INVESTIGATION OF ARCING ACROSS INSULATED RAIL (BLOCK) JOINTS
AT INTERFACE BETWEEN NEW AND EXISTING DC TRACTION POWER
SYSTEMS
Thomas W Palfreyman
Parsons Brinckerhoff Ltd
Westbrook Mills, Godalming
Surrey, GU7 2AZ
UK
[email protected]
KEYWORDS: Traction, Power, Arcing
ABSTRACT
Arcing has been occurring during the passage of new Electric Multiple Units (EMUs) across insulated rail
(block) joints at the interface between the separate traction power return systems of a new DC third rail
electrified railway and an existing established DC electrified railway. These systems are separated by a
long (conductor rail) section gap, with the new system designed for high levels of running rail to earth
resistivity. Results of a detailed technical investigation undertaken to understand the phenomena, consider
and eliminate possible causes, and propose potential solutions for further development are presented. The
investigation comprised a detailed review of designed and constructed traction power systems, and
predictive modelling of stray leakage current flow. In addition, detailed examination in conjunction with
the train manufacturer of arcing taking place on particular wheelsets in relation to EMU traction
equipment behaviour during passage through the gap and EMU bonding arrangements. Extensive visual
observations, high speed video photography, and transient voltage measurements with simultaneous onboard train monitoring provided further insight. The probable cause was stray leakage current flow
between traction power systems via EMU wheelsets, considered to be exacerbated by EMU traction
current when coasting not employed. A range of solutions is identified, including snubber and switching
arrangements, and an update on recent developments (and of further planned testing) presented.
INTRODUCTION
This paper describes the results of detailed investigations into arcing across Insulated Block Joint (IBJs)
at the interface between the DC traction power (conductor rail) systems of the new East London Line
(ELL) railway system and the existing Network Rail (NR) system. These IBJs are positioned at the midpoint within 80m long traction system gaps at this interface. This arcing has been occurring since
introduction of trial train operations on the ELL system in early 2010, at which point regular train passage
across the system gaps on both Down and Up lines was established. Passenger service operation
commenced several months later and arcing continues to the present day. The new ELL railway system is
described in detail in a separate paper presented at the same conference entitled ‘Systems Integration on
the East London Line’ (author A G Shepherd) which also provides an overview of the arcing issue.
The focus of the investigations has been the IBJs between the two signalling track circuits (MF-1 and
MF-2) which form a traction system gap on the Down Line since arcing was first observed at this point.
Also, due to its location, this gap forms part of NR infrastructure and it was understood that failure of the
IBJ in this gap would cause track circuit operation having the potential to affect NR main line service
operation. However, the risk of such failure due to arcing damage has been managed through regular
inspection and, as a temporary measure, replacement of IBJs as necessary. There were also concerns over
longer term damage to the rail, although both the NR track inspector and London Overground
Infrastructure (LOI) maintenance engineer consider that this is minor and not a threat to safety.
The Up line gap, by contrast, is located such that this forms part of LOI. A lesser amount of arcing had
been observed at this location. Again the LOI maintenance engineer viewed the damage as being
relatively minor and not a concern, although subject to regular inspection as a precaution. Detailed
investigations at this location have not been undertaken.
The paper presents a description of the ELL and NR traction power systems and system gap
arrangements, arcing observations and voltage measurements, together with a detailed understanding of
the particular rolling stock in operation which was a key component. The possible candidate mechanisms
for arcing that were identified, and the process by many of these were eliminated, are described in detail.
Results of testing undertaken to date to further inform understanding of the arcing process and confirm
the probable cause are presented. Possible solutions are also outlined.
TRACTION POWER SYSTEM CONFIGURATION
The ELL DC traction power system is electrically separate from that of NR and comprises three multiple
rectifier traction substations located at Hoxton, Shadwell and Canal Junction, operating at a nominal
voltage of 750V. It employs a floating traction return system designed for high levels of rail to earth
resistivity. The ELL traction supply includes New Cross Gate Depot which was originally conceived to
be electrically isolated (both positive and negative poles) from the running lines, however this approach
has since been modified with the depot now being interconnected with the traction power system of the
running lines. The NR DC traction power supply system operates at a nominal voltage of 660V in the
vicinity of the New Cross Gate connection, and employs a floating traction return system having (lower)
rail to earth resistivity levels as might be expected for an established system that has been subject to many
years of in-service operation.
The conductor rail system gaps are located at the southern end of the ELL traction power system, at the
limit of single end fed electrical sections supplied from the nearest traction substation (Canal Junction)
which is at around 8.7km point. The NR traction power system on the other side of the gaps is that which
is partway between two traction substations at the location of the (New Cross Gate) Track Paralleling Hut
(TPH). The gaps are located on completely separate Down and Up line connections between the ELL and
NR systems.
The Down line gap is located just to the south of, and beyond the departure from, New Cross Gate
platform with mid point at approximately 9.8km and on an uphill grade. The Down line gaps are at this
point alongside the NR Down Slow line to which connection is made further to the south. The ELL/NR
boundary on this line is at 9.7km which is towards the north limit of the system gap. The Up line gap is
located on an ELL flyover over the NR main lines at approximately 8.9km and on a downhill grade. The
ELL/NR boundary on this line is at 9.4km, with the short (450m length) conductor rail section between
the gap and the boundary being fed from the NR TPH but forming part of the ELL railway system.
TRACTION SYSTEM GAP ARRANGEMENT
The passage of a train over the Down line system gap as originally configured is illustrated in Figure 1.
(Subsequently, as described later, the short sections of conductor rail in the centre of the gap, known as
‘rescue rails’, were removed but the diagram remains generally valid). The rolling stock that has been in
operation across both Down and Up line system gaps is Bombardier Transportation (BT) ‘Class 378/1 DC
Only’, the definitive train consist and car configuration for which is presented in Figure 2, for clarity
illustrated alongside and aligned with the system gap in Figure 1. The train is orientated such that
DMOSA car is always facing North and DMOSB car is always facing South as the train crosses the gaps
(this is also known and referred to as ‘with PTOS to the south’). Only three of the four cars (DMOSA,
MOS and DMOSB) contain traction power equipment and are DC powered, whereas PTOS is a trailer
vehicle without traction power excepting DC shoegear and internal busline interconnecting DMOSB car.
Fig 1: Train Passage through the Down Line 80 m
Gap
Fig 2: Configuration of Class 378/1 DC only
North
South
Direction of Travel
Figure 1: Train Passage through the Down Line 80m System Gap
Figure 2: Configuration of Class 378/1 DC only Rolling Stock
On the Down line, traction return is double rail on the NR side of the IBJs at the mid-point of the system
gap. On the ELL side of these IBJs, there is a short 36m length of single rail track circuit within the
system gap up to an IBJ in the ‘signal rail’ just to the south of the commencement of the ELL conductor
rail, north of which traction return is double rail. For this track circuit, traction return is single rail (using
the cess side rail) with the other rail (closest to NR Down Slow line) as the ‘signal rail’. The arrangement
of the system gap on the Up line is the same as that for the Down line (rescue rails having been removed),
with the direction of travel reversed and double rail traction return either side of the IBJs.
Trains are expected to cross the Down line gap at an approx speed of around 15 to 20mph having
accelerated away from New Cross Gate platform. A time for each car of the train to fully cross the system
gap of around 10 seconds is expected, and as a consequence of shoegear at the front and rear of each 2 car
unit the conductor rail feed to (each 2 car unit of) the train is lost for around half of this period (5
seconds). It was understood that, in relation to train operation over the Down line gap, drivers are under
instruction to coast through the gap (with traction power controlled in ‘neutral’ position); however there is
no lineside ‘coasting’ instruction board installed and, based on evidence obtained during later testing, it
was considered that drivers may in many cases be motoring through the gap (with traction power
controller in a ‘notch’ position). Trains are expected to cross the Up line at a higher speed of around
40mph and, as this is a downhill gradient, some traction regenerative braking may be expected.
Trains had been in operation (without passengers) over both Down and Up line gaps as part of Test
Running and Trial Operations for a sustained period of time prior to introduction of full passenger service
in mid May 2010. From early May, 18 trains per day were operating over both Down and Up Line gaps,
with 84 trains from mid-May which increased to 250 trains since introduction of passenger service.
ARCING OBSERVATIONS
Arcing damage was observed on the IBJ on the cess side rail of the Down line, as shown in the
photograph in Figure 3. The arcing damage was on the trailing edge (ELL side) of the IBJ, from which it
could be concluded that arcs were being drawn between the wheel tread and the trailing edge (ELL side)
of the IBJ. This was consistent with current being broken as the train traverses the system gap in a
southbound direction from ELL to NR system.
Figure 3: 21/04/10 Arcing Damage on Down Line
Cess Rail showing Direction of Travel
Wednesday 21/04/2010 Down Line Cess Rail - Showing Direction of Travel
Figure 4: 22/04/10 Arcing Damage on Down
Line ’10 foot’ rail (same direction of travel)
Prior to 22/4/10 there was no evidence of arcing on the other ’10 foot’ rail of the Down line, however as
of that date there was evidence of some (very) low level arcing taking place, as shown in the photograph
in Figure 4. This appeared to have resulted in some minor level of arcing material being deposited on the
NR side of the IBJ. Although apparently contrary to the direction of travel (and expected location of
current breaking), it was considered possible that such material deposit may be the result of arcing taking
place on the trailing edge as for the cess side rail.
In addition to photographic evidence of arcing damage, (non high-speed) video observations of arcing
taking place across the cess side IBJ of the Down line during passage of two ‘Class 378/1 DC only’ trains
through the Down line gap, under daytime conditions, were made in early May. These were a critical
early input into the investigation process in that they indicated (visible) arcing to be taking place on
wheelset numbers 4, 12 and 16 only, which are the last wheelsets of DMOSB, MOS and DMOSA
(powered) cars to cross the IBJs. In addition, the observations indicated that the arcing on wheelset 16
may be more pronounced than others.
Photographic evidence of arcing having taken place on the Up Line IBJs is shown in Figures 5 and 6,
which indicates arcing has taken place equally on both running rails.
Figure 5: 12/04/10 Arcing Damage on Up Line
‘First side rail’, direction unknown
Figure 6: 12/04/10 Arcing Damage on Up Line
‘Second side rail’, direction unknown
VOLTAGE MEASUREMENTS
To enable understanding of the traction power system conditions during passage of a train through the
Down line system gap, profiles of rail to earth voltage were measured on both sides of the Down line cess
side rail IBJ. These were measured between the rail (on each side of the IBJ) and a (common) electrode of
a Copper/Copper sulphate cell at a distance around 5m normal to the Down Line IBJ. A four hour period
commencing 20.00hrs was monitored using a one second measurement resolution during which a normal
planned ELL service (of around 12 trains per hour per direction on the ELL core section) was in operation
with zero passenger load. During this period a total of 11 train passages over the Down line system gap
would have occurred (and correspondingly 11 over the Up line gap). From these results, the potential
difference across the IBJ was also calculated as the difference between the two measured values. An
analysis of the measurements is presented in Table 1.
Date
13/4
20/4
V rail to earth (V)
ELL side
V rail to earth (V)
NR side
PD across IBJ (V)
(calculated VELL-VNR)
Average
-2
8.3
-10.3
Max +ve
21.6
39.77
18.01
Max –ve
-19.86
-18.64
-46.19
Average
-2.58
8.3
-10.9
Max +ve
22.29
38.88
26.36
Max –ve
-18.38
-25.66
Table 1: Analysis of rail to earth voltage profile measurements
-39.71
The rail to earth voltage trace for the ELL side was observed to be mostly negative with some positive
peaks considered likely to be due to train operation south of Canal Junction traction substation,
southbound to/from New Cross Gate station and also to/from New Cross turnback. This was as might be
expected due to the nearby proximity of Canal Junction traction substation and other traction loads on the
ELL system generally to the north of this substation. The rail to earth voltage trace for the NR side was
observed to be mostly positive in value. This might be expected due to New Cross Gate being located
near to an NR TPH partway between two NR traction substations. As the values measured were those
during late evening these will be lower than those during peak operational periods on NR system, under
which higher rail to earth voltage values would be expected.
It was also possible to identify from the voltage traces some incidences where a train may have been
crossing the system gap on the Down line and bridging the IBJs. This was seen in the form of a rail to
earth voltage ‘positive peak’ on the ELL side followed by a period of around 10 seconds during which the
potential difference across the IBJ was close to zero Volts and, in some cases, exhibited a ‘sawtooth’
variation. An example ‘snapshot’ is presented in Figure 7.
Figure 7: Snapshot of Voltage Trace with likely Train Passage through System Gap
DETAILED UNDERSTANDING OF ROLLING STOCK
As arcing was taking place on train wheelsets and in the presence of a train passage, the train was a key
component in relation to the arcing process and also a possible contributor to the arcing. Therefore it was
necessary to obtain a detailed understanding of the configuration of the ‘Class 378/1 DC only’ train (in its
2x2 car electrical formation, comprising DMOSA, MOS, PTOS and DMOSB cars as in Figure 2) and its
behaviour during passage through the traction system gap. This was achieved through various discussions
with and the helpful assistance of the train manufacturer, Bombardier Transportation (BT). The following
key points were established:
i.
DMOSA, MOS and DMOSB cars each have a Line Filter, comprising a series inductor and
parallel capacitor, connected between the DC positive bus line of the 2 car unit and negative
(DC traction return) ‘wire’ (for its respective car) from which separate radial traction return
cable connections are made to axle end return brushes on all wheelsets (on its car). These
connections employ a large conductor size and are of extremely low resistance and inductance.
The PTOS car does not have any axle end return brushes (nor DC traction return connections).
ii.
Neither the car body nor bogies form part of the traction return arrangements; there is only a
single point at which reference earthing is made between the car body and the third wheelset
in each car via the frame of the second bogie. Also, electrical isolation (and insulation) is
provided at the gangway/coupling between each of the 4 adjacent cars, and between all 4
wheelsets of PTOS car.
iii.
DMOSA, MOS and DMOSB cars each have a Motor Control Module (MCM) (shown as
‘Traction Package’ on Figure 1) which consists of an (inductor/capacitor) Input Filter
connecting to a VVVF inverter three phase induction motor drive system, with the Input Filter
interfacing with the Line Filter of that car.
iv.
DMOSA and DMOSB each have an Auxiliary Control Module (ACM) which has an
(inductor/capacitor) Input Filter and is connected between the (same) Line Filter and (same)
negative ‘wire’ as the MCM in that car. There is no ACM in MOS car (nor in PTOS); AC
auxiliary power for these cars is distributed from the (AC) output side of ACM in the adjacent
DMOS (A or B) car.
v.
When a feed to the DC positive bus is ‘lost’ (such as when contact with a conductor rail is
completely broken, as in the system gap) the MCM transitions into a ‘partial/self’ low level
regenerative mode whereby a small part of the kinetic energy of the train is used to support the
train auxiliary ACM power supply loading. This is achieved by the MCM seeking to hold the
DC bus voltage constant at a ‘target value’ of an order similar to that prior to loss of DC
positive supply. This ACM support is for train speeds above approximately 15 to 18mph,
noting that the speed for train passage through the Down line gap is of a similar order.
POSSIBLE CANDIDATE ARCING MECHANISMS
A range of possible candidate mechanisms to explain the arcing was identified. Several of these were able
to be eliminated either completely or to the extent that arcing by that mechanism was explicable for and
restricted to occurring on a particular wheelset or wheelsets only. The possible candidate arcing
mechanisms, including whether and how these were eliminated, are described thus (noting that all but the
first are equally applicable to Down and Up line system gaps):
i.
Since removed Rescue Rail (Down line only) - the earlier presence of a (de-energised)
rescue rail gave rise to the possibility of some arcing since it provided a ‘false’ positive DC
feed to the rear 2 cars (DMOSA & MOS) via the DC positive bus line of the front 2 cars for a
period of the order 0.2 seconds. This was an incorrect arrangement since the traction return for
most wheelsets of the rear 2 cars is for that duration positioned on the wrong (ELL) side of the
IBJs. This resulted in auxiliary current (and traction current if line controller not in neutral
position) being drawn momentarily, which could have been contributing to arcing on the 10th
wheelset of the train (2nd wheelset of MOS car) as this passes over the IBJ. However no arcing
on this wheelset had been positively observed and following removal of the rescue rail the
degree of arcing was found not to have diminished. It was therefore unlikely to have been
contributing to arcing on the Down line. Nevertheless it was helpful to eliminate this to enable
focus on other possible causes.
ii.
Stray/leakage current flow across IBJ when bridged by train – the overall circuit path for
this current flow is formed by the leakage resistance to earth of the running rails on each of
ELL and NR systems interconnected by general earth (and any other earth paths), which is
completed when the IBJ is bridged by a train. This current flow had been predicted earlier in a
report by the designer [1] with a maximum value of 11A calculated under normal traction load
and operating conditions based on a simple Matlab model with a single train on each of ELL
and NR systems and various combinations of traction current. This was for a worst case
(running) rail to earth resistivity value of 1 Ohm.km per track on both systems.
Supplementary (SPICE based nodal analysis) modelling using a simple lumped resistance
model was undertaken by the writer to provide supporting indicative information in respect of
this stray/leakage current flow. This considered a total of four trains drawing various
combinations of (zero or maximum) traction current positioned on either side of the system
gap and at locations beyond the ‘first’ traction substation on each of the ELL and NR systems.
High and low values of (running) rail to earth resistivity were modelled on ELL and NR
systems respectively, as might be typical for new and existing established railway systems.
Prior to bridging across the IBJ in the model to obtain predicted current levels, differences
between the rail to earth voltages on ELL and on NR systems in the model were of a similar
order magnitude to those measured on site as reported earlier.
Figure 8a: Diagram of SPICE model, no
bridging
Figure 8b: Diagram of SPICE model, IBJ
bridged
With the ‘as designed’ high value of rail to earth resistivity on the ELL system (of circa 50 to
100 Ohm.km per track) the predicted stray/leakage current flow across the IBJ when bridged
was limited, predominantly by the high rail to earth resistance of the ELL system, to be of a
very low order of magnitude not exceeding under any scenario a current level of the order 1A.
This would not be of a sufficient level to explain the arcing. However when an additional
discrete low value rail to earth resistance component in the vicinity of Canal Junction traction
substation or New Cross Gate depot (of the order 1 to 2 Ohms) was introduced, representing a
point at which the ELL rail to earth resistance may be compromised, the current level
significantly increased to approximately 20 to 35A depending on the particular traction load
conditions applied in the model. Furthermore, it was assessed that since the stray/leakage
current is a small proportion of traction load, the circuit for which contains significant
inductance associated with the Line and Input Filter chokes of the train, it follows that the
effective circuit path for stray/leakage current is also inductive albeit complex to analyse. This
current level and circuit inductance would be sufficient to explain the arcing.
An initial conclusion was drawn that if the rail to earth resistivity of the ELL system is ‘as
designed’ then stray/leakage current flow across the IBJ when bridged will be very low and in
all probability insignificant in terms of causing arcing effects. However, if there should be any
points of low rail to earth resistance on the ELL system then these could causing appreciable
stray/leakage current flow across the IBJ when bridged thus representing a possible cause of
arcing as (potentially any or all of) the trains wheelsets cross the IBJ.
It was confirmed that, further to some original issues identified and addressed in early 2010,
no similar unintentional or inadvertent ‘earthy’ connections to the ELL running rails were
identified including those within the interconnected New Cross Gate depot, these having been
thoroughly inspected throughout their entire length. However subsequent review and
investigations into stray current on the ELL [2] questioned the accuracy of earlier rail to earth
resistivity measurements for track sections (of greater than 100 Ohm.km) particularly for
‘existing’ slabtrack and identified particular issues of greater than expected stray/leakage
current in the ELL Thames Tunnel. This had not been replaced as part of the ELL Project,
having been rebuilt by London Underground in 1997.
Further investigation and measurements within the Thames Tunnel have been recently
undertaken [3] which have confirmed very low resistance between the running rails and the
stray current drain system in the track slab, and that the trackslab itself is excessively wet.
This track slab was installed in 1996-98, and its stray current drain mat has been linked into
the new ELL stray current drain system. (Note the stray current mats are not normally
connected back to the traction substation negative busbars; this is only done for test purposes
or as a short-term measure to deal with specific issues). To mitigate the problem in the short
term, the sections of the drain mat within the Thames tunnel will be disconnected, the slab
drainage inspected and cleared as necessary and the rail baseplates and their fixings cleaned as
there is also evidence some are excessively contaminated. A further problem has also been
identified in the adjacent Rotherhithe station. Further investigations will be undertaken to what
if any other mechanisms are compromising the rail to stray current drain mat resistance.
This supports the possibility that the rail to earth resistivity (or negative to earth resistance) on
the ELL side of the system gap is being compromised to be well below its intended 100
Ohm.km per track design level, and therefore of arcing taking place by this mechanism.
Regarding which wheelsets could be experiencing arcing by this mechanism it was determined
for the three powered cars (DMOSA, MOS and DMOSB) that, due to the configuration and
low resistance and inductance of traction return connections to all wheelsets on each car,
arcing on the final wheelset of each car only (Nos. 4, 12 and 16; for the Down line system gap
with ‘PTOS to the south’) might be expected. Also that arcing on PTOS wheelsets, which do
not form part of the traction return arrangements and are insulated from each other, may be
non-existent or very minor only (possibly of a level not visible under daylight conditions).
This is because these will therefore only individually (and momentarily) bridge the rails on
either side of the IBJ and as a result the current value reached in this short time period, which
is that which would be broken during arcing, is likely to be limited by the ‘time constant’ for
the inductive stray/leakage current path.
iii.
Rail to earth system capacitive discharge/equalisation across IBJ when bridged – as a
consequence of rail insulation measures in place on both systems there exists distributed
capacitance between rails and earth which is charged to the rail to earth voltage, and
charges/discharges with changes in such voltage. The value of this capacitance is likely to be
much higher (relatively) for the new ELL system than the existing NR system. Therefore it
was considered possible that when the two ELL and NR rail systems are connected together
by bridging across the IBJ some capacitive current flow takes place from the system with
higher rail to earth voltage as the voltage is equalised between the two rail systems. The
existence of appreciable differences between rail to earth voltages on ELL and on NR systems
(both measured and predicted) would support this.
This could be appearing as arcing current flow at the wheelsets as these cross the IBJ, the
magnitude of which is not yet predicted and would need further investigation, including
establishing likely values of rail to earth capacitance. It was expected that the RC time
constant of the discharge/equalisation circuit is not significant such that current flow and
equalisation would take place fairly rapidly. In this case such arcing would be expected as the
first few wheelsets only cross the IBJ without arcing being exhibited on wheelsets of
subsequent cars as these pass over the gap.
iv.
Internal train effects during transition between loss of (conductor rail) power and
‘partial/self’ regeneration mode, when either coasting or motoring – it was envisaged that
there could be internal effects within the train as it enters the system gap during the time
period immediately following loss of connection to the conductor rail. This was in relation to
the very short time interval (around 0.5 seconds) between the rearmost shoegear leaving the
conductor rail and the front wheelset of each 2 car unit crossing the IBJ, which depended on
the detailed behaviour of and transition/ramp times for the train traction power equipment.
For the front (DMOSB and PTOS) 2 car unit it was difficult to envisage how any such internal
train effects could contribute to arcing since there is only one MCM and ACM (co-located on
DMOSB) and therefore no (circulating current) circuit between the 2 cars which would
otherwise use the running rails as the negative return path. Furthermore, as the MCM and
ACM of the DMOSB car are connected between the same positive DC bus line and negative
wire (for that car), with radial connections from the negative wire to the axle return brushes, it
was judged that there could be no internal train effects for the single DMOSB car, and
therefore no issues for the front 2 car (DMOSB and MOS) unit.
For the rear (DMOSA and MOS) 2 car unit as there are two MCMs and one ACM there exists
the possibility of such internal effects (for example relating to a differential in line filter
capacitor voltages, and the time for these to equalise and become stable, as part of
transitioning to the ‘target value’ of DC bus voltage for ‘partial/self’ regeneration) resulting in
a circulating current flow between the two cars which would involve the running rails as the
negative return path. The duration of such effects would need to be significant and of the order
2.5 seconds in order for the last of the four wheelsets of MOS car to pass over the IBJ thereby
severing the rail return path between cars and giving rise to appreciable levels of arcing. At
most, minor arcing on the first few wheelsets of MOS car might be predicted as the impedance
of the return path via the vehicle negative wiring changes with successive wheelset passes.
However, discussions with BT confirmed that the response time for the train traction power
(line filter and MCM) equipment to transition from coasting, motoring or regenerative braking
state to the ‘partial/self’ low level regenerative mode to support ACM auxiliaries (and vice
versa) is of the order of ‘tens of milliseconds’ and therefore too short to be a cause of any
arcing. This was further confirmed through ‘non-intrusive’ testing undertaken subsequently
with MOS car isolated (reported later), under which arcing was still evident.
v.
Internal train circulating current between DMOSA and MOS cars (of rear 2 car unit) as
train passes through system gap – this could be due to any minor voltage equalisation
between the line filter capacitors of the MCM in each of these 2 cars, and/or due to the MCM
in MOS car supporting the ACM auxiliary power load (located in DMOSA car) during
‘partial/self’ regenerative mode as the train crosses the system gap. This circulating current
would use the running rail as the negative return path and thereby give rise to the possibility of
arcing, albeit principally only on the 12th wheelset of the train (4th wheelset of MOS car).
Arcing on other wheelsets, if present, would be minor only as the impedance of the return path
via the vehicle negative wiring changes with successive wheelset passes.
In was confirmed by BT that, during passage of a train through the gap, the MCM in the MOS
car will provide an auxiliary current contribution to the ACM in DMOSA car, with a
circulating current path, the return part of which is via the running rails between the traction
return wheelsets of these two cars. Under maximum HVAC conditions (e.g. high summer
ambient temperature with heavy passenger load) this current level could be of the order 30 to
40A; however - at the time of the observations and voltage measurements, with moderate
ambient temperatures and without passenger loading - the current level would have been much
lower than this and was considered not significant in relation to arcing. However, some
contribution to arcing on the 12th wheelset (presently or during future operation) due to
circulating current between MOS and DMOSA cars was acknowledged by BT as a possibility.
Also, as a variation on this candidate mechanism and in relation to arcing on the 12th wheelset,
BT acknowledged that there could be some charge/discharge equalisation effect between the
line filters/MCMs on each of MOS and DMOSA cars due to a change in the return rail voltage
reference for MOS car as it passes from ELL to NR side of the IBJs, noting that such
differentials in rail voltage are present and have been measured. As for the auxiliary current
contribution to the ACM, this would employ a circulating current path the return part of which
is via the running rails between the traction return wheelsets of these two cars.
Note that the possibility of any resonance (ringing) between the line filter capacitors of the two
MCMs on DMOSA and MOS cars, associated with any differentials in capacitor voltage and
resulting charge/discharge equalisation, as the train passes through the system gap (as had
been earlier suggested) was considered by BT to be highly unlikely due to the damping effect
of the various filters.
For all of the above arcing mechanisms, arcing across IBJs of both rails would be predicted. This includes
the Down line gap case where there is single rail traction return on the ELL approach side of the IBJs.
This is because, with the exception of the last (16th) wheelset, the (solid) axles of the wheelsets already
within the first half of system gap will effectively bridge both (traction and signal) rails together such that
passage of current between the ELL and NR system as each wheelset passes over the IBJs will be
possible across both IBJs. Although initially it was not fully understood why arcing is observed
predominantly on the cess side rail of the Down line gap, it is now firmly believed that this is due to a
slight offset in the longitudinal position of the two IBJs. Photographic evidence does suggest minor
occasional sparking on the ‘10 foot’ side IBJ but this is thought to be due to the occasional wheelset that
had a slight skew from the perpendicular due to the characteristics of the bogie/primary suspension.
PROBABLE CAUSES OF ARCING
Based on those candidate mechanisms above which remained as possibilities, and the direct observations
up to that point as to which wheelsets were experiencing arcing, an interim conclusion was drawn that the
most probable cause was:
i.
arcing on the 4th wheelset of DMOSB car of the front 2 car unit due to stray/leakage current
flow (in conjunction with a low rail to earth resistance on ELL system, possibly due – inter
alia - to issues in Thames Tunnel) [mechanism ii] and/or capacitive discharge/equalisation
[mech. iii] between ELL and NR systems. Similarly, arcing on the 16th wheelset of DMOSA
car of the rear 2 car unit due to the former [mech. ii] only. This is on the basis that, critically,
there was no identified credible mechanism arising from internal operation of the train, or its
interaction with the railway infrastructure, which would explain arcing on wheelsets 4 and 16;
ii.
arcing on the 12th wheelset of MOS car of the rear 2 car unit due to stray/leakage current flow
(in conjunction with a low rail to earth resistance on ELL system) [mech. ii] and/or internal
(train) circulating current flow between the two traction packages (MCMs) on DMOSA and
MOS cars (possibly involving support to the ACM) [mech. v].
To further understand the mechanism by which arcing was occurring, and confirm whether or not there is
any contribution from the train to arcing on the 12th wheelset, a series of ‘non-intrusive’ and ‘intrusive’
testing to be undertaken for the Down line gap was proposed. The details of these tests were formulated in
conjunction with BT.
TESTING FOR DOWN LINE GAP
The ‘non-intrusive’ tests were devised to determine the contributory effect to the IBJ arcing of any power
drawn by vehicles traversing the gap, compared to the “background effect” of other trains drawing power
on the ELL and NR systems at the time the IBJs in the centre of the gap are bridged by wheelsets. The
tests involved the 4 car ‘Class 378/1 DC only’ test train operating with different combinations of traction
and auxiliary equipment cut-out on individual vehicles in turn over a series of pre-service timetabled runs.
This had the effect of individually varying the current drawn from the conductor rails by the three
powered vehicles of the train either side of the gap. Under these tests, voltage (potential) difference across
the cess side IBJ was measured/recorded (with a 10kHz sampling rate) and high speed video recording
undertaken. The results of these tests, conducted in May 2010, are reported below.
The proposed ‘intrusive’ tests, which are yet to be conducted, involve the introduction of a bond
connection across the cess side rail IBJ which is made/broken and through which current flow (with 0 to
10kHz frequency response) is measured/recorded, together with voltage difference as for the ‘nonintrusive’ tests. This will be undertaken ‘without train passage’ under which making/breaking of the bond
is applied to simulate train wheelset passage. In addition this will be undertaken ‘with train passage’,
under which firstly the bond is in place for a time period during which the train passes and secondly
removed and re-applied just prior to and following train passage respectively.
RESULTS OF NON-INTRUSIVE TESTING
The measured voltage (potential) difference traces for each train passage clearly exhibited arcing voltage
spikes time coincident and consistent with arcing on the 4th, 12th and 16th wheelsets, examples of which
(for two scenarios where the train was coasting and powering through the gap) are presented in Figure 9.
The voltage difference across each IBJ was measured with ELL side as positive with respect to NR side
as negative.
Figure 9a: Voltage Difference Trace (10kHz
sampling) for Test Train 2, coasting
Figure 9b: Voltage difference Trace (10kHz
sampling) for Test Train 7, powering
The general trend, for all runs, was that the voltage difference was positive and seen to be increasing
(with a maximum mean value of the order 10-15V) as the train approached the gap from NXG platform.
The voltage difference then turned negative as the train departed the gap on the NR side. This is
consistent with the train drawing power from ELL and NR DC conductor rail systems respectively
causing a local rise in rail voltage on its respective rail return system. For the ELL side this is consistent
with the train drawing increasing power/current as it accelerates away from NXG platform. Also, a
collapse in the voltage difference to around 0 Volts is generally observed as the first wheelset crosses the
IBJ at which point (and beyond) the two rail return systems are then interconnected.
There was evidence of some background interference in the test leads and/or trackside located measuring
equipment particularly during the periods over which each of the 4 cars crossed the IBJ, broadly
consistent with the motor cars and level of HVAC that were in operation. This may have been due to the
emissions from on-board traction and/or auxiliary equipment as it passed the point of measurement.
Without a local train passing through the gap a ‘background’ voltage difference exists across the IBJs
generally negative in polarity and of the order -10 to -15V. This was consistent with the location of ELL
and NR substations in relation to the Down line gap, which predicts that the ELL and NR rail voltages
would generally be negative and positive at this location respectively.
Variations in the polarity and magnitude of arcing voltage spikes for successive test runs and different
train operation modes, under which MOS car traction and/or HVAC equipment were in several cases ‘cut
out’, are summarised in the Table 2:
Test Departure
No
time from
NXG - actual
1
2
3
4
5
6
7
Train
Operation
Mode
Potential Difference* –
arcing voltage spikes &
polarity (number refers
to wheelset)
Failed to record
High Speed Video – Arcing
level observed (number
refers to wheelset)
MOS off
16th very small
(coasts
through
gap)
1015
Normal
4th negative (-84.8V peak) Failed to record
(coasts
12th negative (-74.8V)
through
16th negative (-88.5V)
gap)
1114
Normal
4th positive (+56.7V)
4th very small
th
(on power
12 negative (-88.3V)
12th medium
through
16th negative (-95.4V)
16th ‘medium+’
gap)
1145
MOS off
4th positive (+76.8V)
Failed to record
(on power
12th negative (-68.2V)
through
16th negative (-86.2V)
gap)
1214
HVAC off
4th positive (+73.1V)
4th small
th
(on power
12 negative (-106.6V)
12th medium
through
16th negative (-111.3V)
16th ‘medium+’
gap)
Cancelled
---------1346
MOS and
4th positive (+74.2V)
4th very small
HVAC off
12th negative (-80.2V)
12th medium
th
(on power
16 negative (-91.3V)
16th ‘medium+’
through
gap)
Table 2: Results of polarity and magnitude of arcing voltage spikes
(*across cess side IBJ however results for other rail very similar)
0944
The table also includes whether or not each test train was coasting or powering through the gap, based on
scrutinising downloads from the On Train Monitoring Recorder (OTMR) for the test trains which
contained time based traction power notch selection data, an example of which is shown in Figure 10:
Figure 10: Example OTMR Download (30 seconds after departing NXG platform) for train with coasting
Several key conclusions were drawn from these test results, thus:
i.
There was no evident impact on the existence, magnitude or polarity of the arcing voltage
spikes from train operation in modes other than normal operation, whereby MOS motor car
and/or HVAC are turned off. Therefore – with whatever levels of auxiliary current were in
place during testing – circulating current effects (or any other effects due to auxiliary current
flow) on board the train [identified mechanism v] are not affecting or contributing to arcing.
The consequence of this was that arcing is likely to be due to stray/leakage current flow
between the two rail return systems [mech. ii], and this being broken as each of wheelsets 4,
12 and 16 crosses the IBJs. The existence of a measured voltage difference between ELL and
NR systems, which confirms earlier measurement results, supports this.
ii.
The generation of an arcing voltage, of the order approaching 100V, was considered to be due
the stray/leakage current being a small proportion of the traction current (due to train(s) in
either the ELL or NR system), for which the overall electrical circuit contains significant
inductance present within the train(s’) traction package(s). The reduction in current gives rise
to a (negative) Ldi/dt voltage component within that inductance which, in terms of the overall
circuit, appears as an arcing voltage spike across the IBJ as the wheelset crosses. Review of
the polarity of the arcing voltage spikes indicated that these were generally positive for 4th
wheelset and negative for both 12th and 16th wheelsets. An arcing voltage (as measured across
the IBJ) of positive polarity would be consistent with a direction of stray/leakage current flow
from ELL side to NR side. Conversely an arcing voltage of negative polarity would be
consistent with a direction of stray/leakage current flow from NR side to ELL side.
iii.
On the basis that test trains 3,4,5 & 7 were powering through the Down line gap (as
established through OTMR downloads) at the point the 4th wheelset is crossing the IBJ the 2
rear (DMOSA and MOS) motor cars would still be drawing traction power from the ELL side
conductor rail. This would cause the ELL running rail voltage to be generally more positive
than the NR rail voltage. Therefore a component of stray/leakage current flow due to the test
train itself would then flow across the IBJ from the rails on ELL to the rails on NR side
(returning to the ELL negative return system by means of the leakage resistance of NR and
ELL systems via earth). Conversely, for the points at which the 12th and 16th wheelsets cross,
the situation is reversed with the front 2 cars (DMOSB motor car) drawing traction power
from the NR side conductor rail. This would cause the NR running rail voltage to be generally
more positive than the ELL rail voltage such that a component of stray/leakage current flow
due to the test train itself would then flow across the IBJ from the rails on NR side to those on
ELL side. The general results for voltage difference as the train approaches and departs the
gap were consistent with this. Also it was significant that during test 2 (with coasting) this
voltage reversal is not seen due to the fact that the train is coasting through the gap, nor is the
associated predicted pattern of arcing voltage polarity (as seen for the trains which were
powering) exhibited. Similarly test 1 (with coasting) exhibited little visible arcing. This was
considered to be a highly significant finding, and strongly suggested that the effect of the train
powering through the gap is greatly contributing to the magnitude and severity of arcing that is
occurring due to (its own component of) stray/leakage current flow. This in an inherent feature
of the gap design and not of the train itself. Implementation of coasting through the gap,
instructed by trackside coasting boards, was therefore recommended.
iv.
Lesser transient arcing voltage spikes were seen during other train movements in the area.
The nature of these suggested that they were generated by the passage of Class 378/1 trains
crossing the the Up line system gap since they had a reversed timing signature associated with
arcing on the 4th, 12th and 16th wheelsets (since the train is operating in the reverse direction).
This was consistent with the understanding of arcing developed through this investigation and
the probable mechanism by which this is taking place, and was predictable on that basis.
CONCLUSIONS
The results of the ‘non-intrusive’ testing were extremely useful in eliminating the train as a potential
contributor to the arcing on the 12th (or any other) wheelset by mechanism v, and provided some
confirmation that the arcing on 4th, 12th and 16th (or any other) wheelsets is due to stray/leakage current
flow between ELL and NR systems (mechanism ii). In addition, identifying that the train itself whilst
powering through the gap may be greatly contributing to the magnitude and severity of arcing by this
mechanism, resulted in the recommendation that a coasting regime should be implemented.
This is in the context of an ELL rail to earth resistivity which is compromised by a point(s) of low rail to
earth resistance to be well below its 100 Ohm.km intended design level, the possibility of which has been
identified in the area of the Thames Tunnel (for which remedial actions are proposed) and is strongly
supported by [2] and [3]. The proposed ‘intrusive’ testing still to be undertaken is expected to provide
further confirmation of arcing being by this mechanism and also information regarding the level of arcing
current.
A number of possible solutions have been outlined. These require further design development for
implementation as a permanent measure for the interface at New Cross Gate on the ELL, noting that
opportunity for option iii may be limited, and include:
i.
Arc suppression equipment connected across the IBJ (likely required on cess side rail only),
such as in the form of a current limiting resistor, snubber arrangement (e.g. RC network), or
(voltage selectable) diode banks and/or Metal Oxide Varistor combination.
ii.
A normally open switching device connected across the IBJ, to be closed during train passage
then re-opened and synchronised via signalling train detection, such as in the form of a
Controlled Track Switch (re-used from abandoned ‘rescue rail’ application) or solid state
(power electronic) switch.
In addition these require to be developed in consultation with the ELL Phase 2 team with regard to the
future similar interface (between ELL and NR systems) at Old Kent Road Junction, where a similar
arcing problem is predicted. In this case the opportunity for a further option may exist as that enters the
detailed design stage with which the author may be involved. This option consists of a revised longer gap
arrangement with (minimum) 80m long central ‘floating section’ and additional set of IBJs to present
cross-connection of ELL and NR rail return systems by a Class 378/1 train (or Class 378/2 dual voltage
also operating). With this arrangement, traction power could be continued to be drawn and then re-applied
such that coasting distance would be around 80m. Feasibility of this would depend on gradients, train
speeds, and signal stopping positions to ensure the risk of train stranding in the gap was acceptably low.
This option would not be practical for the existing New Cross Gate Down line gap having a 1 in 73 up
gradient not far away from the station, with attendant low value train speeds; nor would it generally be
suitable for retrofitting to an existing gap.
Since undertaking this investigation a similar interface, with the same 80m system gap configuration on
Up and Down lines, has been introduced at the far north end of the ELL as part of Phase 1a works. In this
case the ‘system’ on the other side of the gap is a short (around 2km long) further two track section of the
ELL where this runs parallel to (and will in future interface via a dual electrified ‘transfer’ track with) the
25kV AC overhead line electrified North London Line. This section contains two stations, terminating
beyond the second of these, and uses some existing (previously NR) track having rail to earth resistivity
levels expected to be of low order value. Also there is a new traction substation local to and with a
rectifier connected immediately either side of the gaps, without interconnection between rectifier outputs.
Interestingly, however, no arcing or arcing effects have been observed at this interface with regular
passage of the same Class 378/1 trains. This appears to support the train not contributing by mechanism
v. Regarding an absence of stray/leakage current flow (which would otherwise result in arcing by
mechanism ii), it is provisionally considered by the author that this could be due to the limited number of
trains (some of which could be berthed at a station) and therefore low level of traction/auxiliary load in
this section, and/or due to the minimal influence of any point of low rail to earth resistance (such as
Thames Tunnel) towards the south end of the ELL resulting from the appreciable intermediate (around
8km) distance. Also this could be due the effect of the local substation, having two rectifiers each feeding
separately to one side only of the system gap IBJs, in controlling rail to earth voltages and the flow of
traction return and associated stray/leakage current. This requires further understanding to be obtained,
including whether or not trains are coasting or powering through the gaps, which could provide useful
input into the development of solution options.
The results of the investigation have demonstrated that interactions at interfaces between traction power
systems where the train is not under traction power, but is able to bridge between rail return systems, can
be complex and subtle in nature and therefore may not be readily predictable at the design stage. They
require careful consideration and planning based on practical experience of similar interfaces on other
systems such as at this interface on the ELL, whilst taking into account the possible significant impact of
any differences in application in terms of traction power system, traction return, and train configurations.
ACKNOWLEDGEMENTS
The author wishes to thank the Director, London Overground Infrastructure, for granting permission to
publish this paper, and A G Shepherd for assistance in reviewing and finalising this paper.
REFERENCES
1)
Mott Macdonald ‘80 Metre Gap (Down Line) Traction Power Interface Modelling Report of
Electrical Conditions across IBJs’ ref ELM-TEC-209-14-08-0005 Issue 1 July 08
2)
BBC JV Report ‘Stray Current Investigation and Results Summary’ ref. ELLP-000-REP-0205847-A01, August 2010
3)
TfL LOI IMC Report ‘Stray Current Corrosion Potential Monitoring Brunel Tunnel 22 to 26
January 2011’ Ref CCL-5058-COR-REP-0001-A0