) = - (36) The finite element method is a versatile numerical analytic
x technique in widespread use for the solution of static and
and time-varying electromagnetic field problems. Its application to
4).
IC, = W ! / ( P O € , - JPoU, Re (k,) >0 (37)
problems involving magnetic stored and dissipated energy is
well established, and computation times for systems with two-
or three-dimensional geometrical symmetry are reasonable
typical values being ko = 1.05 x lop6 m-l and k, = using personal computer workstations. FEM is attractive for
1.98 x 1 0 - ~m-l for CT = S/m; the solution of electromagnetic fields around rail track where
the propagation constant is much less than the ground analytic modeling is difficult because the rails and trackbed
phase constant do not have a regular geometric shape, and because the
rails are in close proximity to each other compared with
lYsl Ikgl their individual dimensions. The difficulties in applying the
a typical value for 7sat 50 Hz being 15.7 x m-’. technique to rail track modeling are concerned with optimizing
the computational mesh size to achieve reasonable calculation
The condition that the rail radius is much less than the height
times, and setting the model boundaries to achieve acceptable
above the ground is, however, violated.
accuracy [31].
The theoretical basis of FEM requires a description of the
E. Physical Interpretation of Admittance Model system to be modeled in electromagnetic field terms based
The track transmission line model consists of several paral- on Maxwell’s equations. The equations are manipulated in
lel rails excited with respect to a remote, weakly conducting either an integral or differential formulation and an efficient
ground. From the impedance viewpoint, this is equivalent numerical algorithm is selected for the solution. Elementary
to regarding the system as a set of conductors which inject accounts of FEM theory may be found in [32] and [33], with
current into the ground by induction (eddy currents) and a more detailed exposition in specialist texts such as [34]
conduction (conductive currents). If the conductors are at and [35]. The package used in the present modeling uses a
different potential with respect to each other or absolute differential formulation of the field equations, and following
ground, currents will also flow between them or to ground. is a summary together with interpretive information in terms
The in-phase or conductive components of these currents will of the fields around rail track.
HILL AND CARPENTER RAIL TRACK TRANSMISSION LINE MODELING AND EXPERIMENTAL RESULTS 23 1
232 IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 42, NO. 2, MAY 1993
@)
Fig. 4. Finite element method elemental nets. (a) Running rail. @) Track and ground substructure.
After formation in outline form, the mathematical model is C. Application of FEM to Rail Track Modeling
discretized into a numerical model by dividing each region In the case of rail track with its substructure, the problem
into a mesh linked by nodes. Qpical elemental nets for space is characterized by separate regions as follows:
an individual rail and for the complete problem space are
ground, with low conductivity which is variable with
shown in Fig. 4. The number of mesh elements that can be
depth, zero permeability and significant permittivity which
used is limited by the computational processing time, about
5000 elements being required for most problems covering may also be variable with depth;
the complete space. At the time of publication, processing rails, which have moderate conductivity, high nonlinear
power is continually increasing, with a concomitant increase permeability and permittivity equal to that of free space;
in possible model complexity. To improve the model accuracy, air, with zero conductivity and permeability and permit-
more elements are necessary in regions of rapidly changing tivity equal to that of free space.
fields, since the solution error is proportional to h2 where h is To simplify the problem, ferrous hysteresis in the rail iron is
the largest dimension of the element. A further variable is the ignored, although saturation is permitted through a predefined
interpolation methocLa quadratic rather than linear technique B-H relationship.
will improve the accuracy, albeit at the expense of computation After the problem region is divided into the elemental
time. net and the boundary conditions are defined, the numerical
-
HILL AND CARPENTER: RAIL TRACK TRANSMISSION LINE: MODELING AND EXPERIMENTAL. RESULTS 233
solution of (50) determines values for the vector potential ‘ I L b : PN3F1000 I /Y€W
within every element for the minimum system energy state. MIlS 1 L 2 Bath Unlv
and
Fig. 6 . Finite element method derived ground current density profiles for running rail excitation.
are integrated to find the system impedances, the impedance static conductance problems is found by exploiting aspects
values will be insensitive to rail geometry if the problem area of electromagnetic-electrostaticduality given in Table 11. With
is large. appropriate scaling, it is possible to map A to V and l / p
to (T. The electromagnetic stored energy J B . HdS is then
E. Track Admittance Modeling equivalent to the dissipated energy J E .JdS. In the air region,
Conductance, The self conductance of a rail in continuous (T = 0 implies l / p = 0 or p -+00, so air must be modeled by
contact with the ground may be obtained by integrating the removing all elements from the region. For the dynamic case
current flow from the rail to remote ground. It may be where aA/at # 0, a more complex duality relationship based
represented by on (50) must be used. For the present conditions, this would
be required only for initially quiescent conditions during the
settling down period.
To solve the self conductance problem, a field calculation
where l i e is the total current injected into the ground per was performed by FEM to obtain the minimum energy con-
unit rail length, V, is the rail voltage, and Jie is the current dition with flow of current from one rail to remote ground. A
density in the ground over surface dS. Similarly, the mutual conventional top-contact power-rail electrified single rail track
conductance may be determined by integrating the flow of with the dimensions of Fig. 2 was modeled, and values of con-
current between two parallel rails, represented by ductance corresponding to tie, ballast (0.65 m) and four ground
layers (0.65, 2, 4,and 22 m) were estimated from previously
(55) measured conductivity-depth-frequency profiles [16]. The in-
tegral of (54) converges because the conductivity decreases
where Iij is the total injected current per unit length, V,, is the with depth. A sample equipotential contour for this condition
rail-rail potential difference and Jij is the current density in the is shown in Fig. 7(a). The contour interval has again been
ground over surface dS.In (54) and ( 5 9 , the current density normalized assuming a 1-V rail excitation.
depends on the local values of conductivity and electric field. The mutual conductance was found by considering current
The conductivity is a function of both depth and frequency flow between two rails, and the corresponding equipotential
and the electric field is a function of depth. is given in Fig. 7(b) (only half is required due to symmetry).
FEiM cannot be used to solve these equations directly. In both the self and mutual cases, the conductance was found
Instead, the electromagnetic fields are calculated in the various from the dissipated energy as gV2.
regions of the system being modelled, then integrated to Capacitance. For displacement current, the dual between the
find the corresponding dissipated energy. The solution for electromagnetic and electrostatic problems is required. Table
~
HILL AND CARPENTER RAIL TRACK TRANSMISSION LINE MODELING AND EXPERIMENTAL RESULTS 235
FILE : COND250000
mWMCE: RAILS 2 b 3
M€GA ILE : ONERUP6000 1 A€
Bath Univ
Fig. 7. Finite element method derived equipotentials for conductance Fig. 8. Finite element method derived equipotentials for capacitance deter-
determination. (a) Single-rail excitation. @) %o-rail excitation. mination. (a) Two-rail excitation including power rail. @) Excitation between
running trails.
I1 shows that the appropriate mapping is V to A and E to
l/p. The electromagnetic field stored energy J B . H d S is significant differences in the capacitance between the power
then equivalent to the electrostatic stored energy E . DdS rail to either running rail, and to both running rails in parallel.
through the relationship The results clearly show crowding of the equipotentials in
the air between the power rail and the ground surface, the
/E.DdS=/VV.cVVdS (56) ground being sufficiently conductive to prevent charge storage.
Thus the power rail shape and its ground proximity alone
which is equivalent to the stored electrostatic energy of 0.5
determine the mutual capacitance between the power rail and
cv2. the other rails, and the inter-rail spacings have no effect. The
Since no information was available about the variation of
results also show an order of magnitude difference in the
permittivity with depth, a constant effective permittivity for the
mutual capacitances between the two running rails, which are
tie, ballast and ground structure was assumed. To obtain the
in good conductive contact with the ground compared with
mutual capacitances between the rails, FEM field calculations
those involving the power rail.
were performed to solve for the minimum energy condition
when each pair of the three rails was excited in turn. Because
of the perceived dominance of ground effects, an analysis was v. EXPERIMENTAL MEASUREMENTSOF
TRACK IMPEDANCE AND h M l T T A N C E
also carried out with an excitation voltage applied between the
power rail and the running rails connected in parallel.
Examples of the resulting FEM equipotential plots are A. Impedance Measurements
shown in Figs. 8(a) and (b) and confirm that there were no Since the track impedance is much less than the track
236 IEEE "RWSACTIONS ON VEHICULAR TECHNOLOGY, VOL 42, NO. 2, MAY 1993
Rail 2
;G T ? t
Is *L
(57)
H I U AND CARF'ENTER RAIL TRACK TRANSMISSION W N E MODELlNG AND EWERMENTAL RESULTS 237
%GC
Rail 3 (Running)
-
VI
1 :
$ - 8 :
5
Qa
.- 100f
:
c
3 10:
u -
2 : s :
10 r 1:
-
1: 0.1 F
0.1 : 0.01 :
'
/ ' ' ""'
0.001 0 1 # * * . * * * """" """"
0.01 10 100 lk 10k 1 OOk 10 100 lk 1Ok 1ook
Frequency (W Fresuency(W
0.1
10 100 lk 10k 1OOk
(W
Fig. 15. Validation of finite element method capacitance models.
Frequency (W
VII. DISCUSSION
(b)
The sample calculations of self and mutual impedance at
Fig. 13. Track mutual impedance results from experiment, finite element
method, and analytic theory. (a) Mutual resistance. @) Mutual reactance. 50 HZ given in I show that the difference
single-layer and two-layer stratification models of ground
A comparison of the FEM and experimental results for conductivity is small. The sensitivity of the results in general
capacitance is given in Fig. 15. Because of the constant ground to changes in ground conductivity is thus also small. The
HILL AND CARPENTER: RAIL TRACK TRANSMISSION LINE MODELING AND EXPERIMENTAL RESULTS 239
to running rail capacitance. The unavailability of accurate J. Holtz and H.-J. Klein, “The propagation of harmonic currents gen-
ground permittivity data was of little consequence because erated by inverter-fed locomotives in the distributed overhead supply
system,” IEEE Trans. Power Electron., vol. 4, pp. 168-174, Apr. 1989.
the capacitance is dominated by the rail-ground air gap rather W. Machczynski, “Currents and potentials in earth-return circuits ex-
than the exact value of ground permittivity. The experimental posed to alternating current electric railways,” IEE Proc., vol. 129B,
results for self capacitance and susceptance could not be no. 5, pp. 279-288, Sept. 1982.
A. Rosen, “Interferencein railway lineside telephone cable circuits from
checked by FEM using the constant ground permittivity model, 25 kV 50 Hz traction systems,” Institution of Railway Signal Engineers
because the result would be dependent on the assumed depth pro^., pp. 55-79, 1958.
S. D. Jacimovic, “Maximum permissible values of step and touch
of the ground plane, rather than on the value of permittivity voltages with special consideration to electrified railroads,” IEEE Trans.
chosen. A decreasing permittivity profile with depth would Ind. Appl., VOI. IA-20, pp. 935-941, JulylAUg. 1984.
have been required for the self capacitance to be independent R. E. Shaffer, “Design of DC powered rail transit systems to minimize
stray currents,” Materials Performance, vol. 21, no. 9, pp. 17-22, Sept.
of the position of the ground plane in the model. 1982.
P. Vernon, “Stray-current corrosion control in metros,” Institutwn of
Civil Engineers Proc., vol. 80, pt. 1, pp. 641-650, June 1986.
VIII. CONCLUDING REMARKS T. A. Kneschke, “Design considerations for the DART traction electri-
fication system,” in Proc. 1987 IEEEIASME Joint Railroad Conference,
Analytical models for rail track self and mutual impedance Toronto, Canada, April 21-23, 1987, New York: IEEE, pp. 1-11.
B. Mellitt, “Data-transmission characteristicsof railway track,” Electron
based on the established Carson-Pollaczek theory have been Lett., vol. 9, no. 23, pp, 550-551, Nov. 1973.
evaluated using ground conductivity data obtained from local J. R. Carson, “Wave propagation in overhead wires with ground return,”
measurements of ground conductance. Both single-layer and Bell Syst. Tech. J., vol. 5 , pp. 539-56, Oct. 1926.
F. Pollaczek, “Uber das Feld einer unendlich langen wechselstrom-
two-layer conductivity data have been used. The finite element durchflossenen Einfachleitung,” E.N.T., vol. 3, no. 9, pp. 339-359,
method has been applied to establish the validity of the 1926.
analytical models, and a practical method has been described G. Haberland, “Theorie der Leitung von Wechselstrom durch die
Erde,” Zeitschrifi f i r Angewandte Mathematik und Mechanik, vol. 6,
for the measurement of the mutual impedance between the pp. 366-379, 1926.
individual rails in an electrified rail track. The experimental R. J. Hill and D. C. Carpenter, “Determination of rail internal impedance
for electric railway traction system simulation,” IEE Proc.-B, vol. 138,
technique relies on the separability of the track impedance and no. 6, pp. 311-321, Nov. 1991.
admittance for short track lengths. Due to the inaccessibility W. M. Telford, L. P. Geldart, R. E. Sheriff, and D. A. Keys, Applied
of the remote ground, the method requires prior knowledge Geophysics. Cambridge U K CUP 1976.
R. J. Hill, D. C. Carpenter, and T. Tasar, “Railway track admit-
of the rail self impedance, and this is obtained using the tance, earth-leakage effects and track circuit operation,” in Proc. 1989
FEM model. Self and mutual impedance results are given IEEEIASME Joint Railroad Conf., Philadelphia, PA, April 25-27, 1989,
for a frequency range between 25 Hz and 40 kHz. Good New York IEEE, pp. 55-62.
D. C. Carpenter and R. J. Hill, “A continuous-function ground conduc-
agreement is obtained between all the methods, and the results tivity model for the determination of electric railway earth conductance,”
are interpreted physically utilizing the concept of induced and IEEE Trans. Geosci Remote Sensing, to be published.
conductive currents in the weakly conducting ground. F. W. Grover, Inductance Calculatwns. New York Dover 1973.
R. J. Hill and D. C. Carpenter, “Modelling of nonlinear rail impedance
The self and mutual conductances and capacitances have in AC traction power systems,” presented at the Int. Conf. Harmonics
been obtained by a combination of field modeling using in Power Systems: ICHPS N,Budapest, Hungary, Oct. 4-6 1990, pp.
268-274.
FEM and practical tests on the same short track length. F. C. Alvarado and R. Betancourt, “An accurate closed-form approxi-
The model of ground conductivity used as data in the FEM mation for ground retum impedance,” Proc. IEEE, vol. 71, pp. 279-80,
model has a multi-layer variation with depth and frequency. Feb. 1983.
M. Nakagawa and K. Iwamoto, “Earth-return impedance for the multi-
Good agreement is obtained between FEM and experimental layer case,’’ IEEE Trans. Power Apparatus Syst., vol. PAS-95, pp.
results for the running rail self and mutual conductances. 671-4576, MarJApr. 1976.
The assumption of an equivalent constant permittivity with A. Ametani, “Stratified earth effects on wave propagation - frequency
dependent parameters,” in Proc. IEEE PES Wnter Meeting, New York,
depth and frequency is sufficient to model the power rail to Jan. 27-Feb. 1, 1974, New York IEEE, pp. 1233-1239.
running rail mutual capacitances to a high degree of accuracy. J. P. Bickford, N. Mullineux, and J. R. Reed, Computation of Power
However, exact knowledge of the permittivity depth and System Transients. London, U.K.: Peter Peregrinus, 1976.
M. R. G . Klewe, Interference Between Power Systems and Telecommu-
frequency dependencies would be required for more accurate nications Lines. London, U.K.: Edward h o l d , 1958.
modeling of the self capacitance of the power rail. E. D. Sunde, Earth Conduction Effects in Transmission System. New
York: Dover, 1968.
A. Rosen, “Formulas and tables for mutual inductance between pardel
circuits with earth retum,” Proc. IEE, vol. 118, no. 9, pp. 1227-1236,
ACKNOWLEDGMENT Sept. 1971.
The authors acknowledge Prof. D. Rodger and Dr. P. A. Den, G. Tevan, A. Semlyen, and A. Castanheira, “The complex
ground return plane. A simplified model for homogeneous and multi-
Leonard at the University of Bath, England, for the use of layer earth return,” IEEE Trans. Power Apparatus Sys&, vol. PAS-100,
the FEM package MEGA. pp. 3686-93, Aug. 1981.
W. H. Wise, “Propagation of high-frequency currents in ground return
circuits,” Pmc. IRE,vol. 22, no. 4, pp. 522-527, Apr. 1934.
REFERENCES R. G. Olsen and T. A. Pankaskie, “On the exact, Carson and image
theories for wires at or above the earth’s interface,” IEEE Trans.Power
R. J. Hill, S. L. Yu, and N. J. Dunn, “Rail transit chopper traction Apparatus Syst., vol. PAS-102, pp. 769-778, Apr. 1983.
interference modelling using the SPICE circuit simulation package,” R. J. Mayhan and R, E. Bailey, “An indirect measurement of the effec-
IEEE Trans. Veh. TechnoL, vol. 38, pp. 237-246, Nov. 1989. tive dielectric constant and loss tangent of typical concrete roadways,”
R. J. Hill, “Train position detection and track-train data transmission IEEE Trans.Antennas Propagat., vol. AP-23, pp. 565-569, July 1975.
using audio frequency track circuits,” J. Electrical and Electronics D. C. Carpenter and R. J. Hill, “The effect of magnetic saturation,
Engineering Australia, vol. 5 , no. 4, pp. 267-277, Dec. 1985. hysteresis and eddy currents on rail track impedance,” in Pmc. 1989
HILL AND CARPENTER:RAIL TRACK TRANSMISSION LINE: MODELING AND EXPERIMENTAL RESULTS 241
IEEEIASME Joint Railroad Conj, Philadelphia, PA, April 25-27, 1989, David C. Carpenter was bom in England on
New York IEEE, pp. 73-79. September 8, 1956. He received the B.Sc. degree
[32] M. N. 0. Sadiku, “A simple introduction to finite element analysis from the University of Southampton, England,
of electromagnetic problems,” IEEE Trans. Education, vol. E-32, pp. in 1978 and the M.Phil. degree from Coventry
85-93, May 1989. Polytechnic, England in 1989.
[33] M. N. 0. S a d h , “A further introduction to finite element analysis In 1979 he hecame a Computer-Aided Design
of electromagnetic problems,” IEEE Trans. Education, vol. E-34, pp. Engineer at GEC Machines Ltd, Rugby, England,
322-329, NOV.1991. where he was involved in the design of large
[34] J. F. Eastham and D. Rodger, “Differential methods, finite elements and induction motors. In 1982, he became a Lecturer
applications,” in Industrial Applications of Electromagnetic Computer (Senior Lecturer from 1984) in Electrical Engineer-
Codes, Y . R. Crutzen, G. Molinari, and G. Rubinacci, a s . Dordrecht, ing at Coventry Polytechnic, England. From 1986
The Netherlands: Kluwer, 1990, pp. 51-79. to 1987 he was a Visiting Professor at Lalcehead University, Canada, where
[35] P. P. Silvester and R. L. Ferrari, FiniteEZementsforEZecrricaZEngineers. he extended his research interests to intersystem satellite communications
Cambridge, U K CUP, 1983. interference analysis. Between 1988 and 1991 he has been a Research O a c e r
[36] D. C. Carpenter and R. J. Hill, “Rail impedance modelling for DC- at the University of Bath, England, working on the modeling of electric
fed railway traction simulation,” in Proc. M T E D Int. Symp. Applied traction systems, with special reference to the development of simulation
Simulation and ModeZZing, Lugano, Switzerland, June 18-20, 1990, techniques for EMC. Mr. Carpenter is now an Applications Engineer at Vector
Anaheim, Ck. Acta Press, pp. 105-108. Fields Ltd., Oxford, England, responsible for developments in electromagnetic
field computation with finite-element methods.
Mr.Carpenter is a Member of the IEE (UK).