Destructive Arcing of Insulated Joints in
DC Electrified Railway
Brandon Swartley, P.E.
SYSTRA Consulting, Inc.
Philadelphia, PA
Dave Male, P.E.
SYSTRA Engineering, Inc.
New York, NY
Al Santini, P.E.
Metro-North Railroad
New York, NY
Bob Jaccino, E.I.T.
Long Island Rail Road
Hollis, NY
deliveries of M7 starting in 2002. Similarly, the new
MNR M8s (used for testing) that were part of this study
demonstrate the same arcing.
Visual inspections during testing have shown that
large transient electrical arcs are observed across the IJs
from passing trains. While arcing is seen when some or
many of the axles pass over the IJs in the middle of
crossovers, the largest arc is typically observed when the
last axle passes over the last IJ. Arcing has also been
observed at locations on LIRR that are in straight track
and have impedance bonds (IBs) for traction current
by‐pass around IJs. This problem has not been observed
on MNR.
SUMMARY
Metro-North Railroad (MNR) and Long Island Rail
Road (LIRR) have been experiencing excessive
destructive arcing from electric multiple unit (EMU)
trains passing over the insulated joints (IJs) at various
locations in their third rail electrified territory. The
destructive arcing has greatly reduced the life-cycle of
insulated joints at these locations. The Railroads have
undertaken a study of the phenomena to ascertain the
cause(s) and develop possible mitigation of the
destructive arcing [1]. The study followed a systems
engineering approach, analyzing the elements of the
traction system, vehicle propulsion circuits, and signaling
system that utilize the track running rails as electrical
elements of their circuits. The study included research,
testing, analysis, and development of possible mitigations
for the Railroads to consider going forward to address the
problem. This paper will provide background into the
problem, a brief discussion of arcing theory, overview of
the testing performed, summary of the findings from the
testing, and overview of the potential mitigations
developed for the railroads to assess in the stage of
resolving the problem.
1.0
Figure 1. Last train axle crossing IJ with no Z-bond and a resulting
voltage of 291 V across the IJ
INTRODUCTION
Metro-North Railroad (MNR) and Long Island Rail
Road (LIRR) (referred to in this report as “the Railroads”)
have been experiencing excessive destructive arcing from
train wheels of electric multiple unit (EMU) trains passing
over the insulated joints (IJs) at various locations in the
third-rail electrified territory. It has been observed that
the arcing across the IJs and the degradation of the
insulating material have increased since the introduction
of the M7 multiple unit cars on both Railroads, with
The current from the train’s propulsion system is
returned through the wheels and into the rails back to the
rectifier substation. The action of a train passing over the
IJs in crossovers between two tracks results in the train’s
return current utilizing the running rails of both tracks
while straddling the IJs. When the last axle of a train
crosses the IJs, the return current through one of those
tracks is quickly broken – like a switch opening under
1
Destructive Arcing of Insulated Joints in DC Electrified Railway
IJs constrain signal current to the rail sections
between adjacent sets of IJs or blocks.
load. The arc is created by the current being drawn out
between the rail and the wheel, inducing a voltage across
the IJ due to the rate of change in current in the rails
(from a large value to zero) and proportional to the
inductive characteristics of the return circuit, including
the combination of all wayside and onboard inductances.
Aside from the inductive characteristics of the traction
return system elements, including impedance bonds and
running rails, two inductors are part of the train/traction
system circuits that are the subject of this study: the M7
propulsion circuits and the traction system itself.
M7 propulsion circuits (as well as those of M8s and
the future M9s) are significantly different in electrical
makeup and characteristics than earlier EMU propulsion
circuits. In addition to higher train performance requiring
more electrical current from the traction system, the ac
drive equipment includes an inductive-resistive-capacitive
(LRC) circuit for EMI filtering and other purposes that
contributes
to
the
aforementioned
inductive
characteristics of the return.

The running rails must also carry the train
propulsion and auxiliary current back to the
traction power substation (TPS) to complete the
electrical circuit. The traction currents for a
single 12-car M7 train range from 1200 A dc
during coasting to 10500 A dc at full power.
Each train’s dc propulsion current in the running rails
must by-pass the signal system’s insulated joints. This is
accomplished by installing impedance bonds at the IJ
locations to allow train propulsion current to pass while
blocking the signal current. It is critical to the safety of
the system that signal current of a block not by-pass the
IJs.
Figure 3. Impedance bond across IJ
The action of impedance bonds to block signaling
current while being very low resistance to dc traction
current is critical to the successful operation of the two
systems. Ensuring that signaling current is confined
within the individual signal blocks defined by IJs is also
critical to satisfying the FRA requirement for broken rail
detection. Preventing signaling current from running
around the condition of a broken rail through the traction
system is a major challenge of the railroads’ signal system
when electrified traction systems provide for dc bypass of
IJs and electrical connection of adjacent tracks via
traction substation negative busses.
In multi-track territory, adjacent tracks represent
multiple electrical paths. Tracks are electrically tied
together on the negative bus at substations. Electrical ties
between tracks are also provided in order to reduce the
resistance of traction return circuits. Tracks that are tied
together electrically at substations and at other locations
between substations are referred to as cross bonds. Cross
bonds need to be located far enough apart such that when
a rail breaks, signaling current does not utilize the cross
bonds and adjacent track running rails as run around
paths. This is depicted in Figure 4 where the signaling
current intended for the rail with the break conducts
through the tracks and substation cross bonds to complete
the circuit around the break.
Figure 2. Rail damaged at IJ from arcing
Also contributing to the inductance in the traction
system are the reactors (air-core coils) in series with the
return circuit to increase the impedance of the traction
return system in order to maintain broken rail protection.
Starting in the late 1980’s, Metro-North and LIRR began
installing the reactors as a means of ensuring FRA
mandated broken rail detection of the signaling system.
While there were cases of arcing prior to the advent of the
M7s, they were at few locations, less severe, and
generally mitigated with cross bonding.
2.0
TRACK CIRCUIT
On an electrified railroad with block signaling, the
running rails serve two purposes in addition to supporting
and guiding trains:
 The rails are used as electrical conductors for
track circuits that detect the presence of trains
within signal blocks, delineated by the
installation of insulated joints (IJ) in the track.
2
Destructive Arcing of Insulated Joints in DC Electrified Railway
different between the M3 and M7 EMUs. The M7s
incorporate conductors other than the car body to carry
the return currents to the ground brushes, providing a
better conducting path than the M3s for the propulsion
return current to the running rails. Depending on the
return circuit impedance in the track, more current can be
returned through the last axle with the M7s than the M3s
because the train has a lower resistance than the track
circuit under the train. This may contribute to more
severe arcing when operating M7s than M3.
Figure 4. Signal Current Runaround Path
MNR and LIRR install reactors in the substation
return circuit to increase the impedance of the run around
paths to the 100 Hz +/- signaling current. Increasing the
impedance of the traction return system to signaling
currents is necessary where the quantity of substations
required for high performance trains results in substation
spacing closer than permitted for the signal system to
guarantee broken rail detection. The wayside reactors
have an adverse effect on IJ arcing by increasing the
inductance of the traction return system. The inductance
of the wayside reactors in combination with the inductors
used for filtering on ac drive propulsion circuits provide
the energy storage for bolstering the arc at IJs.
Cross-overs and turnouts create electrical connections
between running rails of adjacent tracks. Accordingly, IJs
are inserted in the rail sections to prevent the flow of
signal current between adjacent tracks.
When trains change tracks at cross-overs and
turnouts, current is interrupted when the wheels pass over
the IJs separating the tracks. When this interruption takes
place, an arc is generated that produces intense heat,
enough to melt a part of the steel rails. Each arc removes
a part of the steel and results in rapid erosion of the
running rails near the IJ and destruction of the end post.
This irregular surface results in a mechanical shock each
time a wheel passes over it. After a period of time, the
deformed rail head shorts the IJ and signal block isolation
is lost, causing the signal system to assume a danger state
restricting train movements. At that point, repairs are
needed to restore the proper function of the insulated
joints. The problem experienced at MNR and LIRR is
that arcing at some IJs is so severe that the IJs are failing
rapidly after only a few months of operation.
4.0
MNR AND LIRR DIFFERENCES
There are a few differences between the LIRR and
MNR trains and wayside circuits as noted in Table 1.
LIRR limits their 12-car trains to 10500 A, whereas MNR
limits their 10-car trains to 7000 A. Since the longer
LIRR trains draw higher currents, their respective
impedance bonds utilize 3-500 kcmil cables at the rail
connections with a solid neutral bus bar between bonds.
These high currents can present a problem if the rail
return paths do not have relatively equal resistances. For
example, a typical IB requires that the impedance (at
signal operating frequencies) shall not decrease by more
than 10% with an out-of-balance current in one half of the
winding exceeding that in the other half by the margin of
12% of the full continuous rated current. Incorporating a
continuous current rating of 2000 A means that the
difference in current between the rail leads needs to stay
within 240 A. An early theory was that impedance bonds
near the IJs may be saturating and causing voltage spikes
leading to arcing. This theory was dismissed after
measuring the current in impedance bond connections and
finding them to be within unbalance ratings.
LIRR and MNR Equipment
Item
Impedance
bonds
Train current
LIRR
3-500 kcmil per side
lead cables
MNR
2-500 kcmil per side
lead cables
Solid copper busbar
neutral connection
12-car trains limited
to 875 A per car
(10,500 A limit per
train)
3-500 kcmil cable
neutral connection
10-car trains limited
to current draw of 8car train (7,000 A
limit per train)
Table 1.
5.0
Figure 5. Adjacent track electrical isolation
3.0
ELECTRICAL THEORY
The train wheels crossing an IJ are similar to a switch
opening in the dc circuit. An arc may be struck by
drawing apart two touching electrodes between which a
current is passing [2]. The components of the circuit
include:
TRAIN CIRCUIT
The electrical connections of the car body and
electrical circuits to the axles via ground brushes are
3
Destructive Arcing of Insulated Joints in DC Electrified Railway
Lr, Lt
Rr
Lb
Vc
Rc
r
t
τ
Carol C1166.21.01, coax
cable, RG58/U, 50 ohm, 18
AWG, 1000 ft roll
Rail clamps
wayside and train inductance
rail and IJ (contact) resistance
IB inductance across the IJ, if present
voltage across the IJ (contact)
time-varying IJ (contact) resistance
contact resistance under full area
time
total time for interruption
6.2
These equations show that the voltage developed
across the IJ is a proportional to circuit inductance, the
magnitude of current in the circuit, and the variable
resistance [3] caused by the wheel breaking contact with
the rail as it crosses the IJ.
FIELD TESTING
In order to examine the traction return circuit,
quantify the magnitude of the arc, and categorize the
factors contributing to the arc, field testing was conducted
to measure the current and resulting voltage during arcing
conditions at a variety of locations and train performance
conditions. Testing was conducted at three MNR and
LIRR locations where arcing had been observed to be
significant. The premise was that if electrical limits could
be established as a function of the arcing conditions, then
methods of mitigation could be categorized and
investigated.
Test Equipment
Table 2 provides a list of the major equipment used
during the field testing. Hall effect type clamp-on current
probes were selected to measure the direct current traction
circuits at fouling wires and negative return feeders.
Voltage transducers were used to isolate the rails from the
test recorder.
Field Testing Equipment
Item
Astro-Med TMX, data
recorder, 18 channel, with
three (3) UNIV-6
AEMC MR561, clamp-on
current sensor, 1500 A dc
AYA P3C-7500-C, clamp-on
current sensor, 7500 A dc
Probe-Master 4231 HV,
voltage sensor
APC BE550G, UPS to power
recorder on board train, 550
VA, 330 W, 120 V ac output
Notes on Testing
The Astro-Med recorders, cable, and all measuring
devices were set up and fully tested prior to transporting
to the field. Current and voltage probe polarities were
checked and marked on the devices.
Coax cable
attenuation was checked to ensure that the expected
signals would propagate up to 1000 ft.
One lesson learned was that a GPS-based time card
should have been rented with each recorder in order to
synchronize the wayside and onboard recorders. This
would have made post-processing of the results easier.
RG58 coax cable was suitable for distances up to
1000 ft due to the relatively low frequency of the
transients compared to the attenuation ratings of the cable.
There were two factors that contribute to interference on
long coax cable runs: 1) electromagnetic fields from
underground cables and 2) wet conditions. It was noted
during testing that when rain made the cable wet, there
was noticeable interference on the test signals. Also, at
Harold, there was periodic signal interference, but this
was eliminated by routing the coax cables using a
different path.
Large split-core clamp-on hall effect current probes
are subject to electromagnetic interference. They are
sensitive to how they are positioned. For best results,
they need to be zeroed each time they are initially setup.
When zeroed (set to 0 A when reading no current), they
should ultimately be set after positioning where they will
be placed during testing. A small battery-powered coil of
wire, current limited by a resistor, and multimeter can be
used in the field to test the current probe when zeroed at 0
A and some higher test current, such as 40 A. The test
coil is also handy for verifying current probe polarity
prior to installing on the test circuit.
Eq. 2
6.1
Connect voltage probes to
running rails
Table 2.
Eq. 1
6.0
Connect measuring devices
to TMX recorder
Use
Record voltage across IJ and
current in fouling wires and
negative returns
Measure current in fouling
wires and on train
Measure current in negative
return feeders
Measure voltage across IJ
Keep Astro-Med TMX
powered on train during 3rd
rail gaps
4
Destructive Arcing of Insulated Joints in DC Electrified Railway
Figure 6. Battery-powered coil and multimeter
Figure 7. Crestwood test configuration
6.3
MNR Crestwood
Crestwood has experienced rapid destruction of IJs at
the #21 crossover due to electrical arcing observed when
electrified trains pass over the IJs. These IJs are in a
crossover and do not have impedance bonds. A series of
tests was performed in order to quantify the magnitude of
current and voltage in the track and train circuits under
various operating scenarios.
The closest substation, B16, was located just south of
the #21 crossover. This substation supplies the majority
of current to a train at #21 crossover. This test
configuration was instrumented by placing 4000 A
current probes around the IB rail leads in the return path
of train currents flowing to Substation B16. The fouling
wires on the crossover were monitored using 1500 A
current probes. Voltages across each IJ were measured
using voltage probes that could be adjusted to either 140
V or 1400 V full-range. Two recorders, A and B, were
used wayside and two recorders, C and D, were used on
the train. The equalizer bond between track 1 and 2 was
installed for certain tests and is not used during normal
operation.
In Figure 7, as the train moves south, it crosses the
21E IJ first. The last axle of an 8-car train will break the
21W IJ last. The potential probes were adjusted from 140
V to 1400 V once high transients were measured.
The TMX recorder plots are shown below.
Recorders A and B measure currents in fouling wires,
current in impedance bond leads, and voltages across IJs.
Recorder C measures train voltage and current in one car.
21W IJ Volts
21E IJ Volts
21S-FW-1 Amps
21S-FW-2 Amps
21N-FW-1 Amps
21N-FW-2 Amps
Figure 8 . Crestwood Recorder A
The first two plots in Figure 8 show the voltage
measured across each IJ. Note the large spike near the
end of the plot. This is the transient arcing voltage that
occurs as the last axle crosses the IJ. The bottom four
plots in Figure 8 provide indication of the fouling wire
currents as the train passes over the IJs.
Figure 9 shows the current in the IB leads. Prior to
crossing the IJ, the track 1 IB probes B10 and B11 have
2210 A and 2000 A flowing at one instant. This is about
a 10% difference in current, demonstrating that the lead
resistances are balanced. The other IB leads for track 2
on probes B9 and B10 have 2252 A and 2357 A at on
instant, representing a difference of about 4.5%.
Figure 10 shows the train increasing from almost 0 A
to 757 A in 6.36 s for a rate-of-rise of 119 A/s per car
during controlled acceleration. For an 8-car train, this
5
Destructive Arcing of Insulated Joints in DC Electrified Railway
equates to 952 A/s. At one point, the car current suddenly
dropped to 0 A, then back to 760 A when passing over a
third-rail gap. The rate-of-rise during this re-energization
occurred in 684 ms for 1111 A/s per car or 8.9 kA/s for an
8-car train.
across the IJs, and no equalizer bond in place are shown
in Table 3. The highest voltage across the IJ was
measured at 291 V. At this time, the current in two
fouling wires from one rail peaked at 2012 A as the arc
was beginning to form. The train was drawing about
6088 A during this test. About 4250 A returned through
IB1 as the train began to cross the IJ and 4633 A was
returned through IB2 after the train crossed the IJ. The
remaining return current is assumed to have been flowing
to the adjacent substation to the north.
The arcing duration, from beginning of the voltage
rise to the peak, is about 30 ms. Figure 11 shows that the
current in the west fouling wires flowing to track 1 almost
doubles just prior to the last axle crossing 21W. This is
because the train just traversed 21E, the IJs are staggered,
and the voltage potential in the track circuits requires the
current to rebalance, causing current to flow back to
through the train to track 1. The time required for the
current to rise is due to the track inductance.
The current being broken by the last wheel to cross
the IJ is the sum of the two fouling wires measured by
probes A5 and A6, which are about 2012 A just prior to
the arc development. This current collapses to 160 A in
30 ms, for a di/dt = 61.7 kA/s.
The following table provides the maximum voltages
and currents recorded at Crestwood.
Crestwood Test Results, Test M-1b
2-B16-IB-W Amps
2-B16-IB-E Amps
1-B16-IB-W Amps
1-B16-IB-E Amps
Figure 9 . Crestwood Recorder B
Train voltage
Probe Designation
A1-21W-IJ
A2-21E-IJ
A3-2-21S-FW-1
A4-2-21S-FW-2
A5-1-21N-FW-1
A6-1-21N-FW-2
B8-2-B16-IB-W
B9-2-B16-IB-E
B10-1-B16-IB-W
B11-1-B16-IB-E
Train current (reverse polarity)
Figure 10. Train Recorder C
The train current in Figure 10 is shown with reverse
polarity, i.e., as current increases, the above plot shows it
increasing in the negative direction.
A7-21-EB
Train Voltage
291 V transient across IJ
Train Current
Value
291 V
254 V
749 A
718 A
982 A
1030 A
2264 A
2369 A
2228 A
2022 A
3505 A
(121 A just prior to last axle crossing)
576 V
761 A average per car
876 A peak per car
6088 A average for 8-car train
Table 3.
Subsequent testing at Crestwood showed that the
voltage transient was improved by connecting an
equalizer bond between tracks 1 and 2 (measured by
probe A7-21-EB).
The number of train cars can be counted crossing an
IJ by observing the pattern in the voltage measurements.
A sample observation has been annotated in Figure 12.
982 A to 73 A in 29.4 ms
Figure 11. Voltage across 21W and current in fouling wires
Significant test results for a train operating at 100%
power, moving south across the #21 crossover, no IBs
6
Destructive Arcing of Insulated Joints in DC Electrified Railway
Figure 12. Counting Cars Using a Voltage Plot
6.4
LIRR Harold
The tests at Harold interlocking shown in Figure 13
were performed using a 12-car train. For full-power run,
the 12-car M7 test train started on Track #ML2, at Harold
Interlocking, West of the 848E signal. The train Engineer
moved the train at low speed into the 849 crossover
reverse, stopping the front of the train at the test starting
point. The starting point for an eastbound run is defined
as the lead car at the 849E turnout. Upon test initiation,
the engineer accelerated to full power (100%) up to the
permitted crossover speed of 40 mph. After clearing
interlocking signal 854E, the engineer stopped the train
and the test concluded. Screen captures of the complete
Test #L3 and definition of the configured probes are
provided in the following figures.
An impedance bond was connected across the IJs
prior to the start of the tests. However, the neutral
connection between the bonds was not connected. While
this left the IJs without a connected return path across
them, it did allow current to flow between rails. For
example, when the last axle of the train crossed the south
IJ, the IB connecting the rails allowed current to flow
from the north rail through the IB to track ML2, which
would not normally have occurred. This was noticed in
the recorder plots when the current in probes A3 and A4
dropped an average of 190 A each when the train crossed
the south IJ. These IBs could have resulted in lower arc
voltages than during normal operations.
Figure 13 . Harold test configuration
Figure 14. Harold Recorder A
A magnified view of the IJ voltages and fouling wire
currents is shown in Figure 15.
Figure 15. Harold IJ Voltages and Fouling Wire Currents
The following table provides the maximum voltage
and currents recorded at Harold.
7
Destructive Arcing of Insulated Joints in DC Electrified Railway
Harold Test Results, Test L3
Probe Designation
A1-849N-IJ
A2-849S-IJ
A3-ML2-849W-FW-1
A4-ML2-849W-FW-2
A5-ML4-849E-FW-1
A6-ML4-849E-FW-2
A8-ML2-G02-IB-N
A9-ML2-G02-IB-S
A10-ML4-G02-IB-N
A11-ML4-G02-IB-S
Train Voltage
Train Current
Value
124 V
114 V
536 A
583 A
742 A
777 A
1387 A
1378 A
4205 A
3718 A
696 V
870 A average per car
1097 A peak per car
10440 A average for 12-car train
Table 4.
Testing proved that fouling wires and IBs were
connected properly and that the IB imbalance did not
exceed 13%.
The maximum arcing voltage at Harold was half that
of Crestwood. Some of this is attributed to the impedance
bonds that were connected across the rails for purposes of
connecting an impedance bond across the IJs during
testing. However, in cases where the neutral was not
connected across the IJ, currents that should have flowed
from the crossover out through the fouling wires instead
flowed from the crossover through the IB to the frog.
Later testing with the neutral connected across the IJ
using the IBs reduced the arc voltage to from 124 V to 23
V.
Figure 16. Mineola test configuration
During the 9:21 pm test at Mineola, the eastbound
train has its last axle cross the north rail first, then the
south rail. Arcing was only observed on the south rail.
P1 Volts
P2 Volts
P5 Amps
P6 Amps
P7 Amps
P8 Amps
6.5
LIRR Mineola
P9 Amps
The IJs at Mineola are different than at Crestwood or
Harold in that they already have IBs connected across
them during normal operating conditions.
It was
previously thought that arcs would only develop across
IJs that did not have IBs. However, the current that
passes through the IBs cannot be changed instantly when
the train passes over the IJ. The current in the IBs
provides a stabilizing voltage across the IJs and reduces
the arc across the IJs, but does not eliminate it.
Figure 17. Mineola Recorder A
The following table provides the maximum voltages
and currents recorded at Mineola.
Mineola Test Results, Test 9:21 pm
Probe Designation
P1 IJ-N Trk 2
P2 IJ-s Trk 2
P5 IB-N Trk 2
P6 IB-S Trk 2
P7 G16-Trk 2a
P8 G16-Trk 2b
P9 G16-Trk 2c
Train Voltage
Train Current
Value
27 V
37 V
4600 A
4261 A
3054 A
3028 A
3134 A
Not recorded
Not recorded
Table 5.
The maximum amount of current through the IB
neutral across the IJ was the sum of probes P5 and P6,
8
Destructive Arcing of Insulated Joints in DC Electrified Railway
which equaled 8861 A. The substation return leads
summed up to 9216 A, showing that almost all current
flowing through the IB was returning to the substation.
An interesting current flow was observed as the last
axle crossed the north and south rail IJs. Just before the
north rail IJ is crossed (P1), the north rail IB lead (P5) has
3175 A and the south rail IB lead (P6) has 2737 A, a
difference of 438 A. When the axle crosses the north IJ,
but has not yet crossed the south IJ, the current in the
north IB lead (P5) goes from 4261 A to 4179 A, while the
south IB lead (P6) goes from 3738 A to 2620 A.
Therefore, the difference in the IB leads is 4179 - 2620 A
= 1559 A just before the axle crosses the south rail IJ.
When the last axle crosses the south rail IJ, the current in
the south rail lead (P6) of the IB goes from 2620 A to
3911 A, or a change of 1291 A, while the north rail lead
(P5) only changes by 340 A. It is postulated that if the IJs
were not staggered, then this transient in current in half of
the IB would not take place and the arcing voltage would
be less than the 37 V measured in the worst-case
condition. Note that this current transient was never seen
in the north rail IB lead and, consequently, no arcing was
seen across the north rail IJ. As a follow-up to this test,
LIRR plans on removing the stagger at this location for a
trial period to see if this improves the arcing conditions.
7.0
Crestwood
100%
Crestwood
50%
Harold
100%
Harold
50%
Harold
25%
Crestwood
100%
Harold
Coasting
Crestwood
25%
Crestwood
Coasting
IJ Type
No IB
across IJ
No IB
across IJ
No IB
across IJ,
but in close
proximity
No IB
across IJ,
but in close
proximity
No IB
across IJ,
but in close
proximity
Cross bond
No IB
across IJ
No IB
across IJ
No IB
across IJ
IJ Type
Maximum
Arc
Voltage
Mineola
100%
IB across IJ
37 V
Harold
100%
IB across IJ
23 V
8.0
OBSERVATIONS
The tests conducted at three locations and the trends
in train operating power, rail and track bonding, and
resulting arcing voltages shown in Table 6 lead to the
following observations:
 The highest arc voltage occurred during the
maximum train power and no bonding near the
IJ.
RESULTS
Location
Train
Power
Table 6.
A summary of significant test results for all three test
locations are provided in Table 6. Results are ordered
from greatest to least arc voltage.
Summary of Arc Voltage Testing
Train
Power
Location
Maximum
Arc
Voltage

The lowest arc voltage occurred with impedance
bonding across the IJ even with maximum train
power.

Cross bonding near the IJ lowers the arcing
voltage across the IJ.

Impedance bonding near an IJ lowers the arcing
voltage across the IJ.

Brighter arc flashes were noticeable for
measurements that recorded higher arc voltages.

Higher train currents contributed to higher arcing
voltage.

Reducing the train power while crossing an IJ
reduces the arc transient across the IJ.

The effects of higher train current can be
mitigated by increasing bonding across the IJ.
132 V
While more impedance bonding around the IJs helps
mitigate destructive arcing, these cross bonds affect
signaling and should be addressed on a case-by-case
situation after proper analysis.
124 V
9.0
MITIGATION OPTIONS
9.1
Operational Adjustments
291 V
123 V
Both analysis and test results indicate that trains
under full power at the time the last axle crosses the IJs
are the worst-case conditions contributing to the severe
damage to the IJs and rail. Damage of the rail is
proportional to the magnitude of the current flowing when
the last axle crosses the IJs. As shown in the testing, train
operation whereby the Engineer operates the train at coast
while the last axle of the train is operating over the IJs
reduces the voltage across the IJs and thus the potential
arcing and resulting damage.
87 V
74 V
68 V
47 V
42 V
9
Destructive Arcing of Insulated Joints in DC Electrified Railway
A potential mitigation then would be to establish in
the Timetable Special Instructions "coast zones" whereby
Engineers would move to coast for a distance such that
the train will not be taking power when the last axle
crosses the IJs. The specific limits of the zones could be
marked in the field with signs so Engineers would know
when to go to coast and when they would be able to
resume power. The coast zone would need to account for
the worst-case train lengths, i.e., shortest train for start
coast and longest train for end coast.
The ACSES II system that MNR and LIRR are
implementing for Positive Train Control (PTC) has a
capability to force a train to operate in coast. The feature
was designed to ensure trains were in coast when
operating through phase breaks. The feature includes a
message from a transponder located at the point where the
highest speed train would need to move to coast to ensure
not being in a power mode or having pantographs
extended (where they need to be lowered for phase
breaks.) Due to the complexity of determining train speed
and location relative to the phase break and variations in
train makeup, the function was not implemented. For
similar complexity, the ACSES capability was deemed
not suitable to implement coasting to reduce arcing on IJs.
9.2
effectively eliminating the arcing. This was visually bornout in the testing at Harold by observation of the IJs under
train movement and by the test results. However,
introducing impedance bonds in crossovers both
jeopardizes broken rail protection (BRP) and imbalances
the track circuit, resulting in a destabilized tract circuit.
9.3.3
Shunting IJs While Train Passes
Several methods of shunting IJs until the last wheel
of the train passes have been investigated. These methods
provide an electrical by-pass in parallel to the IJ in the
presence of an electric train, thereby theoretically
eliminating the arc at the IJ when wheels cross the IJ.
The methods investigated for shunting IJs include:
 An active circuit that detects the presence of a
train and closes a contactor creating the by-pass
circuit
9.4
Train Propulsion Circuit
A passive circuit using back-to-back GTOs and a
trigger mechanism, which activates with a
predetermined value of the voltage across the IJ

A passive circuit using varistors that are
activated by the voltage across the IJ.
Concept Solutions
Several concept solutions are offered, although none
have been tested in actual railway operation for the
purpose of reducing arcing across IJs.
Although arcing has occurred on the railroads prior to
the use of the M7s, it is suspected that the M7 propulsion
circuit is contributing to the arcing, which combined with
the high current capabilities used to achieve train
performance, is damaging IJs.
The onboard LRC circuit is integral to the EMI
filtering of the propulsion and auxiliary power circuits
and is viewed necessary by vehicle engineers. Detailed
analysis of the propulsion circuits has been proposed
using EMTP software.
9.3

9.4.1
Additional Return Rail (4th Rail)
The concept of this mitigation is to provide an
additional return path in the area of IJs that have
demonstrated arcing. In addition to the supplemental
return path around IJs, the concept would require the
equivalent of a third rail pickup shoe on all electric rail
vehicles. It is noted that a line of the London
Underground Transit System uses a 4th rail for traction
for the entire system, eliminating the use of running rails
at all for the traction system. The 4th rail is installed in
the gage of the track.
Wayside Systems
9.3.1
Cross Bonding Track Circuits
Additional cross bonding can reduce the voltage
across the IJs and thus the arcing. This was demonstrated
in the testing with equalizer bonds at CP 116 and with
impedance bonds in the crossovers at Harold. The
problem with additional cross bonding is that they
jeopardize the ability of the signaling system to detect
broken rails. Cross bonding must be designed and used
carefully so as to not impact broken rail detection.
9.4.2
Mechanical Shunting
This concept involves placing a non-conductive,
spring-loaded bar on the inside of a running rail at the
affected IJ location. The non-conductive bar is anchored
to the rail at one end and has a free-moving contact at the
other end of the bar. The free-moving contact is
positioned far enough away from the rail that it won't arc
over to the rail. A fixed contact is mounted just below the
movable contact on the tie. The fixed contact is wired to
the rail on one side of the IJ and the free-moving contact
is wired to the other rail on the opposite side of the IJ. As
9.3.2
Impedance Bonds Across IJs
The installation of impedance bonds in crossovers
provides a path for return current around the IJs,
10
Destructive Arcing of Insulated Joints in DC Electrified Railway
the train passes, the wheel flange presses down upon the
non-conductive bar, making the contacts touch and
shorting the IJ. Once the axle passes, the bar opens and
the IJ resistance is restored. Any arcing will occur across
the fixed and free-moving contacts. The contacts can be
periodically replaced during routine maintenance and
should keep destructive arcing from damaging the IJ.
electrical property of graduated resistance to electrical
current in the vicinity of IJs. Such a rail technology
exhibiting this property and having the strength of the
existing steel rail is not known to exist and would require
research and development. The purpose of the graduated
resistance would be to reduce the rapid change in current
as the wheels cross the IJ, eliminating the resulting arc.
10.0
RECOMMENDATIONS
The near term solutions that can be acted on
immediately include operational procedures, such as
reducing train power while operating across IJs, and
increased cross bonding (subject to broken rail analysis).
For the most severe locations that experience arcing
damage on a regular basis, such as the 849 crossover at
Harold and the 21 crossover at Crestwood, railroads
should implement operational procedures to reduce the
arcing.
Each location that is experiencing arcing can also be
looked at on a case by case basis for cross bonding
solutions that include using impedance bonds as
equalizers. As has been noted, this mitigation requires a
thorough analysis of the impacts that the added cross
bonding may have on broken rail protection. While this is
the quickest and least costly approach to mitigating the
worst-case arcing in the near term, this is not considered
an optimal solution and additional mitigation should be
investigated. Other methods, such as using passive
shunts, aligning IJs that already use impedance bonds
(such as at Mineola), and modifying vehicle propulsion
systems, should be further analyzed and tested.
Longer term solutions include additional cross
bonding where possible using reactors, if necessary, to
provide broken rail protection, new IJ technology, and
deploying passive shunts if prototypes prove successful.
Use of reactors with added cross bonding is considered
long term only because it is assumed that the acquisition
and implementation of the reactors is more difficult,
costly, and requires more time to implement than the use
of impedance bonds in the equalizer configuration. Use
of reactors should also be evaluated for their added
inductance to the circuit to ensure that they do not
exacerbate the arcing. Lap joints and new IJ technology
should be monitored and tested where applicable.
Contactors and electronic shunting devices may provide a
solution, but they need significant development to
determine if reliable control circuitry and device ratings
could be developed.
Figure 18. Mechanical Shunt
9.4.3
Long-Lap IJ
The long-angle or long-lap IJ is under test by freight
railroads [4]. The purpose of the long-lap joint is to
mitigate the effects of mechanical stresses that occur on
IJs when each rolling stock wheel crosses over the IJ.
The joint has some promise of mitigating the destructive
arcing MNR and LIRR have encountered by providing a
longer transition of the last axle between the circuit that
the train is leaving to the circuit that the train is entering.
LIRR intends to test this type of IJ in the future. It
has been tested at Transportation Technology Center, Inc.
(TTCI) with good results from the mechanical perspective
of the IJ. However, TTCI was not testing for the
destructive arcing problem encountered on MNR and
LIRR.
Figure 19. Long-Lap IJ
11.0
9.4.4
Graduated Resistance IJ
An ideal solution for the arcing at crossover and
turnout IJs would be a rail technology that has the
REFERENCES
[1] "MNR & LIRR Insulated Joint Arcing Study,"
SYSTRA Engineering, Inc., January 2013.
11
Destructive Arcing of Insulated Joints in DC Electrified Railway
[2] Somerville, J. M., The Electric Arc, London: John
Wiley & Sons, Inc., 1959.
[3] Rudenberg, R., Transient Performance of Electric
Power Systems, New York: McGraw-Hill Book
Company, Inc., 1950.
[4] Akhtar, M., et al., "Revenue Service Evaluation of
Advanced Design Insulated Joints," AREMA
Conference Proceedings, 2008.
12