Computers in Railways XI 425

DC protection calculations –
an acceptable approach
R. Leach, D. Tregay & M. Berova
Parsons Brinckerhoff Ltd, UK

Abstract
As a consequence of substantial changes to the railway track layout and
implementation of a new approach to the renewal of negative bonding as part of
a Major UK Railway Signalling Infrastructure Project, the need arose for
reassessment of the existing electrical circuit impedances. This was necessary to
ensure that circuit breakers and relays, protecting the portion of the third rail DC
traction power supply network in question, had safe protection settings employed
to continue to ensure safety under electrical fault conditions.
This paper outlines the challenges that needed to be overcome in gathering,
processing and reconciling input data to arrive at a complete, coherent and
consistent set of data necessary for the calculation of the maximum fault
impedances seen from each of the circuit breakers. The paper goes on to present
the development, implementation and application of the methodology for the
calculation of the protection settings.
The process, illustrated with a flow-chart, was developed based on the
relevant railway standards and guidance documents reflecting best practice,
taking into account previous experience and the lessons learnt from other recent
Power Upgrade projects. The methodology is based on modelling the feeding
arrangement with an equivalent electrical circuit and was implemented into a
spreadsheet based calculation tool. The tool facilitates the choice of feeding
scenarios, the input and validation of data and enables sensitivity studies of the
effect of various circuit components. A particular challenge of the application of
this methodology and tool was the identification of the worst fault scenarios,
especially in large switches and crossings areas with multiple ‘Tee’ feed
connections at various points along the track. Such problems were solved by
modelling and carrying out simulations for various scenarios to determine the
worst case. The methodology and computational tool were validated for every
application by comparing the results with those obtained by an independent
computer programme.
The analysis of the protection setting calculation results demonstrated
surprising conclusions contrary to original expectations.
Keywords: power supply, DC railway, negative bonding, protection settings,
modelling, computational techniques.

WIT Transactions on The Built Environment, Vol 103, © 2008 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
doi:10.2495/CR080421
426 Computers in Railways XI

1 Introduction
As a result of a major re-signalling project in the UK the permanent way layout
and the third rail DC traction power supply infrastructure underwent substantial
change, both in terms of the switch and crossing (S&C) layouts and in the plain
line sections of the rail network. The installation of new types of running rails
and the renewal of the traction return (negative bonding cables) based on a new
technique [1] meant that existing circuit impedances needed to be revisited to
ensure that circuit breakers and relays protecting the network in question had
safe protection settings employed to continue to ensure safety under electrical
fault conditions. This necessitated carrying out protection setting calculations
for 36 electrical sections, some standard plain line and others, involving
diverging routes, resulting in complex feeding arrangements.

2 Issues relating to data

The initial approach was to update the existing calculations, relevant to the
affected electrical sections, to reflect any changes in the infrastructure.
However, when compiling the data, it was found that before proceeding with any
analysis there were a number of issues to resolve, namely:
• Record drawings often up to 20 years out of date or incomplete
• Recent changes arising from an earlier Traction Power Supply Upgrade
project needed to be included
• Feeder cable lengths not typically included in earlier calculations where
impedance values of some such feeders were actually quite significant
• Variety of protection relay and circuit breaker current tripping device
types needed consideration
• Recent drawings based on metric units whereas original calculations
utilised imperial units (miles and feet) thereby warranting conversion
• Mileages shown on positive (conductor rail) drawings and negative
bonding drawings not always in agreement
• Details of running rail types not readily available, or unknown
• Different running rail types in parallel within electrical sections
• Multiple reference points e.g. record drawings based on miles and feet
from London, permanent way drawings at switch and crossing based on
local datum point, signalling drawings based on separate datum relating
to extent of re-signalling project
• Changes to conductor rail cross-sections, feeder cables and return
circuits arising from changes to permanent-way track layouts

3 Selection of approach and methodology

3.1 Assessment of previous practice
The existing protection setting calculations supplied by the client relating to
previous stages of the traction power system development were based on simple

WIT Transactions on The Built Environment, Vol 103, © 2008 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
 Computers in Railways XI 427

hand calculations, which typically excluded cable lengths and tended to assume
connection points were adjacent to the Traction Substations (TSS’s) and Track
Paralleling Huts (TPH’s). This presented several issues to overcome:
• Hand calculations cannot readily cope with complex layouts, or
multiple changes in conductor rail and running rail types.
• With hand calculations it is easy to make mistakes and difficult to check
the calculations.
• Often the lengths of junctions are now typically much longer. This
meant that physical connection points to conductor rails and
feeder/return cables were rarely adjacent, but often a considerable
distance from the TSS or TPH resulting in greater cable lengths.
Therefore, it became apparent that cable impedances could not be
ignored.
• Significant ‘Overhangs’ or ‘Tee’ feeds arising from extended junctions
were omitted from the original calculations.
• Historically, conductor rails were of smaller cross section, hence
impedance was very similar to the feeder cables and actual connection
point was not considered particularly critical.
Figures 1–3 below show the variety of configurations as typically found on
most schemes of this kind.

S/S 1 S/S 2

Figure 1: Normal feeding (no ‘Tee’ feed or ‘Overhang’).

S/S 1 S/S 2
tee

Figure 2: Branch in section forms a ‘Tee’ feed.

S/S 1 overhang
S/S 2

Figure 3: ‘Overhang’ with connections remote from substation.

WIT Transactions on The Built Environment, Vol 103, © 2008 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
428 Computers in Railways XI

Note: A ‘Tee’ feed or ‘Overhang’ (which is equivalent to a ‘Tee’ feed in

electrical circuit terms) is particularly difficult to evaluate in terms of providing
suitable protection settings. This arises as a fault in the ‘Tee’ or ‘overhang’
results in fault current from the two feeding circuit breakers flowing along the
same path, thereby increasing the apparent impedance seen by the circuit
breakers. It is therefore necessary to determine whether a fault at the end of the
‘Tee’, or a fault at the remote end of the section represents the higher impedance,
in order to determine the required settings (see Figs. 2 and 3). The risk is that
neither circuit breaker may trip under a fault condition.

3.2 Selected approach

As a consequence of the above, it was decided that hand calculations were not
appropriate for this project. Instead, the process was undertaken through use of a
calculation spreadsheet (PB-ProCalc) with data inputted via a customised front
end template, incorporating the relevant formulae contained within Sections 8.3
and 8.7 of the client’s process document [3]. The calculations were also
modified to take account of the client’s specific requirements within the
guidance document [4].
The main changes from the clients process document are clearly defined in
his guidance note [4]. The designer is instructed to exclude any allowance for
rail joints and bonding resistances. A 7.5% tolerance is to be added to the
settings to compensate for impedance relay tolerances. Any cables over 15m in
length are to be separately identified and included in the calculations.
In order to validate the accuracy of the spreadsheet, a full check of the
formulae within the spreadsheet was carried out using an independent Engineer
within Parsons Brinckerhoff (PB). In addition, the spreadsheet was tested
against one of the worked examples in the clients process document [3] and
against the example ‘Tee’ feed calculation results contained in the clients
Guidance document [4]. As a further means of verifying the accuracy of the
spreadsheet, a second independent set of calculations were undertaken using
Network Rail ‘Tee’ feed computer program [5] for all of the electrical sections
on the project.
The Proposed Approach was documented [6] and submitted to the client, and
formal acceptance received prior to implementation.

3.3 Modelling

An equivalent circuit was created for each electrical section based on

information derived from design drawings and record drawings (see section 4
and the Flow Chart in Figure 6 for further details).
In order to simplify the calculation process, and to arrive at worst case, the
following rules were applied:
• Benefits of reinforcement cabling for running rails around S&C
locations and at single rail track circuited areas were excluded from
calculations as worst case, on assumption these could be disconnected
during maintenance, etc. and therefore not available for traction return.

WIT Transactions on The Built Environment, Vol 103, © 2008 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
 Computers in Railways XI 429

• Running rails were assumed to be 109lb/yd (smallest cross section)

throughout as per original, unless confirmed otherwise.
• Conductor rail also assumed to have the smallest cross section
(100lb/yd), unless confirmed otherwise.
• In order to simplify calculations in multi-track areas, where different
types of running rail are in parallel, a particular rail size was selected
and the length adjusted in the spreadsheet to provide an equivalent
impedance.
• Cable lengths to substations and TP Huts were typically scaled from the
conductor rail and negative layout drawings as exact cable lengths were
not typically available.
• ‘Tee’ points and ‘Overhangs’ were worked out separately for positive
and negative circuits, making the calculations simpler to handle. This
produced exactly the same result as if the elements were combined.

3.4 Methodology

The preparation of equivalent circuits for the positive side (conductor rails)
became relatively straightforward, although determination of actual connection
points was in some cases difficult due to imprecise record data for some areas.
The situation with regard to the negative return circuits was, however, often
quite complex.
In some instances it was far from clear which equivalent circuit should be
used for the return circuit, in particular relating to multi-track areas with
substation return connections at differing points for each track (sometimes
hundreds of metres apart) and potentially involving an ‘overhang’ situation. The
dilemma was how to determine the position/length of the ‘overhang’ in relation
to the equivalent circuit for the remainder of the main line in terms of whether a
single point for all the return connections should be assumed, or each track
worked out separately. If a single point of connection is assumed, the issue
arises of where it should be placed: at the electrical mid-point for instance; at the
closest connection point; or at the furthest point from the substation. All of these
options then impact on the length of the assumed ‘overhang’, so in the end it was
decided that all of these options should be tested and the worst case taken.
Figures 4 and 5 below illustrate two extremes at the end of an electrical
section, with a fault occurring just beyond the substation return connections.
The solid lines are assumed to be part of the fault path and included in the
calculations, whereas the dashed lines are excluded. The dashed vertical line
indicates the assumed return connection point to the tracks.

4 Detailed process

The process of carrying out the protection setting calculations is presented in the
Flow Chart illustration, Fig.6 below.

WIT Transactions on The Built Environment, Vol 103, © 2008 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
430 Computers in Railways XI

To next S/S

Fault

Substation

Figure 4: Return connections to rail taken as electrical mid-point ‘along

track’ position.

To next S/S

Fault

Substation

Figure 5: Return rail connections taken at the furthest negative return

connections in the electrical section.

4.1 Preparation of single track schematic diagram

A single track schematic diagram (Fig 7) of the positive and negative circuits
was prepared and marked up with all the information necessary for the definition
of the geometry, circuit elements (sections of rail(s), or cables) and their
dimensions and parameters, i.e. circuit references, boundaries, mileage of
important points, rail types, cable sizes, points of change of rail type and
identification of the circuit elements together with their respective lengths. The
physical circuit diagram was then used to derive the feeding arrangement
diagram (Fig 8), which was in turn converted into equivalent electrical circuit
diagram (Fig 9). The first diagram indicated the absence or presence of ‘Tee’
feed in the positive and/or negative circuits, left-hand or right-hand ‘Tee’ feed,
whilst the second identified the equivalent circuit components, used as input data
to the calculation spreadsheet summarised in Table 1. For the purpose of

WIT Transactions on The Built Environment, Vol 103, © 2008 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
 Computers in Railways XI 431

Provide existing records drawings, including:

•Comprehensive Track Diagrams
•Arrangement of conductor rail, hook switches and jumper cables
•Arrangement of track circuits & negative bonding
•Others, e.g. Signalling plans

Provide survey drawings showing the existing arrangement of the

permanent way, conductor rail and negative bonding

Provide drawings showing the As Built arrangement of the

Permanent Way, Conductor rail and Negative Bonding

Provide existing records for protection settings calculations, including

type of protective devices and current settings

Provide relevant standards, guidance notes and other instructions

Obtain all necessary input data

Produce a schematic diagram for each electrical section based on

the Comprehensive Track Diagram (CTD), including the conductor
rail and the negative bonding layout showing points, substations,
TPHuts, cable connections, types of conductor and running rails, etc.

Is the electrical Yes

section over switches or
crossings?

Analyse the electrical

No section layout to determine
worst case scenario

Establish the type of electrical section equivalent diagram

Using the existing and As Built drawings populate the schematic

diagrams with the coordinates (in Miles and Feet) of important points,
including mile posts, points of cable connections, points of change of
conductor / running rail type, point tips, etc.

A
B

Figure 6: Flow-chart of the protection setting calculation process for Tee

feed case (in plain line feeding case latter comparison stage
omitted).

WIT Transactions on The Built Environment, Vol 103, © 2008 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
432 Computers in Railways XI

A B

Using the information in the schematic diagram populate the Summary

table with the types and lengths of the equivalent circuit components.

Compare the data describing the equivalent components in the schematic

diagram with those in the existing protection settings calculation records

Yes No
Do the data agree?

Establish the reason for discrepancy

Is the discrepancy due to alignment

No remodelling, different contact rail or
negative bonding design or
repositioning of the equipment ?

Yes

Independent check of the schematic

Carry out a check diagram and input data
& correct error(s)

Yes No
Error(s) found?

Run NR Protection Setting Running PB DC Protection

calculation software (Tee-feed) Setting calculation spreadsheet

Protection settings – Results (B) Protection settings – Results (A)

Comparison of protection settings

results (A) and (B)

No Do Results (A) agree Yes

with Results (B) ?

Overall independent check

Figure 6: Continued.

WIT Transactions on The Built Environment, Vol 103, © 2008 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
 Computers in Railways XI 433

N
TA NT
O
BS VA
TI
SU HA
E3
56

on
37M 2023c
37M 2305c

nd
HAVANT

Lo
(j)

To
To Brighton
T
TT

37M 2419c (e) (d)

(b) (a)
(c)

(T connection)
37M 2023c

37M 1795c

37M 1652c
(f)

T ON
AN TI

)
AV STA
ar
sb
H B bu
SU (k)
e
(-v

nd T
T
37M 2305c
37M 2387c

37M 1689c

n
37

T o
M

T
HAVANT Lo 37 10
(n) z M 8
z
To

z 12 4 c
z 38
c T
T
TT
z z T
TT
z zz TT

To Brighton
T T
z TT
z TT
z T
TT
zz
37M 1038c
(o) (m) (l)
37M 1395c [2]
37M 1696c

37M 1446c
37M 2023c
37M 2221c

Figure 7: Extract from a track schematic diagram of the traction power

supply system.

FEEDING ARRANGEMENT
(+)-ve (+)-ve

L3 POS

L1 L2

(-)-ve
L3 NEG

L4 L5

Figure 8: Electrical section feeding arrangement.

WIT Transactions on The Built Environment, Vol 103, © 2008 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
434 Computers in Railways XI

EQUIVALENT CIRCUIT

R3POS R3NEG
R1 R2

VA VB
R4 R5

R1,2,&3POS = ΣLN*rCR+rPOSCABLE
GENERALLY:
R3NEG,4&5 = Σ(LN*rRR / nrails)+rNEGCABLE

Figure 9: Electrical section equivalent diagram.

Table 1: Extract from input data summary table.

Equivalent Total
Type of circuit elements Length
Circuit Ref. length
(cable/rail) (feet)
Component (feet)

(a) Cable type ‘z’ 500

600
(c) Cable type ‘z’ 100
L2
(b) Conductor rail type 150lb/yd 79
307
(d) Conductor rail type 150lb/yd 228
L3 POS (j) Conductor rail type 150lb/yd 282 282

L3 NEG (n) 2 x Running rail type CEN60E1 282 282

(k) 4 x Cable type ‘a’ 150 150
L5 (l) 4 x Running rail type BS113A 301 301
(m) 4 x Running rail type CEN60E1 328 327

traceability and clarity the schematic diagram contained a list of all the source
drawings and documents from which the data has been extracted, together with
any assumptions made.

4.2 Calculation of fault impedances

Following the input of data from Table 1 to the Calculation Spreadsheet (PB-
ProCalc), the sheet automatically determines if a ‘Tee’ feed exists or otherwise
and proceeds to compute the fault impedance seen by each circuit breaker
feeding each electrical section (A and B) of the two track railway (in the example
given). The result can be seen from Fig 10.

WIT Transactions on The Built Environment, Vol 103, © 2008 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
 Computers in Railways XI 435

SECTION LAYOUT (L1, L2, L3POS, L3NEG, L4, L5)

+VE SECTION LENGTH (A to 'T') - L1 (miles) 1.563
+VE SECTION LENGTH (B to 'T') - L2 (miles) 0.058
-VE SECTION LENGTH (A to 'T') - L4 (miles) 1.571
-VE SECTION LENGTH (B to 'T') - L5 (miles) 0.119
+VE 'TEE-FEED' LENGTH - L3POS (miles) 0.053
-VE 'TEE-FEED' LENGTH - L3NEG (miles) 0.053

TEE-FEED CALCULATION

TEE-FEED CIRCUIT R3 [=R3POS+R3NEG] (Ohms) 0.0026384

"EQUATION 23" VALUE 0.0076348
Tee-Feed Calcs Required (Yes/No)? NO
IB/IA at ZTmax 0.0380052
SUBSTATION LIGHT LOAD VOLTAGE VLL (V) 790
VA MINIMUM VALUE (V) 285
VB MINIMUM VALUE (V) 285
VA & VB MAXIMUM VALUE [VLL+6%] (V) 837.4
K' MIN 0.3403391
K' MAX 2.9382456
IB/IA at K' MIN 69.71468077 69.714681
IB/IA at K' MAX 2.423407145 2.4234071
ZA (Ohms) -----
ZB (Ohms) -----
ZTmax (Ohms) -----

RADIAL FEED CALCULATION (remote cb tripped)

MAX FAULT IMPEDANCE (A) VIA TEE (Ohms) 0.0804262
MAX FAULT IMPEDANCE (A) VIA MAIN (Ohms) 0.0862535
MAX FAULT IMPEDANCE (B) VIA TEE (Ohms) 0.0111041
MAX FAULT IMPEDANCE (B) VIA MAIN (Ohms) 0.0862535
MAXIMUM FAULT IMPEDANCE
CROSS-BONDING ALLOWANCE (Ohms) 0
IMPEDANCE CB A (Ohms) 0.0862535
IMPEDANCE CB B (Ohms) 0.0862535
ROUTE SETTING
MAX TRACTION LOAD CURRENT IROUTE (A) 6000
SUBSTN SOURCE IMPEDANCE MIN (RS Ohms) 0.012
SUBSTN LIGHT LOAD VOLTAGE MIN [VLL-10%] (V) 711
ZROUTE (Ohms) 0.1065

'MAXIMUM SAFE' (half-kA from falling voltage curve table)

CIRCUIT BREAKER A (kA) 7
CIRCUIT BREAKER B (kA) 7
RECOMMENDED SETTING
CIRCUIT BREAKER A (kA) 6.5
CIRCUIT BREAKER B (kA) 6.5

Figure 10: Extract from Protection Setting Calculations Spreadsheet (PB-

ProCalc) illustrating the calculation of fault impedances and
protection settings.

WIT Transactions on The Built Environment, Vol 103, © 2008 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
436 Computers in Railways XI

4.3 Derivation of protection settings

Impedance relays and Falling Voltage Overcurrent protection were the two types
of DC traction feeder protection that existed on the portion of infrastructure
involving this project.

4.3.1 Impedance relays

• For impedance type relays, 7.5% was added to the value derived in the
box headed ‘Maximum Fault Impedance’ to determine the actual figure
(‘Maximum Safe’) to be used for the settings.
• Recommended Settings were then derived manually by rounding up to
the next 2.5mΏ, but in any event to be no more than 20% higher than
those given for ‘Maximum Safe’ setting to ensure proper discrimination
with the circuit breakers feeding the next electrical section.
• In order to avoid unnecessarily changing existing settings, where the
existing setting is between ‘Maximum Safe’ and 20% above ‘Maximum
Safe’, then the recommendation has been to leave the setting unchanged
• Additionally a Direct Acting Overcurrent Electromagnetic Trip is set at
a higher ’common’ setting to cater for close up faults.

4.3.2 Falling Voltage Overcurrent protection

In the case of Falling Voltage Overcurrent protection which is principally an
electromagnetic device mounted within and forming part of the mechanism of a
DC circuit breaker, the settings were derived by converting the manufacturer’s
standard protection curves to an impedance table (Fig 11) showing settings (kA)
against impedance. The actual ‘Maximum Safe’ setting is determined by taking
the impedance value from the calculation spreadsheet i.e. ‘Maximum Fault
Impedance’ and then reading off the required setting in kA (rounded down to
next kA value on the chart if the impedance is mid way between settings).

4.3.3 Route setting

This is the setting required on a particular electrical section which will allow
train services to operate without resulting in ‘nuisance tripping’ and is normally
advised by the client. Any settings should normally be above this value.

4.3.4 Summarising of results

The settings required for all of the electrical sections between adjacent
substations (or substation and TP Hut) were then summarised on A4 sheets, in
order that all the information relevant to those sections was available from one
sheet. In all cases ‘Category 1’ safe settings were achieved (based on worst case
data and feeding arrangements, hence not requiring any operational restrictions).

5 Discussion and conclusions

It was originally thought that the enhancements and reinforcement to the
negative bonding arrangements as described in [1], together with changes to

WIT Transactions on The Built Environment, Vol 103, © 2008 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
 Computers in Railways XI 437

Setting
(A)

10000
9500
9000
8500
8000
7500
7000
6500
6000
5500
5000

46 52 58 65.5 73 81 89 98 107 118 129

Impedance (milliΩ)

Figure 11: Chart of Overcurrent Settings versus Impedance including half

(kA) settings (for circuit breaker type RJR530L).

conductor rail size, would logically lower the associated electrical impedances in
comparison to the existing infrastructure. Benefits may have been realised in the
form of allowing increases in (current based) protection settings, thus permitting
higher train currents in section. However, the overall impact was seen to be
minimal. This was due to a range of factors which included modifications to the
S&C layout and associated detailed electrical feeding arrangements, together
with the application of the methodology contained in the client’s process
document. This was further supplemented by the corresponding guidance note
[4], namely working within the parameters of 7.5% when undertaking the
calculations, thus allowing retention of many existing settings. If one was to stay
within parameters of 2.5%, as suggested by [3] then definite changes would have
been warranted.
Consequently, the resultant sensitivity of the calculations to the actual
negative bonding reinforcement was seen to be relatively low. Moreover, the
majority of settings calculated were noted as being close to existing and
therefore not necessitating any changes. Only in a few instances were
recommendations made to change settings (calculated impedances actually
increased). Although no major changes were perceived, this exercise was still
seen to be of great benefit in that it provides a solid basis for future applications

WIT Transactions on The Built Environment, Vol 103, © 2008 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
438 Computers in Railways XI

when ascertaining these types of protection settings. This approach was

considered pragmatic gave a scientific means of calibration between the
spreadsheet mathematics and software. Moreover, the method allows for further
refinement and development as changes take place in the future.

Acknowledgements
The authors would like to thank Network Rail for their permission to publish this
paper.
The support of Thomas Palfreyman, Head of Electrification, PB Ltd in
developing this calculation process and spreadsheet, and in reviewing this paper,
is also recognised.

References
[1] Development of an Improved Traction Return System, Eur Ing Raymond
Leach and Dennis Tregay, paper Railway Engineering – 2007,
9th International Conference, London, UK, 20/21 June 2007.
[2] NR/SP/ELP/21051 ‘Calculation of Protection Settings for DC Circuit
Breakers’, Issue 2 dated Dec.2005
[3] NR/GN/ELP/27006 ‘Calculation of Protection Settings for DC Track
Feeders, Issue 2 dated April 2006
[4] Network Rail Southern Region Power Supply Upgrade Project - Guidance
Note A437-00-DC-32031 ‘DC Protection Setting Calculation’, Issue C1.0
dated 11 May 2004
[5] Network Rail Computer Programme “RUN FILE DC3.EXE”
[6] Parsons Brinckerhoff (PB) Methodology for Undertaking Protection Setting
Calculations associated with re-modelling and negative bonding changes
(Issue 1 dated 07/09/06)

WIT Transactions on The Built Environment, Vol 103, © 2008 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)