The Capacity of Railways and their Control Systems

Daniel Woodland (London Underground Ltd.)

Felix Schmid (University of Birmingham)

ABBREVIATIONS
ARS - Automatic Route Setting
ATC - Automatic Train Control
ATO - Automatic Train Operation
ATP - Automatic Train Protection
CCTV - Closed Circuit Television
DB - Deutsche Bahn AG (German limited company)
IDEF0 - Integration Definition for Function Modelling
LUL - London Underground Limited
tph - Trains per Hour
UIC - Union Internationale des Chemins de fer (International Union of Railways)

1 THE RAILWAY SYSTEM

All modes of transportation have the primary purpose of moving passengers and goods. The rail
mode achieves this by moving individual vehicles and trains, carrying the passengers and goods,
from one location to another. However, the characteristics of rail transport differ from those of
other transport modes in a number of ways.
The most obvious characteristic of the rail mode is the inherent restriction of motions to a single
degree of freedom - the train is forced by the conicity of the wheel tread and the flanges of its
wheels to follow the lateral guidance of the rails. The use of a fixed track infrastructure limits the
set of available routes and greatly reduces the number of route options compared to the near
infinite number of options for road, sea or air transport.
This restriction of route choice is compounded by the need for a safe separation to be maintained
between trains on the same track segment. The low coefficient of friction between steel wheels
and steel rails, combined with high operating speeds and large moving masses results in braking
distances which are too long for safe operation purely on the basis of a train driver’s reaction to
objects within his/her line of sight. As a result of these factors, there is a far greater requirement
for access planning, or scheduling, to ensure that optimum use is made of the available paths 1 on
a particular infrastructure. The term “path” will be discussed further in section 3.3.
A more comprehensive overview of the characteristics of railways is given in Table 1.

1
A path on a railway can be described as the authority to be in a specific location on a train route (geographically
defined link between an origin and a destination) at a particular time with a given tolerance for the schedule time.

Page 1 of 22
 Motion Restricted to One Degree Low Coefficient of Distribution of Load
of Freedom Friction Between over a Large Area
Wheel and Rail

Direct  No steering required  Energy efficiency  Heavy axle loads

Strengths  Predictable motion  Smooth operation  High tonnage /
 Narrow swept path – reduced land  Efficiency of period
use requirement propulsion  Low ground forces
 High standard of safety  High speed operation  High load capacity
Indirect
Strengths  Use of linked consists (trains)  Energy recovery  Long life
 High capacity  Lower environmental infrastructure
 Wayside power supply is possible pollution
 Favours use of automatic control
 High initial system costs for  Distance to stop  Stiff rolling interface
Direct
Weaknesses guidance and control infrastructure  Distance to accelerate  Low inherent
 High infrastructure maintenance damping
costs  Cost of structures
 Low network flexibility  Cost of track
 Can not swerve to avoid a collision
 Overtaking and passing movements
require physical infrastructure
 Limited number of paths available
 No door to door service  No line of sight  Noise generation
Indirect
Weaknesses  Poor access to shipper’s premises working  Cost of inspection
 Reliant on performance of  Seasonal variations in
connecting modes train performance

Requirements  Junctions and stations  Train control  Track design

 Interlocking of variable geometry systems to ensure tailored to loads
infrastructure components safe operation
 Schedule movements in advance  Adhesion control
 Train control systems to ensure  Stringent safety rules
efficient operation

Table 1: Characteristics of Rail Transport (based on Schmid 2002, p4)

1.1 THE NEED FOR TRAIN CONTROL

Based on these characteristics, it is the authors’ opinion that railways require a control system to
achieve compliance with the operating schedule, both to:
 maximise the number of train paths that can safely be accommodated and;
 to assist in maintaining a service that reliably achieves the schedule’s demands.
If such a control system is to be effective in providing this support, it must be designed as an
integral part of the railway system and be based upon a clear understanding of the context within
which it must operate.

Page 2 of 22
1.2 THE CONTEXT OF A RAILWAY SYSTEM
The railway is a sub-system of the transport network of a country, which is in turn an integral
part of the system of a country's economy. In a similar way, the railway system is composed of
subsystems, including the control system (Goddard 1996, D5/24). The control system cannot
operate in isolation from the wider railway system and must be able to cope, therefore, with both
planned and unplanned demands placed on it by the day-to-day operation of the railway. Figure 1
shows a basic outline of this approach, portraying a number of factors, both internal and external
to the railway system, that have an influence (either direct or indirect) on the operation of the rail-
way and its control system.

Government & External Other Transport

Incidents Systems
Stake Holders
Freight Emergencies
Tourists Investment & Disruptions Suppliers
Interchange
Commuters
Regulation
Shoppers Fares Information
Goods / Passengers Charges
(Transport Demand)
Emergencies /
Staff Utilisation
Disruptions Energy,
Equipment
Failures & Delays
Weather & Services
Information
Environment Control System
Rules & Regulations
Staff Equipment
The Railway
Emergency Services System
Train
Optimisation

Transport Profits
Delays Security Comfort
Disruptions Information

Journey Return on
Travel Investments
Quality

Figure 1: The Context of a Railway System (based on Goddard, 1996, p D5/36)

Many engineers (particularly those responsible for signalling) have historically treated railway
control as purely concerned with ensuring the safe movement of trains. However, the introduction
of modern technology has led to the demand for a more rigorous approach to railway control,
encompassing all aspects of customer comfort and well-being. Hence, the railway control system
must now consider the safe control of operations, without neglecting the operational requirements
to keep passengers informed, regulate trains, save energy and provide for optimal usage of the
track.
In accordance with this approach, Figure 2 provides an outline of the main sub-systems that
combine to form the railway control system. It should be noted that, as already outlined in Figure
1, the operation of these sub-systems requires both equipment and human elements (including
drivers, signallers, controllers, maintainers and station staff).

Page 3 of 22
 Staff Utilisation Information
Failures
& Delays Rules & Regulations
Emergencies
& Disruptions Railway Control Station The Railway
System Control Systems System
Revenue
Collection
Train Operation
Systems CCTV,
Fire Detection,
Information Display
Telecoms, Suppression &
(Staff & Passenger),
Train Signalling
Radio,
Public Address, Voice
Ventilation
Data Control
Control Systems Transmission
Communications,
Train Asset Monitoring &
Optimisation Systems Management Lifts, Escalators,
Pumps, Lights,
Service
Control Crowd
Power, etc. Passenger
Systems Control Flow
Network
Planning
Optimisation
Line Control
Network Control Systems
Systems

Figure 2: The Railway Control System

2 THE SYSTEM RAILWAY

The word ‘system’ has several related, but distinct meanings. When talking about a ‘railway
system’, it can be used to refer to the organised body or set of things that we group together under
the title of a ‘railway’. That is, we consider the railway as a single entity, everything inside of
which is a part of the railway. This is the meaning that has been assumed already within this
paper when referring to the ‘railway system’. It is also the meaning implied by the Railtrack
‘Yellow Book’ definition of a system: ‘Any collection of equipment, people and procedures
which work together to achieve a common goal’ (Railtrack 2000, p1). In this context, the word
‘system’ could be replaced by ‘network’ and the same understanding inferred.
Alternatively, the term ‘railway system’ could be interpreted as a reference to the way that the
technology and processes implemented within a railway interact and combine to form a cohesive
whole, or a ‘system’, which happens to be a railway. Using this approach, we consider what the
system does, how it does it and how its sub-systems fit together. In order to ensure clarity, this
second usage will be referred to as the ‘system railway’ within this paper.

2.1 THE SYSTEM RAILWAY DELIVERY PROCESS

A system railway can be represented in terms of the processes required for its development and
operation. This can be done by use of an IDEF0 model (Integration Definition for Function
Modelling). Within IDEF0, a rectangular box represents processes or transformations. Inputs
enter this box from the left and outputs exit to the right. Controls / constraints on the process
enter from the top of the box and mechanisms providing the means of carrying out the process
enter from below.
As identified in the sub-system view, the railway system is subject to external factors. These
govern all activities and operations, whether in a steady state, a degraded mode or a transitional
state, as can be seen from the top level IDEF0 diagram in Figure 3.

Page 4 of 22
 Available
Finance Principles
Political
Direction Standards Rules

Transport
Product
Transport Move People & Goods
Demand Journey
(System Railway) Quality

Equipment Cargo
Staff Passengers
Power
Finance

Figure 3: Top Level View of the System Railway Process

2.2 DECOMPOSITION OF THE SYSTEM RAILWAY

If this top-level view (level 0) is decomposed to represent the processes that are required to create
(develop and operate) a Railway System, the first level IDEF0 representation shown in Figure 4
is obtained.
Political Available Principles
Rules Standards Rules
Direction Finance

Finance Allocation
Transport Plan System
Demand Requirements
Railway Theoretical
Capacity Line Capacity
Topography, Loading Gauge,
Demand OHLE Arrangements
Driving Methodology
Develop
Railway Equipment Types
and Quantities
System

Planned Availability

Service Pattern
Plan Railway
Operation Staff Levels

Journey
Quality
Operate Railway
Transport
System Product

Equipment Equipment Equipment Equipment Cargo

Staff Staff Staff Staff Passengers
Finance Power Finance Power
Finance Finance

Figure 4: First Level View of the System Railway

Page 5 of 22
It should be noted that overall transport demand acts as the initiating input for the entire
transportation system, not just the process or transformation executed by the system railway.
There will also be a limit to the available finance for transportation as a whole and a given
railway system in particular. Some proportion of the transportation demand and available finance
may be acquired by railways and must be translated into capacity demand and finance allocation
for specific railway networks.
The first process identified is, therefore, the planning of system requirements for the railway.
This process can be found in the top left-hand corner of the diagram in Figure 4. The mechanisms
of this process are not of paramount importance in this paper and will therefore not be expanded
further at this stage.
Once the railway capacity demand is known, the railway system itself can be developed in
accordance with the demand. The process by which this is done is shown in Figure 5.
Finance
Standards Allocation Principles Rules

Planned
Railway Develop Availability
Capacity Track
Demand Infrastructure Topography,
Loading
Gauge, OHLE
Platform Lengths Arrangements

Develop
Trains
Train
Characteristics

Develop
Stations Dwell

Equipment
Develop Types and
Control Quantities
System Theoretical
Line
Capacity
Develop
Operating Driving
Startegy Methodology

Equipment Equipment Equipment Equipment Staff,

Staff Staff Staff Staff Finance
Power Power Power Power
Finance Finance Finance Finance

Figure 5: Second Level View of “Develop Railway System”

A feedback loop has been shown between the development of stations and train in Figure 7. This
represents the fact that the length of trains operating on the line may be constrained by the
available platform lengths. In practice, the constraint could, of course, work the other way – with

Page 6 of 22
platform lengths being designed on the basis of planned train lengths. Other similar feedback
loops would also exist between all of the second level sub-processes (they have been omitted
from the figure purely for the sake of clarity).
Based upon the outcomes of the ‘Develop Railway System’ process, plans can then be drawn up
for the railway’s operation. These must account for the demands of the systems users, the
arrangement and types of both fixed and moveable infrastructure, and staffing issues. This
process can be seen within Figure 4, but as its mechanisms are once again not of paramount
importance to this paper, they will not be expanded further.
Once the service pattern and staffing levels have been determined, the actual operation of the
railway can be studied, as shown in Figure 6.
Eqipment Types
& Quantities Staff Levels
Topography,
Loading
Staff Finance Driving
Guage, OHLE
Levels Allocation Rules Methodology
Arrangements

Planned Actual
Availability Maintain Availability
System

Service
Movement
Pattern
Demand
Railway Control Service
Capacity
Demand

Operate Transport
Service Product

Equipment Equipment Equipment Cargo

Staff Staff Staff Passengers
Finance Finance Finance
Power Power Power

Figure 6: Second Level View of “Operate Railway System”

The key point to note from these IDEF0 representations of the railway system is that a number of
different processes must work together, utilising a number of engineering systems, in order to
supply efficiently a transport product that meets the customers’ demand for transportation in
general, and railway capacity in particular. Unfortunately, the system railway in Britain is not
really managed as an integrated whole in this way, but rather as a set of subsystems.

Page 7 of 22
3 RAILWAY CAPACITY
Whilst ‘capacity’ is frequently mentioned in the literature about railway systems, what is meant
by the term is rarely defined. This is perhaps because it can be taken to mean a number of
different things – each of which offers advantages when trying to understand a railway’s
operational performance.

3.1 THEORETICAL CAPACITY

Probably the most commonly used measure is that of ‘technical headway’ (often referred to as
simply ‘headway’):
Technical Headway: Minimum time or distance between trains that the signalling will
permit, so that the train ahead does not affect a following train.
The units of this measure are most often defined as either ‘seconds between trains’ or ‘trains per
hour’ (tph). Unfortunately, the definition given above allows two alternative interpretations. That
commonly adopted by main line railways in the UK could be described as ‘Train Following
Capacity’:
Train Following Capacity: Indicates the maximum throughput at a particular location on the
railway network, such as a signal position, if all trains were to
follow each other at line speed and with a minimum of braking
distance separation (with some element of safety add-on), no
allowance being made for station stops.
In contrast to this, the typical interpretation of technical headway for UK metro railways could be
described as ‘Point Capacity’:
Point Capacity: Indicates the maximum throughput at a particular location on the
railway network, such as a station platform, accounting for
station stops and actual train speeds (McCormick et. al., 1996 pp
A1/7).
In either case, the use of technical headway as a measure of capacity is only meaningful in the
context of a specific location on the railway. Just because one location can achieve a high
throughput in trains per hour does not mean that the whole line will be able to do so. The
technical headway of any line is therefore limited by that of the most restrictive location along
the line.
The measure also fails to account for the speed of the service. The train following headway that
can be achieved with an ‘n’ aspect signalling system is determined in part by the design speed
(V) and in part by the actual speed of operation (Vact). In simplistic terms (ignoring system
delays, sub optimal signal spacing, driver behaviour and other complications – such as station
stops), this means that where b = brake rate, S = sighting distance, O = overlap and L = train
length, the headway time for a section of railway signalled to UK main line practice is given by:
V 2 (n  1) S O L
H tn   for n>2 Equation 3.1
2bVact (n  2) Vact
Dividing 3600 by this headway time gives the headway throughput in ‘trains per hour’ (tph).
Equivalent equations can also be derived for point headways, where all trains stop at a station,
such that:

Page 8 of 22
  2 (O  L) 
  For V  2 a (O  L )
S V (2n  3)  a 
H tn ( Stopping )    Dwell  (for n > 2)
V b(2n  4) V O  L  For
   V 2 a (O  L )
 2 a V 
Equation 3.2
In accordance with this approach, examples of the relationship between speed and technical
headway are given in Figure 7.
The rising part of the technical headway curve is determined by the linear increase in line
capacity as trains run faster over a particular section while the drop-off beyond the point of
maximum capacity is given by the braking distance which increases as the square of the speed.
Figure 7 also illustrates that, regardless of the type of signalling used, speeds may need to be
reduced to increase the throughput in terms of trains per hour. However, reducing speed on a line
would require additional numbers of trains in order to provide the same service frequency and
would also significantly contribute to journey time. This latter issue is an important part of the
performance as far as passengers are concerned. ‘It is not enough just to consider point capacities
along the line. We need also to look at junctions, terminal working, services operating with
different stopping patterns, trains running at different speeds and the rolling stock fleet size’
(McKenna 1998, pA1/4). Technical headway does, however, have its advantages in helping to
identify ways in which the effects of pinch points on the network can be overcome.
Train Following Headways Point Headways
85

34
75
32
Technical Headway (tph)

Technical Headway (tph)

65 30

28
55

26
45
24

35 22
15 25 35 45 55 65 75 85 95 105 10 20 30 40 50 60 70 80 90 100
Line Speed (mph) Line Speed (mph)

3 Aspect 4 Aspect 5 Aspect Moving Block

Headway (tph) Headway (tph) Headway (tph) 80m Safety
Margin (tph)

Train Length 200m, Sighting Distance 275m, Overlap 180m, Braking Rate 0.5m/s/s, Acceleration Rate 1m/s/s, Dwell 30s

Figure 7: Theoretically Achievable Headway

It is possible to demonstrate, using the headway diagram shown in Figure 7, that terminal stations
do not represent a serious limit for the capacity of a line operated at a speed substantially higher

Page 9 of 22
than that which offers the greatest capacity, so long as the terminal is equipped with sufficient
platform tracks to cope with the turnaround time of all trains (conditions that are met by most
main line railways). For example, if a route is operated at 90mph with five-aspect signalling, then
22 trains can be operated per direction on a two-track railway. Additional tracks have to be
provided on the approach to a terminal only once the speed drops to below 15mph. Similar
considerations apply for a station where all trains stop.
In order to overcome the point specific nature of ‘technical headway’, the minimum line headway
can be considered:
Minimum Line Headway: The minimum time interval between a pair of trains, so that a
following train is not affected by the train ahead throughout
its run (Pachl 2000, p5/8).
Perhaps most usefully, this could be measured as seconds between trains, since it will vary for
different pairings of trains. An example of how this works can be seen in Figure 8. If the area of
line shown in the figure represented the entire line being considered, the minimum line headway
between non-stopping trains 1 and 2 would be 77 seconds. Between non-stopping train 2 and
stopping train 1 it would be 152 seconds, etc. Attempting to run two trains over the line at less
than their minimum line headway separation would impede the following train’s progress.
Time (s) Line Speed Achieved
by Stopping Trains
600
Non-Stopping Train 3
G Y R 77 seconds
Y R
500 G Y R
G

152 151 seconds

400 seconds Stopping Train 2
G Y R
Y R 457 seconds
G G Y R
300
151 152 seconds
seconds Stopping Train 1
G Y R
Y R
200 G Y
G Non-Stopping Train 2R
G Y R 77 seconds
77 Y R
seconds G Y R
100 G
G Y R
77 Y R Non-Stopping Train 1 Distance
seconds

500 1000 1500 2000 2500 3000

(m)
Clearance
of Signal
Signal 1 Signal 2 Signal 3 Signal 4 4 overlap Signal 5
Sighting Sighting Sighting Sighting Sighting
Point 1 Point 2 Point 3 Point 4 Point 5

Design Assumptions: Platform

Sighting distance = 275m Overlap length = 180m
Train length = 200m Line Speed = 60mph = 25.8 m/s/s
Braking rate = -0.54 m/s/s Acceleration rate = 0.56 m/s/s
Station Dwell time = 30 seconds
Distances and Times determined by use of Newton’s Laws Of Motion (V=U+at; V2=U2+2aS; S=Ut+½at2)

Figure 8: Variation in Line Headway with Stopping Pattern

Minimum line headway can account for the combined effects of the point capacities achievable
throughout the line, including the effects of all relevant junction layouts, multiple line sections,

Page 10 of 22
station stops and reversals. However, it only considers the headway of two defined trains and, as
it considers the effect of complete journeys, it does not remove the need to also consider technical
headway in order to identify where the system pinch points are and how to alleviate them.
An alternative definition that can be used to consider the passage of any number of trains is that
of ‘Theoretical Line Capacity’:

Theoretical Line Capacity: Indicates the theoretical maximum throughput of a railway

line when all trains complete more than one round trip.
Appropriate units for measuring theoretical line capacity would usually be quoted as train
kilometres per hour2. For a defined railway line, this provides a combined measure of both the
number of trains using the line and the speeds at which they travel, but does not allow these
factors to be differentiated. This means that services with vastly different headways may achieve
the same theoretical line capacity due to operation at different speeds.
For theoretical line capacity to be meaningful, it must represent a defined train service pattern.
This can be achieved by assuming all trains to have the same nominal performance and stopping
patterns (as may prove useful for a metro, but less so on a mixed traffic railway). Alternatively,
some other (more realistic) theoretical service pattern can be assumed. In either case, the
assumptions made and the use of the units ‘train kilometres per hour’ will make the measure line
or network specific. The service provided by a given number of train kilometres per hour could
be vastly different on lines or networks of different sizes and characteristics – or with the use of a
different set of performance and service pattern assumptions.
As theoretical line capacity represents the combined effects of the engineering infrastructure that
supports and enables the railway’s operation, it is included in Figure 4 and Figure 5 as an output
of the ‘Develop Railway System’ process.

3.2 SUSTAINABLE CAPACITY

Technical headway, minimum line headway and theoretical line capacity relate to the ‘theoretical
maximum’ that can be achieved from the railway’s infrastructure. However, the operation of any
railway can be expected to suffer deviations from the ideal operating conditions, due to factors
such as variability in system and human performance, system reliability and external influences
or perturbations. A margin for recovery is often added to the technical headway time to give a
‘service interval’ that allows for such deviations:

Service Interval: The sustainable time interval between trains, such that the train
ahead does not affect a following train (e.g., technical headway +
recovery margin).

The service interval (usually measured as seconds between trains) is also referred to as the
‘operational headway’ (measured as trains per hour). Just to confuse matters further, both railway
engineers and operators often shorten this to ‘headway’ (with the result that it is sometimes
difficult to know which definition of ‘headway’ is being referred to).
The size of the recovery margin must be selected to provide an optimal balance between the
proportion of time for which the margin will be sufficient to allow sustained service and the

2
Alternative measures of passenger or tonne kilometres per hour are considered in section 3.3.

Page 11 of 22
reduction in capacity that it introduces for the unperturbed service. In line with this approach, the
UIC have quoted a desirable norm for the actual operating capacity of a railway as 75% of the
theoretical capacity (Holtzer 1999, p587).
Being based on technical headway, the service interval is again point specific. This limitation can
be overcome by considering a derivative of the minimum line headway. This could be referred to
as a sustainable line headway:

Sustainable Line Headway: The sustainable time interval between a pair of trains, such
that the train ahead does not affect a following train, through-
out its run (e.g., minimum line headway + recovery margin).

On a metro railway (where all trains have similar stopping patterns and performance) the
sustainable line headway is fairly easy to work out. However, on main line railways, with a
multitude of train performance and stopping patterns to accommodate, the train separation
required will vary significantly throughout the day (as already shown in Figure 8). In order to
cope with this for planning and charging purposes, it is now common practice on UK main line
railways to talk about ‘train paths’.

3.3 TRAIN PATHS

Train paths have been defined in an abstract manner in footnote 1 on page 1. The following
definition is based on the terminology developed in the preceding sections.
Train Path: An allocated unit of capacity on a section of line, based on the
sustainable line headway between trains with a defined
nominal performance and stopping pattern (Nominal Line
Headway).
If the behaviour of the non-stopping train in Figure 8 is representative of the nominal train
performance and stopping pattern (with a minimum line headway of 77 seconds and sustainable
line headway of, say, 100 seconds for a 23 second recovery margin), the non-stopping trains
would require one train path, whilst operation of a stopping train would require two train paths.
As can be seen in Figure 9, some of the capacity of the line would be wasted by a rigid adherence
to the allotted train path, but use of this measure provides a convenient method for determining
access charges.
In practice, railways will never actually use all possible train paths. If the railway is to run
efficiently whilst satisfying the users’ demands, it must be expected that the service will vary
throughout the day, with extra trains at peak times and limited service, or possibly even closure
of the railway, due to a lack of demand during some other periods. Engineering access may also
be required to carry out maintenance and renewals work. Failure to account adequately for this
latter constraint would make the service unsustainable in the long term. Therefore, the service
must be operated to an optimal service pattern:

Service Pattern: The sustainable timing of trains in accordance with planned

operation. i.e., the timetable.
As the service pattern represents the planned capacity of the railway, it combines the effects of
the service intervals achievable throughout the line with an operational margin for recovery and
the potential to allow for other requirements and constraints on the railway’s usage. The service

Page 12 of 22
pattern is therefore included in Figure 4 and Figure 6 as an output of the ‘Plan Railway
Operation’ process, an input to the ‘Control Service’ process and a constraint to the ‘Maintain
System’ process.

Time (s) Non-Stopping Train 3

2 train paths
600 (200 seconds)
G Y R
200 seconds
500 (25 seconds over Sustainable Line Headway)
Stopping Train 2

2 train paths
600 seconds
400 (200 seconds)
(+ 143 seconds)
G Y R
200 seconds
300 (26 seconds over Sustainable Line Headway)
Stopping Train 1
1 train path
(100 seconds)
200 Non-Stopping Train 2
1 train path
100 (100 seconds)
seconds
100 Non-Stopping Train 1
100
seconds Distance
500 1000 1500 2000 2500 3000 (m)
Clearance
Signal 1 Signal 2 Signal 3 Signal 4 of Signal
4 overlap Signal 5
Sighting Sighting Sighting Sighting Sighting
Point 1 Point 2 Point 3 Point 4 Point 5

Platform

Figure 9: Schematic Arrangement of Train Paths (For design assumptions see Figure 8)

The capacity offered by a service pattern could be defined as the ‘sustainable line capacity’:

Sustainable Line Capacity: Indicates the sustainable throughput of a railway line when all
trains operate in accordance with the timetabled service
pattern (based on McCormick et. al., 1996 pp A1/7).
The units of this measure could again be train kilometres per hour. However, it should be noted
that neither the measure of trains per hour nor that of train kilometres per hour consider the
loading capacity of the trains themselves. Accounting for this, the units of theoretical line
capacity and sustainable line capacity could be converted from train kilometres per hour to
passenger/tonne kilometres per hour.
In practice, all that is required to convert train flows into passenger and goods flows is an
understanding of vehicle loading. This is a relatively simple matter on railways where all trains
are the same, or have broadly similar characteristics (such as a typical metro railway), as long as
the measure of capacity refers to a theoretical maximum (i.e., assumed full loading). However, if
the measure is taken to be actual loading (which must then be measured or estimated in some
way), the conversion becomes more difficult. This is particularly true if the railway operates
mixed stock and mixed traffic types with long and short trains able to carry differing loads of
passengers or goods. As a consequence, and in order to simplify the issue, most papers referring

Page 13 of 22
to line capacity do so in the context of train numbers only. However, if this kind of over
simplification can be avoided, the resulting measure of capacity will be far more useful.
Sustainable line capacity does not overcome the limitations already outlined for theoretical line
capacity. It simply provides a more realistic measure of what can be reliably achieved by a
railway.

3.4 CAPACITY HIERARCHY

It is important to note that all of the capacity definitions given above are of a theoretical nature,
as there will always be the potential for disturbances in the system that the capacity calculations
did not allow for. In reality, numerous engineering systems are required to work together with
each other, within the context of the railway environment, in order to deliver railway capacity.
Building on the IDEF0 system model and capacity definitions already discussed, the main factors
that contribute to the real capacity achieved by a railway system are shown in Figure 10.
ACHIEVED LINE CAPACITY

Sustainable Line Capacity Passenger or Tonne Kilometres per hour

Service Pattern (Timetabled Train Intervals)

Nominal Line Headway / Train Path

(Nominal Sustainable Train Interval) Seconds Between Trains

System Service Interval User Demand Pattern

Availability (Sustainable Train Interval) Seconds Between Trains Number of People / Goods;
When Required;
Maintenance /
From / To Where;
Upgrade Access Theoretical Line Capacity How Quickly.
Requirements; (Theoretically Achievable) Train Kilometres Per Hour
Reliability;
Repair Times
Minimum Line Headway Train
(Theoretically Achievable) Seconds Between Trains Capacity
Passengers /
Goods
Recovery Margin Point Capacity Track Layout
(Theoretically Achievable Number of Tracks; Layout of
Regularity in Stopping Train Interval) Points; Gradients; Loading
Trains Per Hour Train Fleet
Timetable; Gauge; OHLE Arrangements Train Characteristics
Driving Technique;
(capacity, acceleration,
Signal System Train Following Capacity braking & maximum
(Theoretically Achievable Non- speeds); Fleet Size
Stopping Train Interval)
Trains Per Hour

Service Control Serious Service Regulation Normal Service Platform Dwell Signalling Component Track
Abnormalities Perturbations Component Condition
Train Characteristics (length,
Reversals; Emergencies (fire, Dwell Time Extended Dwell Train Design; acceleration, braking & Adhesion
Major Conflicts; illness, etc.); Management; Time; Low Adhesion Regularity in maximum speeds); Characteristics;
Diversions; Equipment Failure; Coasting Poor Driving Timetable; Platform Line Speeds (track condition, Permissible
Allocation of Stock; Exceeding Normal Strategies; Technique; Management; Station and topography); Speeds
Allocation of Crews Service Perturbations Conflict Temporary Speed Design & Flows; Signalling Design (layout &
Resolution Restriction. Station Management system response times)

Figure 10: Achieved Line Capacity (Capacity Hierarchy)

Page 14 of 22
3.4.1 OPTIMISED CAPACITY
The greatest possible capacity for any line or network has been defined within this paper as the
theoretical line capacity. This could be considered, therefore, to be the ultimate engineering target
for optimisation of capacity. However, simply improving the theoretical line capacity does not
guarantee that the capacity has been optimised. The usage of available capacity must also be
considered. To this end, a measure of capacity efficiency could be defined in operational terms in
the form:
Achieved Line Capacity
Capacity Efficiency =
Theoretical Line Capacity
Improvements could then be made to this ratio by developing the availability of the railway’s
systems and fine-tuning its management and control processes in order to reduce the required
recovery margin, reduce service perturbations and allow better response to any perturbations that
do occur. Including operational measures in the service pattern, such as the ‘flighting’ 3 of trains
with similar stopping patterns and performance, would also improve this measure.
A similar approach could be taken with the number of train paths available as the target for
efficiency, such that:
Number of Train Paths Used
Exploitation Rate =
Total Number of Train Paths
Improvements could be made to this ratio by reducing serious service abnormalities and
including flighting of trains with similar stopping patterns and performance in the service pattern.
Unfortunately, these approaches consider only the output of the system (the capacity that can be
delivered) and not the input (the transport demand and / or railway capacity demand of Figure 3
to Figure 5. As a result of this, they fail to measure the success of the system in meeting the
demand that it exists to fulfil. Focusing too heavily on improving them could, therefore, lead to
inefficiency (e.g. running under-utilised trains).

3.4.2 CUSTOMER FOCUSED CAPACITY

Optimisation from the customer perspective may well be different from optimisation as viewed
by engineers or operators. User requirements for the service (the required frequency of trains,
from where, to where, and how long the journey should take) must be accounted for when
considering the capacity achievable by a railway system. The users’ requirements can then drive
both the development of the system to achieve higher capacities where needed, and also the
optimisation of the way the railway is operated, once the system has been designed. Whilst the
capacity definitions considered so far have allowed for this (through the inclusion of user
requirements in the development of the service pattern), they do not provide a means of
measuring the success with which user requirements have been met.
In order to provide a better measure of how well the system meets the users’ requirements, a
further definition of capacity can be given as the ‘optimum line capacity’:

3
‘Flighting’ is a term used to describe the practice of grouping trains with similar stopping patterns, performance
and/or direction together in the service pattern. The use of this practice allows more efficient usage of the available
capacity than, for example, scheduling a fast train to follow a slow one.

Page 15 of 22
 Optimum Line Capacity: Indicates the sustainable throughput when passenger and goods
travel times and comfort are optimised (McCormick et. al., 1996
pp A1/7-10).
‘Optimised’ here does not necessarily mean minimised or maximised, since the cost implications
of that could be far from optimal. Instead, it implies a cost to benefit consideration of the service
provided to the passenger. Suitable units for this would be social benefit value per passenger
kilometre, where the social benefit would represent the value of the passengers time / comfort
(including both journey and waiting times). This value could be optimised in accordance with
achieving:

Social Benefit Value + Revenues

Benefit to Cost Ratio = = Acceptable Ratio
Cost of providing Service
The social benefit offered by the system would be difficult to measure and, indeed, the authors
have not come across any published attempt to do so in a way relating specifically to customer
demand. One approach that has been used on LUL is an estimation of the value of an average
passenger’s time, together with effects of proposed improvements on journey time and quality
(LUL 1994, pp6, 9). This can even be taken so far as to consider the effects of delays in different
parts of the system, such as at platforms, on trains or whilst travelling to the platform (Millard
1999, pp142-3). Such analysis provides the opportunity to tune the service to customer
requirements, rather than purely technical ones.

4 RAILWAY CONTROL SYSTEM CAPACITY

If the capacity hierarchy shown in Figure 10 is considered, it can be seen that there are five
components through which a railway’s control system can influence achievable capacity:
1. The signalling
2. Control of platform dwell
3. Service control (response to serious service abnormalities)
4. Regulation (response to normal service perturbations)
5. System availability
The remaining factors (track layout / condition, train fleet characteristics and user demand
patterns) fall outside of the influence of real time railway control).
Of the five components listed above, the signalling, service control and regulation components
are controlled by means of the train control system, and the platform dwell component can be
controlled by a combination of train control and station control systems (including manual
systems). System availability is not often properly controlled, but acts as a limiting factor
applying to all aspects of the railway system, including the control system.

4.1 HEADWAY
As shown in Figure 10, with a given train control system, the signalling and platform dwell
components combine to define the theoretical point capacity, or headway. The relationship
between headway, line speed and the train control system’s signalling strategy has already been
considered in section 3.1. That discussion will not, therefore, be repeated here. However,

Page 16 of 22
signalling strategy is not the only component of a train control system that influences achievable
capacity.
Table 2 shows typically achievable point capacity headways for different types of train control
systems on a typical metro railway (note: these examples are derived from LUL experience).

Train Control System Type Typical Achievable

Headway (tph)
Conventional trainstop system, manual driving 28
Conventional trainstop system, manual driving and speed control 31
Coded ATP system with manual driving 31
Coded ATP system, ATO driving 34
Moving block transmission based system, ATO driving 36/37

Table 2: Achievable Headway (White et al. 1998, p4)

It can be noted from Table 2 that the introduction of speed control or full ATP increases the line
capacity by around 3 trains per hour. This is largely due to a reduction in required overlap lengths
when control of the trains’ approach speed can be enforced. The introduction of ATO to the
system also provides an improvement in achievable headways of around 3 trains per hour. This is
largely due to the greater consistency in driving performance that automatic operation allows,
which in turn reduces the number and size of minor perturbations that occur to the service and
enables the use of smaller recovery margins.
Similar headway improvements have also been predicted on main line railways as advanced
control and automation systems are introduced. Studies on behalf of NS Reizigers have, for
example, suggested that a 10-25% increase in line capacity should be possible in the Netherlands
if a modern ATP/ATC system were to be used on their existing infrastructure (Holtzer 1999,
p587).
This shows that the capacity of the railway depends significantly on the type of train control
system that is in place.

4.2 OPERATIONS CONTROL

The provision of a recovery margin within the service interval provides a capability to maintain
capacity despite minor service perturbations and variations in driver performance. However, this
capability must be managed in order to ensure that the promised ‘recovery’ occurs without
serious disruption to the scheduled service. The systems and processes providing this manage-
ment of minor perturbations to the timetable are known as ‘regulation’.
Where more serious service abnormalities occur, it is also necessary to manage the service in real
time to ensure that train destinations are appropriately balanced, that bunching and conflicts are
minimised, and that staff and rolling stock resources are available when and where required. This
function is known as ‘service control’.
The combined functions of regulation and service control are referred to as ‘operations control’.

Page 17 of 22
The structure of an ideal operational control system is outlined in Figure 11. On main line
railways in the UK, the functions required within this are currently divided between a number of
staff roles:
 Signallers deal with minor perturbations and changes to the planned sequence of trains
at junctions, assisted by systems such as Automatic Route Setting (ARS);
 Signalling supervisors deal with local perturbation management, such as re-routing
trains from fast to slow lines and changing the planned train order at junctions;
 Station Supervisors / Managers deal with station control activities, such as re-
platforming;
 Train running controllers deal with more serious perturbations by re-timing,
cancelling, diverting or terminating short of destination any affected train on a given
line or geographic area;
 Stock controllers deal with faulting, re-allocation of stock and ensuring availability of
stock for maintenance;
 Train crew resource supervisors deal with changes to crew allocation following an
incident;
 Infrastructure maintenance controllers deal with infrastructure (signalling, track, etc.)
maintenance and faulting activities.

PLAN
Timetables Trains
Stock Diagrams Stock MONITORING
Crew Diagrams
Crew Deviations
Conflicts
Possessions Plan
Track
System Log
Actual
Train Position Trains
Stock Condition Stock REVISED PLAN
GENERATION Signalling Control
Crew Position
Crew Constraints Station Staff
Track Condition
and Signal Status Track Optimisation Passenger Information

Figure 11: The Ideal Operations Control System (based on Hurley 1999, p132)

In most railway systems service control (major conflict resolution, reversals, diversions and
allocation of stock / crew) is performed manually, with little obvious link to other aspects of the
train control system. Indeed, a survey of European railways in 1993 revealed that there are very

Page 18 of 22
few automatic solutions to dealing with serious perturbations, such as signalling failure (IRSE
1993, p15). However, with modern computer techniques there is an opportunity to consider the
provision of enhanced decision support tools (such as simulators running faster than real time) or
even the automation of some, if not all, of the service control functions in order to provide for
more efficient operation. This is as yet a vision for the future in the UK. To date, the closest we
have come is arguably the use of train graph displays to assist signallers at the Cheriton
Eurotunnel and Ashford CTRL control centres, or perhaps the ‘Control Centre of the Future’,
providing controllers with up to date train running information for the whole network, installed in
many signal boxes on Network Rail infrastructure. Other initiatives (such as the Network
Management Centre for WCML and the LUL Central Line ATR) have been launched with the
aim of providing improved network control, including predictive simulation for ‘what if’
scenarios, but these have yet to come to fruition. Clearly this is an area where there is room for
improvement in future railway control systems.

4.3 PLATFORM MANAGEMENT

Reducing the time that a train is stationary in a platform (the dwell time) is a major concern for
both main line and metro railways that operate with a high traffic density. Extended dwell times
result from a number of delays, the effects of which can be limited by implementing systems and
techniques for platform management:

 Door opening: Enabling door opening, passenger operation of opening devices and
actual door movement generally takes two to three seconds (Howarth
1999, p157). It is possible to gain reductions in this delay by releasing
the doors at slow speed (as is the practice on Paris Metro). Research on
LUL’s Victoria Line has also shown that poor stopping accuracy with
ATO has significant effects on door opening delays, due to the need for
driver intervention if sufficient accuracy has not been achieved (Horsey
2000, p52);
 Alighting/boarding: Passenger alighting and boarding times are largely determined by the
distribution of passengers within the train and along the platform, the
size of any gap between the platform and train, the number and size of
doors and vestibules on the train and the size of the platforms (Howarth
1999, pp157-8; McKenna 1998, pA1/4). Public address announcements,
audible warnings as doors are about to close and playing military style
music have all proved effective for reducing these delays (Harris et al.
1992, pp53-4; Horsey 2000, p22; McKenna 1998, pA1/4). Elsewhere,
providing platforms on both sides of a train to segregate boarding and
alighting, measures to distribute passengers evenly along the length of
the train, platform queuing and additional platform staff attendance have
also helped (Horsey 2000, p23). The number of people congregating on
the platform also increases with the time between trains, causing
extension of alighting and boarding times (Horsey 2000, p72). The
capacity of the train control system (which defines the minimum steady
state train interval) and its reliability are therefore also of significance
(poor reliability causing train delays and, therefore, congestion on both
trains and platforms);

Page 19 of 22
  Door closing: Studies have shown that operating trains with guards reduces door
closing delays, whilst use of ‘Close Doors’ and ‘Right Away’ signs
activated by station staff increases them (Howarth 1999, p158). Re-
ductions can also be gained by the introduction of countdown clocks to
advise drivers of time to departure and the development of an ‘every
second counts’ culture amongst staff (Horsey 2000, p23).
 Traction application: DB in Germany have reduced the delay between door closure and start
away by allowing pre-selection of the traction system, so that trains
automatically start away when the doors are closed (Howarth 1999,
p159).

Most of these systems and techniques can be implemented without the need for major
infrastructure changes. Platform management therefore offers potential for optimising the
utilisation of capacity in both new and existing railway control systems.

4.4 AVAILABILITY
Whilst improving the service control functionality would enable the railway to cope better with
service abnormalities, it would be far better to avoid the occurrence of those abnormalities and
their consequent disruption to services altogether.
In May 2000, Railtrack’s Performance Director wrote that a major source of overall delay to the
Railtrack network was from ‘problems with reliability of infrastructure’ (Curley 2000, p49).
Indeed, responsibility for around 40% of all railway delays in the UK is attributed to the
contractors responsible for maintenance, repair, minor renewals and rapid response faulting of
fixed infrastructure alone (Winder 1999, p2). “To the extent that basic equipment can be made
more reliable there is a corresponding opportunity to reduce spare capacity and increase delivered
capacity” (McKenna 1998, pA1/4). The availability of railway control systems therefore
represents a major factor in the overall capacity that a railway can achieve.
The availability of the railway control system could be improved in a number of ways, including:
 Improved maintenance - it has been estimated that the combination of poor
specification and inadequate execution of maintenance accounts for 64% of all
infrastructure related delay minutes (Curley 2000, p49)
 Improved equipment access – most railway infrastructure is currently located near to
the track, making maintenance access difficult. Future design of control systems to
keep as much of the equipment as possible out of the track environment would greatly
assist maintainers in gaining adequate maintenance time (Winder 1999, p3)
 Improved equipment reliability – this could include the design of components to
reduce fault rates or the use of fault tolerant design such as active redundancy (with
multiple sub-systems operating concurrently), standby redundancy (with an operating
sub-system and one or more back up systems available to be switched in if a failure
occurs) or graceful degradation (where the system is designed to continue operation at
a lower level of functionality if a sub-system fails) (IRSE 1993, pp5-6, 9)
 Improvements to equipment repair times – use of condition monitoring equipment and
diagnostic tools to assist maintenance staff, together with modular designs for ease of
repair or replacement (Winder 1999, p4)

Page 20 of 22
  Rule changes – railway operating rules tend to impose a heavy delay penalty in the
event of equipment failure. For example, following failure of a track circuit all trains
must be stopped and cautioned. If, instead, the rules permitted the use of suitable
signage to inform approaching drivers that a signal is ‘out of use’ and advise an
appropriate speed restriction, delays could be significantly reduced (Winder 1999, p4).

It is worth noting that the advantages of improved availability go beyond capacity benefits. For
example, control systems may be designed to ‘fail safe’ by stopping all train movements. When
this happens, operation of the railway continues in accordance with rules and human centred
control processes. This transfers safety responsibility to the human operators and, in so doing,
introduces additional risk to the operation of the railway. In fact, it can generally be said that in
the railway domain ‘the lower the system availability is, the lower the system safety is’ (Boycott
et al. 1999, p1). The provision of alternative systems designed with inherent fault tolerance and
ability to operate in degraded modes will therefore not only increase the availability of the system
and its capacity during times of technical system failure, but will also provide a means for the
system to maintain safety besides reliance on the human element (Francis 1994, p1).

5 CONCLUSIONS
It is clear from the definitions given within this paper that there is no one measure of railway
capacity that provides all of the information required by all parties interested in the analysis of a
railway’s performance. However, a number of measures of capacity do exist which, between
them, can provide most of the information required to determine what the capacity of a railway
should be, and how to meet that requirement. Where most of these existing measures fall short,
however, is in providing customer focused measures of capacity that relate the actual service to
the customer’s demands. Therefore, the consideration of optimum line capacity through analysis
of the total journey time imposed by the service offers a significant opportunity to identify
potential gains in service capacity as seen by the user.
Both the IDEF0 system models and the capacity hierarchy outlined within this paper show the
importance of a systems view of the railway if the achieved capacity is to be optimised, both
from a technical viewpoint and from that of the customer. If the customer is to be provided with
an optimal service, it is therefore of great importance that all railway disciplines consider the
wider effects of factors that come under their control and work together towards the goal of
improving the overall system.
Capacity can clearly be designed into train control systems by selection of appropriate
technological strategies for signalling and operation. However, if such systems are ever to
achieve their full potential, the wider railway control system (including both technologies and
processes for train, station and line control) must also be designed to support that operation.
In addition to this, the capacity offered by railway control systems must be available when
required by the railway’s customers. The systems implemented must provide, therefore, adequate
availability and, where necessary, the facility to continue operation following equipment failures
by use of redundancy and degraded modes of operation.
Clearly there is scope for future improvements in the provision and management of railway
capacity.

Page 21 of 22
6 REFERENCES
Boycott P, Berieau M, ‘Assessing Cost Efficiency and Safety of the ERTMS Technology’, 6 th Annual Rail Safety Congress, AIC
Worldwide, 1999
Curley J, ‘Railtrack’s Role In Improving Performance’, Modern Railways, p48, May 2000, 2000
Francis J D, ‘From The Signal Box Window’, IRSE, 1994
Goddard E, ‘Supervision and Operation of Mass Transit Systems’, IEE Power Division Sixth Vacation School On Railway
Signalling & Control Systems, IEE, 1996
Harris N G, Godward E W, ‘Planning Passenger Railways’, Transport Publishing Company, 1992
Holtzer J, ‘NS Reizigers needs ETCS and new lines to handle growth’, Railway Gazette International, September 1999
Horsey C I, ‘Railway System Performance Improvements on an Existing Metro’, MSc Dissertation, Dept. of Mechanical
Engineering, Sheffield University, Sheffield, UK, 2000
Howarth D A, ‘Capacity Achievement on Thameslink 2000’, International Conference on Railways as a System, IMechE, 1999
Hurley J, ‘Optimized Control of Railway Operations’, International Conference on Railways as a System, IMechE, 1999
IRSE, ‘IRSE Technical Committee Report No. 2: Operational Availability Of Railway Control Systems’, IRSE, 1993
LUL, ‘Business Case Development Manual’, LUL, 1994
McKenna P, ‘The Needs of the Business and the Operator’, IEE Power Division Seventh Vacation School On Railway Signalling
& Control Systems, IEE, 1998
McCormick H, McKenna P ‘Signalling and train Control Systems – Requirements of the Business and Operating Company’, IEE
Power Division Sixth Vacation School On Railway Signalling & Control Systems, IEE, 1996
Millard D, ‘Delivering a Metro Train Service-Managing the Performance of Key Components’, International Conference on
Railways as a System, IMechE, 1999
Pachl J, ‘Railway Operation and Control’, Institut fur Eisenbahnwesen und Verkehrssicherung, 2000
Railtrack, ‘Engineering Safety Management Issue 3, Yellow Book 3, Volumes 1 and 2: Fundamentals and Guidance’, Railtrack,
2000
Schmid F, ‘Train Control Research in Europe’, IRSE, 2002
White C, Millard D, ‘Metro Signalling Operations’, IRSE, 1998
Winder K, ‘The Needs and Expectations of Train Operators’, AEA Technology Ltd, 1999

Page 22 of 22