MiCOM P438 Railway Catenary Protection

Application
Railway Notes
Protection
(version –603 with –800 “New” English)

Page 1
RAILWAY PROTECTION
Many high speed, regional and urban rail networks worldwide are electrified, to
provide the motive power for trains. The electrification system serves as the means to
distribute power, and also as the contact interface for current collection by each train.
In general, two philosophies are followed: Power distribution via an overhead catenary
above the track, with current collection by a pantograph; or Live rail electrification,
with current collection via contact shoes. Overhead catenaries are generally viewed to
be safer, as they are above track, out of reach of rail personnel and the public, and so
can utilise high feeding voltages. Typically 11 to 50kV with respect to the running rails
is the nominal feeding voltage, electrified single phase ac, although 1500V and 3kV
dc catenaries are predominant in some countries. Using live rail feeding, the system
voltage is generally 600V to 1700V dc.

These application notes cover protection associated with HV overhead catenary

electrification, using worked examples to aid setting of the most frequently-applied
elements in the MiCOM P438 relay. Due to the nature of the rail route, and limited
electrical clearances, line faults are common, often amounting to one fault per year
for every kilometre of track. The frequency of fault inception, coupled with the high
tension in the contact wire (typically 6 to 20 kN) makes fast fault clearance the ultimate
requirement. Should faults not be cleared quickly, the conductors that make up the
catenary may break due to intense overheating, risking further severe damage.

1. Classical Single Phase Feeding

Classical single phase ac electrification has been used on railways since the 1920s.
Earlier systems used low frequency supplies, such as 25Hz in the North East Corridor
of the USA. In many countries, 16.7Hz and 25Hz electrified systems remain today.

Starting from western Europe and with the influence spreading worldwide,
electrification at 50/60Hz, single phase ac has become the standard. In Figure 1,
25kV classical feeding is shown, with booster transformers (BT) used to force return
current to flow in a return conductor. This serves to limit traction current returning
through the rails and earth, and reduces electromagnetic interference with any
adjacent communication circuits. The traction supply is generally provided by a
transformer connected phase to phase across one phase pair of the utility grid. The
electrical feeding to the train is via the overhead catenary, with the return current
flowing via the rails and the return conductor.
Supply
transformer
Return conductor

Catenary

BT BT

25kV

Rails

Figure 1. Simple 25kV feeding with booster transformers

Page 2
As the running rails are regularly bonded to earth and are nominally at earth
potential, it is only necessary to disconnect the supply to the catenary in the event of a
fault. Thus, single pole switchgear tripping the catenary circuit only, is required.

2. Classical System Feeding Diagram

In practice, single track railway lines are rare, and two or four parallel tracks are more
common. The overhead line equipment is then comprised of two or four electrically
independent catenaries, running in parallel. Figure 2 shows a typical two track
railway electrified with classical single phase ac.

Feeding North
FS SS1 SS2 MPSS
A C E
F1

B D F

F2 BS1
BS2

Feeding South

Figure 2. Classical 25kV feeding diagram

The infeed to the tracks in the “north” direction is via grid transformer F1 at the Feeder
Station (FS). The power is then distributed via catenaries A and B to the location of the
trains. At intervals, it is usual to couple the two catenaries, allowing load current to
flow in parallel paths, and thus reduce the feeding impedance and line voltage drops.
This is done at parallelling/sub-sectioning substations, shown SS1 and SS2 here. As
the terminology implies, providing circuit breakers for each of the outgoing feeds to
the catenaries also allows subsectioning – ie. the ability to disconnect supply from
sections of catenary, in the event of a fault, or to allow maintenance attention. For a
fault on catenary “A”, we would expect to trip circuit breaker A at the feeder station,
and breaker A at SS1. Only then would the infeed to the double-end fed section be
removed, and the feeding to healthy sections B, C, D, E and F would be unaffected.

The infeed from F1 generally feeds only as far as the normally open bus section circuit
breaker (BS2) at the mid-point substation. Beyond the MPSS there is a mirror image of
the feeding shown F1 to BS2, with the remote end feeder station often 40-60 km
distant from F1. BS2 must remain open during normal feeding, to avoid parallelling
of supplies that may be derived from different phase pairs on the utility grid – eg.
Phase A-B at F1, and B-C at the next FS north. The same is true for BS1, which
normally remains open, as F1 and F2 feeds are generally from unlike phase pairs in
an attempt to balance the loading on the three phase utility grid.

Page 3
3. Classical System Protection Philosophy

The grid infeed transformers often are rated at 10 to 25MVA, with a relatively high
reactance (around 10%). Thus even for a fault at the Feeder Station busbar, the
maximum prospective short circuit current is low in comparison to a utility system,
typically only 10 times the rating of a single catenary. Were a fault to occur further
down the track, there would be the additional ohms/km characteristic impedance of
the catenary and return conductor to be added in the fault loop. One ohm per mile
(0.6Ω/km) is typical, meaning that for a fault at the remote end of a protected section
(eg. Catenary section “A” in Figure 2), the current measured at the upstream circuit
breaker location (circuit breaker A at the FS) may be twice rated current. Thus, at
Feeder Stations there is the possibility of applying overcurrent protection, as there is a
sufficient margin between the maximum continuous load current, and the fault current
level at the remote end of sections.

However, overcurrent protection is often used only as time-delayed back-up protection

on railways, for the following reasons:

• The protection needs to be discriminative, and ensure that only the two circuit
breakers associated with the faulted line section are tripped. This places a
requirement that the protection is directionalised, to respond to fault current
flowing into the section, and not to respond for reverse fault current flow. At
location SS1 for example, this means that the protection for catenaries A and B
looks back towards the grid infeed. It can be appreciated that for a fault close to
the FS on catenary A, the remote end protection will measure only the proportion
of fault current that flows via unfaulted catenary B, and then a “hairpin” path to
SS1 and back along catenary A to the location of the fault. This fault current
contribution may be less than rated load current.

• For protection located at SS1, SS2 and MPSS, the prospective fault current levels
are progressively smaller, and again can result in the measured fault current being
lower than rated current.

• During outages of grid supply transformers, alternative feeding may be necessary.

One possible feeding arrangement is to extend the feeding by closing the bus
section circuit breaker at the MPSS. Beyond the MPSS the prospective fault current
levels will be much lower than normal.

Distance protection has been the most proven method of protecting railway
catenaries, due to its inherent ability to remain stable for heavy load current, whilst
tripping for even low levels of fault current. Later in this section it will be explained
how fault current generally lags the system voltage by a greater phase angle than is
usual under load conditions, and this phase angle measurement is an important part
of a distance relay’s trip characteristics.

Page 4
4. Distance Protection Zone Reaches

Distance relays applied to a classical single phase system have two measurement
inputs: a catenary to rail voltage input derived from a line or busbar connected
voltage transformer, and a track feeder current input derived from a current
transformer for the circuit breaker feeding the protected section. Distance relays
perform a vector division of ( Voltage (V) / Current (I) ) to determine the load or fault
impedance (Z). Typical relay characteristics are shown in the R + jX impedance plane,
Figure 3.

X Z3
Max normal
load area

Z2

Z1
R

Max regenerative load area

Figure 3. Polar Impedance Plot of Typical Trip Characteristics

Faults on the catenary will present an impedance to the relay along the dotted line in
Figure 3, and the quadrilateral characteristics shown are directionalised to match the
expected fault impedance of usually 70 to 75 degrees leading the resistive axis. The
closer the fault is to the relay location, the lower the measured fault impedance will be,
and thus the trip decision is given when the measured fault impedance falls within the
tripping zones. Three zones (shown Z1, Z2, Z3) are commonly applied, with the reach
along the resistive axis being constrained, such as to avoid tripping for load current.
All impedance reaches for distance zones are calculated in polar form: Z ∠α, where
Z is the reach in ohms, and α is the line angle setting in degrees. For railway systems,
where all catenaries share the same line angle, it is often convenient to add and
subtract section impedances algebraically, treating Z as a scalar quantity.

! Zone 1

The zone 1 elements of a distance relay are designed to protect as much of the
immediate catenary section as possible, without overreaching and tripping for faults
out of section. In most applications as the zone 1 reach (Z1) cannot respond to faults
beyond the protected catenary, instantaneous operation is set (ie. no deliberate time
delay for tripping). For an underreaching application the zone 1 reach must therefore
be set to account for any possible overreaching errors. These errors come from the
relay, the VTs and CTs and inaccurate line impedance data. It is therefore
recommended that the reach of the zone 1 elements is restricted to 85% of the
protected line impedance, with zone 2 elements set to cover the final 15% of the line.

Page 5
! Zone 2

Zone 2 elements should be set to cover the 15% of the line not covered by zone 1, and
are time-delayed in their operation. Allowing for underreaching errors, the zone 2
reach (Z2) should be set in excess of 115% of the protected line impedance for all fault
conditions. It is often beneficial to set zone 2 to reach further than this minimum, in
order to provide faster back-up protection for uncleared downstream faults. A
constraining requirement is that zone 2 does not reach beyond the zone 1 reach of
downstream catenary protection. This principle is shown in Figure 4, for a four track
system, where the local breaker for section E has failed to trip.

A
B NORMAL
FEEDING
C
D 70% E

A = Protected Section Impedance

E = Shortest Following Section

Z = Equivalent Impedance of A, B, C, D in Parallel

R = Equivalent Impedance of B, C, D in Parallel

Figure 4. Fault Scenario Considered for Zone 2 Reach Constraint

(Normal Feeding)

In order to calculate Z2 for the left-most circuit breaker of protected catenary “A”, a
fault is imagined to occur at 70% of the shortest following section. This is the closest
location that unwanted overlap could occur with Z2 main protection for catenary E.
The 70% figure is the nominal 85% Z1 reach for catenary E protection, minus 15% for
composite measuring errors at that location. The apparent impedance of the fault, as
viewed from catenary A is then calculated, noting that any fault impedance beyond the
right-most busbar appears to be approximately four times its actual ohmic impedance,
due to the fault current parallelling along four adjacent tracks. The setting applied to
the relay is the result of this calculation, reduced by a further 15% to accommodate
any measurement errors at relay A location.

The equation for the Zone 2 reach becomes:

Z2 = (Z + 0.7 E) x ((A + R)/R) / 1.15

The possibility of current following an out and back hairpin path to faults has already
been described, and it is essential that the relay does not overreach under these
conditions too. The feeding scenario is shown in Figure 5.

Page 6
 A
B
C HAIRPIN
FEEDING
70% D

A = Protected Section Impedance

D = Shortest Hairpin Fed Section

Z = Equivalent Impedance of A, B, C, in Parallel

R = Equivalent Impedance of B, C, in Parallel

Figure 5. Fault Scenario Considered for Zone 2 Reach Constraint

(Hairpin Feeding)

Figure 5 depicts a fault that has been cleared at one end only, with the remote end
breaker for section D failing to trip. The fault is assumed to be on the lowest
impedance catenary, which is an important consideration when there are more than
two tracks. In a four track system it is not uncommon for mutual induction to cause
inner (middle) track catenaries to have an ohms/km impedance 13% higher than for
the outside tracks.

The calculation principle is similar to that for normal feeding, except that now the fault
current is parallelling along three (= number of tracks minus one) adjacent tracks.
The three catenaries concerned are the protected catenary A, and the remainder of the
unfaulted catenaries (R), ie. catenaries B and C.

The equation for the hairpin Zone 2 reach becomes:

Z2 = (Z + 0.7 D) x ((A + R)/R) / 1.15

To avoid overreaching for both normal feeding and hairpin fed faults, the lowest of
the two calculated impedances is used as the Zone 2 reach setting.

! Zone 3

Zone 3 elements would usually be used to provide overall back-up protection for
downstream catenary sections. The zone 3 reach (Z3) is therefore set to at least 115%
of the combined apparent impedance of the protected catenary plus the longest
downstream catenary. Figure 6 shows the feeding considered:

Page 7
 A
B NORMAL
FEEDING
C
D 100% E

A = Protected Section Impedance

E = Longest Following Section

Z = Equivalent Impedance of A, B, C, D in Parallel

R = Equivalent Impedance of B, C, D in Parallel

Figure 6. Fault Scenario Considered for Zone 3 Minimum Reach

(Normal Feeding)

The equation for the Zone 3 reach (normal feeding) becomes:

Z3 = (Z + E) x ((A + R)/R) x 1.15

It can be appreciated that hairpin feeding scenarios too must be considered, and this
is depicted in Figure 7:

A
B
HAIRPIN
C FEEDING
100% D

A = Protected Section Impedance

D = Longest Hairpin Fed Section

Z = Equivalent Impedance of A, B, C, in Parallel

R = Equivalent Impedance of B, C, in Parallel

Figure 7. Fault Scenario Considered for Zone 3 Minimum Reach

(Hairpin Feeding)

Page 8
The equation for the Zone 3 reach (hairpin feeding) becomes:

Z3 = (Z + D) x ((A + R)/R) x 1.15

To avoid underreaching for both normal feeding and hairpin fed faults, the highest of
the two calculated impedances is used as the Zone 3 reach setting. Occasionally the
zone 3 reach requirement may be raised further, to offset the effects of regenerative
braking trains providing an additional current infeed to the fault. However, an
additional 5% reach increase is generally easily sufficient.

! Reverse Reaching Zones

Reverse directional zones typically provide back-up protection for the local busbar,
where the offset reach is set to 25% of the zone 1 reach of the relay. Typically Zone 3
is set with a reverse reach in this way to also satisfy the requirements for Switch on to
Fault protection.

5. Distance Zone Time Delay Settings

The zone 1 time delay (tZ1) is generally set to zero, giving instantaneous operation.

The zone 2 time delay (tZ2) is set to co-ordinate with zone 1 fault clearance time for
downstream catenaries. The total fault clearance time will consist of the downstream
zone 1 operating time plus the associated breaker operating time. Allowance must
also be made for the zone 2 elements to reset following clearance of an adjacent line
fault and also for a safety margin. A typical minimum zone 2 time delay is of the
order of 200ms. This time may have to be adjusted where the relay is required to
grade with slower forms of back-up protection or circuit breakers for downstream
circuits.

The zone 3 time delay (tZ3) is typically set with the same considerations made for the
zone 2 time delay, except that the delay needs to co-ordinate with the downstream
zone 2 fault clearance. A typical minimum zone 3 operating time would be in the
region of 400ms. Again, this may need to be modified to co-ordinate with slower
forms of back-up protection for adjacent circuits.

6. Load Avoidance

Figure 3 shows how the distance relay trip characteristics must avoid regions of the
polar plot where the traction load may be present. This has historically been achieved
by using shaped trip characteristics, such as the lenticular or lens aspect. Since the
1990’s, the benefits of applying quadrilateral characteristics have been realised,
whereby the right-most line (resistive reach line) of the quadrilateral is used as a load
blinder. Quadrilateral (or polygon) characteristics permit the resistive reach line to be
set independently of the calculated zone reach, which sets the position of the top line
of the quadrilateral. The resistive reach is then calculated merely to avoid the traction
load by a suitable margin, and provide as much coverage for faults of a resistive
nature as is possible.

Page 9
Figure 8 shows how the resistive reach settings are determined:

X
C

LINE
100

Z3

Z2
REGEN.
Z1 D E F
B MOTORING
LOAD
70O A
100 R
Primary ohms

Figure 8. Resistive Reach Settings for Load Avoidance

For polygon characteristics, impedance point B is the critical loading to avoid. The
magnitude of the impedance is calculated from Z = V / I , taking the minimum
operational catenary voltage and the maximum short-term catenary current. It should
be noted that it is not uncommon for the catenary voltage to fall to 80% of nominal,
and for short term loading to measure 160% of nominal – these worst-case measured
values should be used when aiming to find the lowest load impedance.

In order to find the phase angle of B with respect to the resistive axis, this is
determined as:

θ = Cos –1 (Maximum lagging power factor)

The diagram shows how resistive reach E-F for Zone 1 has been chosen to avoid the
worst-case loading by a suitable margin of 10 to 20%. Zones 2 and 3 reach further,
thus the effect of any angular errors introduced by CTs, VTs etc. will be more
pronounced. It is therefore common to set the resistive reaches progressively
marginally smaller for zones with longer reaches. A practical setting constraint to
ensure that zones with long reaches are not too narrow, and not overly affected by
angle measurement tolerances, is for the resistive reach not to be less than 1/7 th of
the zone reach.

Page 10
7. Enhanced P438 Relay Characteristics

Figure 9 shows typical use of polygon characteristics to enhance the protection offered
by P438 catenary distance protection.

Forward
X
Z1 fwd

β1
αL1
R1
fwd
R1 rev

γ1

Reverse

Figure 9. Polygon Distance Characteristics

Certain systems may demand reverse operation of zones, although for the Zone 1
characteristic shown, forward only operation is common. For systems with
regenerative braking, trains usually regenerate at a leading power factor to avoid the
creation of overvoltages on the catenary. Where a regenerating train contributes to
fault current, the fault impedance measured by distance relays may shift up to 10
degrees greater than α. To ensure that the fault impedance remains within the trip
characteristic, and does not stray outside the top left hand resistive boundary of the
polygon, the reverse resistive reach (Rrev) can be set greater than the forward resistive
reach (Rfwd). For most applications however, Rfwd is set equal to Rrev.

Applicable settings (example for Zone 1, are):

Location in menu: Settings/Function Settings/Setting Group SGx/DIST:

Cell (for Description Purpose

SG1)
076.093 Z1 Forward Forward zone reach in secondary ohms, at line
angle α
076.100 Z1 Reverse Reverse reach, set to “blocked” if not required
076.115 R1 Forward Intercept of right-hand resistive reach line
076.121 R1 Reverse Intercept of left-hand resistive reach line
012.013 Alpha_L1 α is the characteristic angle of the catenary

Page 11
 076.247 Gamma 1 γ is the start angle of the directional line
072.208 Beta 1 β is the end angle of the directional line
012.028 tZ1 Zone trip time delay (usually zero for Zone 1)

Figure 9 shows that it is possible to make the directional line into a “V” shape by
means of angles γ and β. However, this is not necessary in most applications, and the
directional line can be maintained as shown in blue. This is achieved by setting:

• Gamma = - 45o

• Beta = 135o

The P438 also allows the use of rate of change of current and voltage level detectors
(di/dt and dv/dt) to control the time delays associated with time-delayed Zones 2 and
3, allowing better discrimination between load and fault impedances. In most
applications this is not required (set dx/dt = “Disabled”), and thus is not covered in
these notes.

8. Switch on to Fault and VT Supervision (Fuse Failure)

The P438 includes two important functions closely associated with distance protection:

• Switch on to Fault (SOTF)

• VT Supervision

SOTF is used to make sure that any time a circuit breaker is reclosed, particularly for a
manual close command, that any fault present will be cleared instantaneously. SOTF
is used (1) to make the operation of distance elements non-directional, and hence not
dependent on the readiness of a memory voltage (this would be the risk any time the
VT is dead prior to the CB closure), and (2) to allow faults at the remote line end to be
cleared without waiting for their usual time delays (for example to allow a fast trip if a
catenary earthing clamp were inadvertently left in place at a point beyond the Zone 1
reach limit). SOTF is primed by assigning an opto input to the function 036.047
“Manual close EXT”. This must be energised by the same close pulse as is used to
initiate the closing of the CB, not by using breaker auxiliary contacts.

• Recommended settings:
Location in menu: Settings/Function Settings/Setting Group SGx/SOTF:

Cell (for Description Purpose

SG1)
011.061 Operating Suggest set “Trip with Starting” to temporarily
Mode bypass all CB Zone delays on closure.
011.060 Man. Close Length of SOTF time window – leave at 1s
Timer default.

Page 12
VT Supervision

Distance protection requires both a current and voltage input, in order to measure an
impedance. Essentially, a distance relay is an under impedance relay, and should the
VT input to the relay be lost for whatever reason, then the division of voltage (now
zero) by any flowing load current, will give a result of zero, and risk a spurious trip.
VT Supervision monitors the health of the VT circuit, and is used to detect a VT fuse
blow, or inadvertent removal. In both cases this happens quickly enough so that the
relay will automatically block the distance protection, preventing it from maloperating.

To enable the fuse failure detection, set cell: 014.204 Settings/Function

Settings/Setting Group SG1: VTS/Fuse Fail. enab = “Yes”. (Example for setting
group 1).

It is better to use this function rather than the V< V-monitoring function, as the
undervoltage function may pick up and alarm under genuine operational conditions,
when the VT is dead just due to isolations/switching.

9. Autotransformer Feeding

Particularly for high speed routes with trains running at speeds in excess of 200 km/h
(125 mph), autotransformer feeding is typical. Each train would be rated at around
8MW, and the heavy load currents drawn would cause significant voltage drops across
the catenary feeding impedance. To avoid a decrease in train performance, feeder
stations and parallelling substations for classical systems would then need to be sited
prohibitively short distances apart.

Autotransformer feeding uses a centre-tapped high voltage system, with the trains
typically supplied at half this voltage via a catenary to rail connection at the
pantograph. Each catenary is accompanied by a feeder wire, whose supply voltage is
in antiphase to the catenary. The feeder wire is aerially mounted, supported on
insulators along the back of the overhead line masts. Figure 10 shows
autotransformer feeding for the typical 25-0-25kV system found in western Europe:
Supply
transformer

If Feeder

50kV Ic

AT AT Catenary

25kV

Rails

Figure 10. 25-0-25kV autotransformer feeding

Page 13
The use of autotransformers (AT) offers distribution losses lower than for simple 25kV
feeding, and allows running of high power 25kV traction units. Note that the
switchgear is shown to act double pole, isolating both the feeder and catenary circuits
in the event of a line fault. This is not always the case – some systems switch single
pole, with one distance protection relay per catenary and another per feeder.
Protection would then protect the two “halves” of the system independently, with Zone
1 and Zone 2 protection set to typically 85% and 120% of the section, similar to
protection of a classical system.

Figure 10 illustrates the distribution of load current for a train situated midway
between AT locations. Each arrow shows a unit of current (equal to ¼ of the train
load). Assume that two units of current are drawn from the 50kV supply, with the flow
to the train via the catenary. This current must return to the supply transformer, and
does so via the upper half of the AT windings shown, with the load shared between the
neighbouring ATs. Each unit of current in the AT winding sets up an equal and
opposite current in the lower winding, due to the 1 : 1 turns ratio. The only path for
these currents is to circulate via a catenary-train-rail loop, doubling the train current
compared with the current drawn directly from the supply transformer.

The topology of the AT system is often similar to the classical system shown in Figure
2, except that the supply transformer secondary winding is wound as a centre-tapped
AT winding, and ATs are connected catenary-rail-feeder at each downstream
substation.

Page 14
10. Autotransformer System Protection Philosophy

It can be noted from Figure 10 that wherever in the system the addition of the catenary
current and the inverted feeder current is made (Ic – If), this sums to be equal to the
load downstream. The same is true for fault current, and performing this current
addition by either a parallel connection of CT secondaries, or an addition in the relay,
forms the basis of most protection methods applied.

For discrimination between load and faults, distance protection is common, with (Ic –
If) the current input for relaying. The voltage input taken is generally the catenary to
rail voltage, as the low reactance of the ATs – typically 1% on a 10MVA base – ensures
that any fault voltage drop on the feeder will be exactly mirrored on the catenary too.

When using distance zones to provide protection for AT systems tripping double pole,
it should be realised that it is not usually possible to provide fully-discriminative
protection. This is because when a catenary and feeder current are combined, the
usual linear relationship between distance to fault, and impedance measured, is lost.
The result is that it then becomes difficult to set an underreaching Zone 1, and
overreaching Zone 2 element in the normal way. The approach taken is then to detect
all faults on any catenary or track, right from the Feeder Station to the Mid-Point
Substation, by distance relays installed at the Feeder Station. In the event of a fault,
the supply to all tracks is isolated by simultaneous tripping of all track feeder circuit
breakers at the FS. Inevitably this removes supply to more sections of track than
necessary, so autoreclosure is a must, to restore supply to all but permanently faulted
sections of catenary and feeder. The momentum of moving trains ensures that little
speed is lost during the dead time of autoreclose, and with high speed lines generally
being better fenced, and having less overbridges or limited electrical clearances
compared to classical systems, the infrequent losses of supply cause few operational
problems. As all line faults are isolated by tripping of circuit breakers at the FS, there
is then not the need to have switchgear at downstream substations rated to interrupt
fault current. For economy, load breaking switches are used instead of breakers at
SS1 and SS2 in Figure 2.

Page 15
11. Distance Protection Zone Reaches

Figure 11 shows how for any one track at the FS, the variation of the P438 impedance
calculated as:

Z = ( V catenary – V feeder) / ( I catenary – I feeder), varies with distance to fault.

Z
C
Z max
FAULT IMPEDANCE (ohms)

B
Z min
MPSS

A SS2
Solid line
shows Z
measured
SS1

FS
DISTANCE TO FAULT (km)

Figure 11. Variation of impedance measurement with fault location along track

For clarity, only the impedances measured for a catenary to earth fault located
upstream of SS2 are plotted. The hump-like characteristic has a number of noticeable
trends:

• The initial slope of the characteristic, in ohms/km shown as line “A”, is equal to the
characteristic impedance of a catenary to rail out and back fault loop (the 25kV
loop in Figure 10).

• At AT locations, slope “B” shows how the effective ohms/km trend is equal to half
the characteristic impedance of a catenary to feeder out and back loop (the 50kV
loop in Figure 10), which is lower due to the feeder being a better conductor than
the rails.

• Beyond SS1, the parallelling effect means that slope C is greater than slope A. In
Figure 11, the system simulated is four track, thus the gradient of C will be
approximately four times that of A (marginally higher than four for the inner
tracks, and less than four for outer tracks).

Distance relay reaches would typically be calculated as follows:

Page 16
! Zone 1

The zone 1 elements of a distance relay should not overreach and trip for faults
beyond the MPSS, when the mid-point bus section breaker is closed. If it is known that
the MPSS is definitely open, then there is no real reach constraint for distance
protection. However, if the mid-point breaker is closed, or no status information is
communicated to the protection, then the relay must not trip for the lowest impedance
fault out of zone. It can be appreciated that in Figure 11 this fault impedance would
be equivalent to a fault like Z min, but located at the MPSS busbar, following slope B.
The applied Zone 1 setting would be restricted to 85% of this impedance, to allow for
tolerance errors.

Depending on the response of relay characteristics to inrush current, a lower reach

setting may be applied. This avoids the relay from mistakenly interpreting AT inrush
current at switch-on as an in-zone fault. For relays that have magnetising inrush
restraint, such current will not cause a problem, and this reach constraint is
unnecessary.

! Zone 2

Allowing for underreaching errors, the zone 2 reach (Z2) should be set in excess of
115% of the protected line impedance for all fault conditions. The relevant impedance
in Figure 11 would be the Z max peak between SS2 and MPSS. The Z max figure for
the highest “hump” is used, which must then be increased by 15% or more. A higher
figure of 120% x Z max would also offset regenerative braking effects. Thus with Z1
and Z2 applied, the relays at the Feeder Station provide complete protection up to the
MPSS.

! Zone 3

Zone 3 may be applied to protect for faults beyond the MPSS, or with a longer reach
to cover instances where ATs are switched out of service and the effective feeding
impedance becomes higher.

12. Distance Zone Time Delay Settings and Load Avoidance

The principles used are identical to those for classical feeding, with just one exception.
Zone 1 may sometimes be subjected to a short time delay – of the order of 50ms – if
magnetising inrush current is expected to be severe.

It should be appreciated that due to the relay using (Ic – If), which is measuring the
combined load current of all trains at their pantographs, the load impedance to avoid
is that measured from catenary to rail (the “25kV” impedance in Figure 10).

13. Implications of Using Two-Pole Switching and Autoreclosure

A full discussion of operational implications is beyond the scope of these notes, thus
just a few important points are listed:

• To avoid repetitive tripping of unfaulted catenary sections, during the dead time of
the first shot of autoreclose it is usual to remove all parallelling between tracks.
This is achieved by opening motorised isolators at all SS and MPSS locations. On

Page 17
 reclosure the tracks are then radial fed, and a persistent fault would cause tripping
of the faulted track only.

• Once radial-fed, the relays at the FS need only trip their own track, not cross-trip
the parallel tracks as before.

• Protection at the FS will also trip for downstream AT faults. As there are no circuit
breakers at SS and MPSS autotransformer locations, AT protection waits for loss of
line voltage during the dead time of FS circuit breakers, and only then opens
switches locally. Thus on reclosure of the FS breakers the faulted AT will have
been disconnected.

• Multiple shots of autoreclose, with tracks radial fed are often used. For the last
shot, it is common to disconnect all ATs downstream of the FS prior to the
reclosure attempt. The reason is that with all ATs and paralleling removed the
relay can now apply a linear relationship distance to fault measurement – the
impedance measured in ohms is directly proportional to the distance to fault,
allowing conventional fault location algorithms to be applied. Fault rectification
crews could then be directed to the correct site to investigate a permanent fault.

• Zone reaches of distance elements may be increased for the final autoreclose
attempt due to the relay measuring the out and back catenary to rail fault
impedance right up to the MPSS. This may be achieved by switching to an
alternative setting group with Z2 set higher than previously.

14. Classical System Protection Worked Example

Figure 12 depicts a typical 25kV system, with the settings for the relay protecting track
feeder TB-1 at Trent Bridge substation to be calculated. The inputs to the relay are
derived from the track feeder CT adjacent to the circuit breaker, and a busbar voltage
transformer (although a catenary side VT individual to TB-1 section would be equally
suitable).

Other system data:

• Each catenary has primary impedance characteristics: 0.26 + j 0.68

Ω/km

• Booster transformers are sited every 3km, their impedance: 0.051 + j 0.21Ω
each

• Vacuum circuit breakers are used with max. trip time: 65ms

• Maximum assumed Zone 1 protection trip time: 45ms

• Load current measured never exceeds 150% of CT primary rating.

• Catenary rated current at 20oC ambient temperature 540A

• Limiting catenary temperature for full speed train running 48oC

• Catenary trip temperature to prevent mechanical damage 56 oC

Page 18
 Off-load Track Feeder
Voltage CT Ratio
26kV 600/1

Trent Bridge Edgbaston Lords

(TB) (EB) (LS)

1 3

TB-1

275/26kV
2 4
15 MVA
Ζt = 10% ∠88°

Busbar VT
Ratio
26400/110

12.2km 13.7km

Figure 12. Assumed Major Feeding Diagram, VT and CT Locations

The first step is to calculate the primary impedance for the catenary sections to be
protected. Zone 1 for relay TB-1 protects section 1, however the backup protection
offered by Zones 2 and 3 must discriminate with downstream relays and so the
impedance of sections 2, 3 and 4 needs to be calculated too. In this example each
pair of catenaries runs between the same two substations, and so the impedance of
adjacent sections will be identical. (This would not be the case for four track railways,
or feeding configurations where the two tracks follow different routes and could be of
dissimilar length.)

The equivalent section impedance per kilometre is as follows:

= Line impedance per km + (BT impedance / BT spacing)

= (0.26 + j 0.68) + ( (0.051 + j0.21) / 3 ) = 0.277 + j 0.75 Ω/km

= 0.8 ∠ 69.7o Ω/km

This will be rounded up to an assumed 70 o common characteristic line angle, α, set

on the relay.

Most protection relays are set and injection tested in secondary impedance terms, so
actual (primary) impedances on the system must be converted to secondary. The
equation to be used is:

Zs (secondary impedance) = Zp (primary impedance) x CT ratio / VT ratio

Zs = Zp x (600/1) / (26400/110)

= Zp x 2.5

Page 19
Section impedance calculations:

• Sections 1 and 2: Zp = 12.2 km x 0.8 Ω/km = 9.76Ω

Zs = 9.76 x 2.5 = 24.4Ω

• Sections 3 and 4: Zp = 13.7 km x 0.8 Ω/km = 10.96Ω

Zs = 10.96 x 2.5 = 27.4Ω

Zone 1 Reach Calculation for TB-1:

• Forward Reach: Z1fwd = 24.4 x 85% = Ω

20.75Ω

• Reverse Reach: Z1rev = Blocked (forward directional operation required)

Zone 2 Reach Calculation for TB-1:

Firstly to calculate the follow-on fed fault scenario. Figure 4 shows the condition to
consider, with two track feeding only in the Trent Bridge area.

Z2 = (Z + 0.7 E) x ((A + R)/R) / 1.15

Where: Z = Impedance of sections 1 and 2 in parallel

= 24.4 / 2 = 12.2Ω

A = The track of interest, section 1 = 24.4Ω

R = Parallel fault current path (section 2) = 24.4Ω

E = Shortest following section (3 or 4) = 27.4Ω

∴ Z2 = (12.2 + (0.7 x 27.4)) x 2 / 1.15 = 54.6Ω

Notice how for two track feeding, (A + R)/R above becomes 2, due to a fault current
split between two identical parallel paths.

Now for hairpin feeding. Referring to Figure 5, it is apparent that with only two tracks,
inner tracks B and C are not present (B = C = ∞), so once circuit breaker TB-2 at
Trent Bridge is open, the impedance to the fault is merely 170% times the impedance
of track section 1 or 2. Thus:

Z2 = (24.4 + (0.7 x 24.4)) / 1.15 = 36.1Ω

For Zone 2 always the lowest of the two calculated results is used:

Page 20
• Forward Reach: Z2fwd = Ω
36.1Ω

• Reverse Reach: Z2rev = Blocked (forward directional operation required)

Zone 3 Reach Calculation for TB-1:

Now for the follow-on fed fault scenario with two tracks. Figure 6 shows the condition
to consider: It is apparent that the calculation is exactly as for Zone 2 follow-on,
except that the multiplier of 0.7 (70%) is replaced by 1 (100%), and that as Zone 3
must tend to overreach rather than underreach we can take 120% of the fault
impedance calculated.

∴ Z3 = (12.2 + 27.4) x 2 x 1.20 = 95.1Ω

Repeating the same for hairpin feeding:

Z3 = (24.4 + 24.4) x 1.20 = 58.6Ω

For Zone 3 always the highest of the two calculated results is used:

• Forward Reach: Z3fwd = Ω

95.1Ω

• Reverse Reach: Z3rev = Z1fwd x 25% = Ω

5Ω

Zone Time Delays:

Zone 1 will be set to instantaneous operation (tZ1 = 0) – it is not common practice to

time-grade with primary protection on board trains.

Zone 2 (tZ2) should be delayed as follows:

t2 " CB max. trip time + Relay max. trip time + 50ms margin

= 65 + 45 + 50 = 160ms (or greater).

As all of the protection and circuit breakers are alike, then the 160ms figure can be set
as tZ2. If downstream relays were electromechanical (typically 40 to 70ms slower
than numerical), or the circuit breakers were oil insulated (OCBs typically 40 to 60ms
slower than VCBs), then the tZ2 delay would need to be extended accordingly.

The Zone 3 time delay can typically be set double the minimum calculated above.
However, as Zone 3 is often most at risk of unwanted pickup due to train starting
currents or momentary overloads, a longer setting of: tZ3 = 500ms will be
applied.

Additional settings:

• I> sets the current sensitivity of the distance protection, typically 10% In.

• Wrong phase coupling (WPC) is often not employed, and thus will not be
explained in detail.

Page 21
15. Overcurrent Protection

Railway systems often use overcurrent protection as time-delayed back-up to the main
distance protection elements. Two different philosophies of setting overcurrent
elements are typical:

• Definite-time overcurrent protection (DTOC) continually in service - in parallel to

the distance elements.

• Back-up overcurrent protection (BUOC) switched in service only during periods

when the distance protection is out of service. A typical example is where a VT
supervision or measuring circuit monitoring function detects a VT fuse blow. In
such instances the distance protection is automatically blocked, and the BUOC
elements can be automatically brought into service, such that catenary protection
is not lost.

16. Classical System Overcurrent Protection Worked Example

Overcurrent protection can be applied to the 25kV system in Figure 12. As

overcurrent protection is most commonly used non-directional for railways, the easiest
application is for track feeders at Feeder Stations, such as TB-1. At this location and
with normal feeding, any fault current will naturally be flowing away from the busbar,
and so no reverse operation can occur. At downstream substations it will not be
possible to apply overcurrent protection in a similar way, and any elements enabled
would tend to be set with long time delays to ensure that the distance protection zones
are all given chance to trip beforehand.

Back-Up Overcurrent (BUOC) at Feeder Stations:

Should the distance protection be out of service, two BUOC overcurrent elements
could be set. Firstly a high set overcurrent element, typically I>> is set to underreach
the protected section, mimicking Zone 1 operation. This can be set for instantaneous
tripping. Secondly, a lower set overcurrent element can be applied to complete
protection for TB-1 section, overreaching the end of the protected section at
Edgbaston. I> is set for this purpose, operating with a definite time delay.

In order to determine the overcurrent settings, the fault current measured by TB-1 CT
for a fault next to the Edgbaston busbar needs to be calculated.

Scenario 1:

• Fault current for a fault at the end of section 1, with two tracks in-service.

If (TB-1) = Source volts / (Zt + (Section 1 and 2 parallel impedance) ) / 2

Where: Zt = ( kV2 / MVA ) x % impedance

= ( 262 / 15 ) x 10% = 4.5Ω

Page 22
∴If (TB-1)= Source voltage / (Zt + (Section 1 and 2 parallel impedance) ) / 2

= 26000 / [( 4.5 ∠ 88o ) + ( ( 9.76 ∠ 70o ) / 2 )] / 2 = 1.4kA.

Note that the fault current splits into two parallel paths, fed via TB-1 and TB-2, and so
the final division by 2 in the equation results in the per track current being calculated.

Scenario 2:

• Current for a fault at the end of section 1, with section 2 isolated for maintenance.

If (TB-1) = Source voltage / (Zt + (Section 1 impedance) )

= 26000 / [( 4.5 ∠ 88o ) + ( 9.76 ∠ 70o )] = 1.84kA.

Overcurrent Setting for BUOC Instantaneous Stage:

To prevent overreach, set at least 20% above the higher of the two fault scenarios:

I>> " 1840 x 1.2 = 2.2kA

The secondary current setting on the relay is found by dividing by the CT ratio:

I>> = 2.2kA / 600 = 3.68A

Overcurrent Setting for BUOC Delayed Stage:

To ensure complete coverage for short circuits in the protected section, set at least 20%
less than the lower of the two fault scenarios:

I> ≤ 1400 x 0.8 = 1.1kA

I> = 1.1kA / 600 = 1.86A (secondary)

A setting no less than the Zone 2 distance time delay would tend to be used, possibly
tI> set to 250ms.

All overcurrent protection must have a pickup in excess of the maximum expected load
current. Assuming that the maximum overloading would never exceed 150% of CT
rating, the I> and I>> settings look acceptable.

Definite Time Overcurrent (DTOC):

It is not general practice to set instantaneous protection elements running in parallel to

the distance zones, thus often just one definite time delayed stage is used. This setting

Page 23
can be applied common to all locations, and must be in excess of the maximum load
and overload current expected.

I> ≥ 600A full load x 150% overloading allowance = 900A

I> = 900 / 600 = 1.5A (secondary)

The time delay applied must be longer than the t3 distance zone delay, so tI> =
800ms would be acceptable.

17. Thermal Protection

It is essential that railway catenaries remain in the correct position relative to the track,
thus ensuring good current collection by train pantographs. The catenary is designed
to operate continuously at a temperature corresponding to its full load rating, where
heat generated is balanced with heat dissipated by radiation etc. Overtemperature
conditions therefore occur when currents in excess of rating are allowed to flow for a
period of time. It can be shown that the temperatures during heating follow
exponential time constants and a similar exponential decrease of temperature occurs
during cooling. It is important that the catenary is not allowed to overheat, as this will
lead to contact wire supporting arms moving beyond acceptable limits, and loss of the
correct registration position above track.
The tension in the catenary is often maintained by balance weights suspended at each
end of tension lengths of contact wire. Overtemperature will cause the catenary to
stretch, with the balance weights eventually touching the ground. Further heating will
then be accompanied by a loss of tension, and excessive sagging of the contact wire.
The P438 relay incorporates a current based thermal replica, using load current to
model heating and cooling of the protected catenary. The element can be set with
both alarm (warning) and trip stages.
The heat generated within the catenary is the resistive loss (Ι2R x t). Thus, the thermal
time characteristic used in the relay is therefore based on current squared, integrated
over time. The heating leads to a temperature rise above ambient temperature, so in
order to calculate the actual catenary temperature, the relay must know the ambient
temperature. This can be either set as an assumed “default” ambient temperature, or
measured, typically using a PT100 platinum probe mounted externally to the
substation building.

18. Classical System Thermal Protection Worked Example

Thermal protection can be applied to the 25kV system in Figure 12. The thermal
ratings shown are for typical UK catenaries (a temperate climate) with characteristics:

• Catenary design temperature range for correct tension -18oC to 38oC

• Typical assumed max. winter temperature (610A rating) 10oC
• Typical assumed max. spring/autumn temp (540A rating) 20oC

Page 24
• Typical assumed summer temperature (515A rating) 23oC
• Worst-case assumed hottest ambient 28oC
• Balance weights touch ground 38oC
• 20% loss of tension, train speeds must be restricted 48oC
• Possible damage due to clashing of supports at overlaps 56oC

In order for the protection to operate to mimic the heating of catenaries most
accurately, an absolute temperature replica is used (ie. Where the heating model is
used to estimate the real or absolute temperature reached by the catenary). The
Common THERM settings must thus be set to:
• 022.065 THERM “Absolute Replica” = Yes
• 022.064 THERM “Relative Replica” = No

The P438 requires a thermal rated current or reference current, Iref, to be set that
corresponds to full load current. This rated current is qualified by stating the ambient
temperature at which this applies. For the UK example, Iref should be set to:

Iref = 540A

The secondary current setting on the relay is found by dividing by the CT ratio:

Iref = 540 / 600 = 0.9A

The (default) ambient “Coolant temp.” at which 540A is the rated current is set at
20oC. The catenary temperature at which mechanical damage may begin to occur is
56oC. This must correspond to the P438 thermal trip command, and so the “Max.
object temp.” setting is set at 56oC.

If ambient temperature compensation is being used, the probe should be ideally

mounted in a location to most accurately model the coolant air around the catenary:
• If the catenary is exposed to direct sunlight for the majority of the protected
section, then the probe should be mounted to face the sun.
• If the catenary is shaded from sunlight, such as running in a tunnel, then the probe
should be mounted on an exterior wall facing away from the sun.
• If the catenary runs in a cutting, shielded from wind, the probe should be mounted
in the lee of the substation.
• If the catenary is exposed to the wind, the probe should also be mounted on an
exposed wall.

To avoid chattering of contacts when the load current is close to the trip threshold, a
hysteresis setting is provided on reset. Typically the hysteresis is set to 2%, such that
following a trip the thermal model must cool by 2% before the trip contacts will reset.

Page 25
An alarm should be issued to warn the rail operator when speed restrictions are
necessary to avoid the risk of dewirements. This should occur at 48oC in the UK
example. The appropriate setting is:
θ alarm = 48oC

When ambient temperature compensation is not being used, an assumed default

coolant temperature ambient must be chosen. In the example, if the ambient were
assumed to be 20oC, that would provide adequate protection, except for a calculated
risk on certain hot summer days. Recall that Iref is the rated thermal current at this
ambient.

Having set the current and temperature settings, it is then necessary to define how fast
or slow the catenary temperature responds to changes in load current. This is
achieved by setting heating and cooling time constants. For most catenaries, the
heating and cooling time constants would be expected to be equal. However, in the
UK catenary example, the time constant often varies from 5 minutes during daytime,
to 5 – 7 minutes at night. In order to set conservative settings that assume the worst
case time constants for heating, and ensure protection at all times during the day, the
settings recommended are:

• Heating time constant, “Time const. 1” = 5 minutes

• Cooling time constant, “Time const. 2” = 7 minutes

The P438 also allows the thermal rating of the protection to be modified, based on
signals from opto inputs. However, the functions related to Parallel lines (“Parall.”) will
not be used – set cell 022.062, location in menu: Settings/Function Settings/Common
Settings/THERM: = “1 par. Line”.

Page 26
19. Classical System VT Supervision Worked Example
Again, assuming a UK system operation, where:

• Nominal syatem voltage 25kV

• Off-load voltage 25 to 27.5kV
• Usual minimum running voltage 22 to 24kV
• Allowable prolonged running voltage during infeed outages 16.5kV
• Short duration minimum (2 mins.) for double outage 12.5kV

The nominal VT ratio is 26400/110 V.

The nominal CT ratio is 600/1 A.

Settings required:

Cell (for Description Purpose

SG1)
014.204 VTS/Fuse F. Set = “Yes” to enable the supervision.
enab
014.208 V< VTS/Fuse The relay voltage must fall below this level in
Fail order for VTS to operate. Suggest leave at 0.10
x Vn default.
014.212 V> VTS/Fuse The resetting voltage, any recovery above this
Fail voltage will indicate a VT healthy condition. Set
marginally below 12.5kV, ie. = 0.45 x Vn.
014.216 Delta V The change of voltage, at the time of the VT
failure is used for VTS pickup. Typical setting is
the difference between V< and V> VTS (=
0.35 x Vn).
014.221 Delta I Any change in current above the delta I setting
at the same time as a drop in voltage will
prevent VTS operation, as such a current change
is indicative that a catenary fault is present.
Typically set equal to the DIST I> sensitivity in
cell 072.195 (usually 0.10 x In).

Page 27