Derailment Containment Provisions PDF

Ref: R00673
Report on the findings of:

Current practice and
effectiveness of derailment
containment provisions on high
speed lines
Issue 1
Zoetermeer, Netherlands
30 September 2004
HSL-Zuid
Submitted by:
This report is confidential and intended solely for the use and
information of the company to whom it is addressed.
Executive Summary
Overview of Current Practice
The findings indicate that there is a general acknowledgment that Derailment Containment
Provision (DCP) is a positive approach to minimising derailments although there was:
• No single current practice to derailment containment provision,

• No substantive reasons for the provision of DCP,
• No standard used in applying DCP,
• No definite opinion on the effectiveness of DCP,
• No DCP is 100% effective.
The majority of projects and organisations looked to use DCP Type 1 or DCP Type 3 at high
risk locations and there was widespread agreement that DCPs are not used as a matter of
course throughout the system.
Estimation of Effectiveness of DCP
There is insufficient information to make a quantified decision on the effectiveness of DCPs

to confidently predict how a train will behave in a derailment at speeds above 200 km/h.
Furthermore, determining effectiveness is problematical, due to the complex variable factors
to be considered when trying to evaluate the effectiveness of DCP and every accident is
unique.
After a high level qualitative assessment, DCP Type 3 is the most effective form of DCP on
high speed lines based on high level causal events. This high level assessment is supported
by organisations which operate high speed lines such as CTRL and Skinkansen who
implement DCP Type 3.
The effectiveness of DCP can be enhanced through mixing DCP Types, with DCP Type 3 at
switches in conjunction with DCP Type 1 to be potentially effective. The least effective form
of DCP was Type 2.
The most effective approach to managing the risk of derailments is to focus on “Step Zero”,
the prevention of the causal events. Many of the causal factors associated with derailment
could be mitigated to some degree by the detection / prevention of faults on the vehicle or
the infrastructure.
This attention to managing derailment risk includes the system design phase, where there
needs to be a realistic expectation regarding the maintainability of the equipment. The
designer should be careful about making design assumptions which can have long term
consequences.
Furthermore, through the lifecycle of the project and its operational lifecycle, when a change
to the system is proposed, the relevant custodian of the system should evaluate the effect of
the change across the whole system environment.
Adverse Effects of DCP
It can be stated that under certain conditions, DCP could contribute to the escalation of a
derailment, or to the derailment in the first place. These being:
• The overall continuity of the DCP,
• Adverse effects related mainly to the maintenance elements,
• Debris between the running rail and the DCP rail causes or escalates derailment,
• Deutsche Bahn indicated that they limit the use of DCP Type 1 because of the
difference in passenger carriage bogie and axle designs,
• Under coach arrangement of brake rigging and traction equipment may also interfere
with the guide rails and cause a more dangerous situation after a derailment,
• An increase in escalation with DCP Type 3, since the barrier could cut through the
passenger areas on a double-decker train.
HSL-Zuid Application
On the HSL-Zuid line there is a combination of ballasted track and Rheda slab form track. In
the instance of a derailment on Rheda slab form track there is no ballast to act as resistance
to slow the train down. Furthermore the structure of the Rheda track, such as the concrete
boots for the base plates could accelerate the escalation of the derailment and derailment
escalation since the wheel set could “fall off” the Rheda track bed slab.
Infraspeed's proposed concrete upstand between the rails would appear to be a sensible
provision, though it should be designed so as to ensure the outside derailed wheel remains
securely upon the track slab. There is a concern with the interface design on HSL-Zuid
between the vehicle and the DCP, since the brake disc may ride on top of the concrete plinth
DCP and cause potentially serious consequences.
Recommendations
1. Undertake a critical evaluation of the design assumptions and inputs for the civil
deign and system design (especially with regards to the Rheda Slab Form Track) in
and validation of the design within the boundary of the HSL-Zuid Transportation
System.
2. Independently review the HSL-Zuid Transportation System safety management

system to ensure it is effective, as far as reasonably practicable, in eliminating
derailments since this is the most effective solution to reduce escalation. This would
include for example:
• Ensuring that appropriate risks are being adequately managed i.e. fires in tunnel
systems;
• Ensuring that a consistent approach (whilst understanding where this is not
applicable) to derailment issues are taken into account in operations, training and
procedure definition as appropriate;
• Appropriate assurance regimes in place in order to ensure that installation,
inspection and maintenance are undertaken.
DERAILMENT CONTAINMENT PROVISION
ISSUE 01 SEPTEMBER 2004
Table of Contents
1 Introduction 6
1.1 Background to the Assignment 6
1.2 The Assignment 7
1.3 Methodology and Approach 7
1.4 Report Structure 12
2 Technical 13
2.1 Part A: Overview of Current Practices and Existing Knowledge 13
2.1.1 Interviews 13
Results 14
2.1.2 Media Search 19
Results 19
2.1.3 Part A: Summary 21
2.2 Part B: Effectiveness & Part C: Adverse Effects 23
2.2.1 Causes of Derailments 24
2.2.2 Relationship of Derailment Cause and Escalation 24
2.2.3 Review of Derailment Containment Provisions 25
2.2.4 Expert Panel Assessment 26
2.2.5 Expert Panel Commentary & Methodology Step 10 38
3 Summary 45
3.1 Overview of Current Practice 45
3.2 Estimation of Effectiveness of DCP 45
3.3 Adverse Effects of DCP 47
3.4 HSL-Zuid 47
3.5 Recommendations 48
4 References 50
5 Glossary and Abbreviations 51
6 Appendices 53
A Standard Questionnaire
B Detailed Questionnaire Results Table
C Summary of Questionnaires, Interviews & Media Search
D Causes of Derailments
E Infraspeed Concrete Plinth Illustrations
F Accidents for Expert Panel Assessment
G Derailment Accidents - Further Reading
Page 4 of 92
REPORT REFERENCE: R00673

List of Tables
Table 1 Activities for Part A 8

Table 2 Activities for Part B prior to Expert Panel Session 9
Table 3 Activities for Part B during the Expert Panel Session 9
Table 4 Activities for Part C 10
Table 5 Railways/Organisations for Face to Face Interviews 13
Table 6 Railways/Organisations for Telephone Interviews 14
Table 7 Methodology Step B1. Accidents Identified For Expert Panel Investigation 27
Table 8 Methodology Step B3. Expert Panel Findings 28
Table 15 Qualitative Assessment of DCP’s 43
Page 5 of 92

Section
1 Introduction
This report forms the Booz Allen deliverable in response to the letter from the HSL-
Zuid organisation (ref HAVL/517575 [AD1]) requesting a study into the current
practice and effectiveness of derailment provisions on high speed lines.
1.1 Background to the Assignment
The HSL-Zuid organisation, are responsible for the delivery of the HSL-Zuid
Transportation System with the requisite level of safety. Derailment risk is one of the
main issues in the safety management of the project. The high traffic density, large
number of overpasses/structures/transitions and extensive use of elevated track
significantly increase the consequences of a derailment on HSL-Zuid.
A key element of the HSL-Zuid derailment risk limitation strategy is limiting the
consequences of a derailment. This approach provides a second line of defense in
managing derailment risk, (the first line of defense being the prevention of the
derailment occurring in the first place, the last being emergency response). In the
event that a train has derailed, the measures introduced as a result of this element of
the derailment risk limitation strategy, aim to reduce the loss of life, injury, property
damage and economic damage of that derailment.
Whilst a train might be derailed, the consequences of that derailment may be limited
provided the train does not collide with another object (e.g. civil structure or another
train) nor fall from a structure (e.g. elevated viaduct, bridge or embankment) nor roll
over. This principle is most graphically demonstrated by comparison of the
derailments of TGVs at Haute Picardie station in 1993 and the ICE at Eschede in 1998.
In the case of the TGV, the train remained upright and did not encounter any
obstructions. There were no deaths and few injuries. In the case of the ICE, the
derailed train collided with an overbridge, resulting in many deaths and extensive
damage. Whilst there are a number of variables involved with these two significantly
different results, it is clear that in the case of the ICE, escalation of the derailment was
a major factor as the train encountered a rigid object due to travel beyond the
planned track alignment.
The goal of Derailment Containment Provisions (DCP) is to reduce the probability of

a derailment escalating by increasing the probability that a derailed train will:
• Follow the track;

• Not enter the free space profile of an adjacent track;
• Remain upright;
• Not encounter a rigid object.
Provision of DCP falls within the scope of the Infrastructure Provider to the extent
that it is required to achieve the Safety Case and specific targets detailed in the
Implementation Agreement.
Page 6 of 92

Recent reports commissioned by the Infrastructure Provider have cast doubts on the
viability of some of these targets. The Infrastructure Provider has presented a
position where targets are not achievable. A major part of the Infrastructure
Providers argument is founded on the degree to which DCP can reduce the
probability that a derailment will escalate. It has also been stated that DCP could
increase the probability of escalation.
1.2 The Assignment
In managing the delivery of the HSL-Zuid Transportation System and the

Infrastructure Provider contract in particular, the HSL-Zuid organisation wishes to
act as the informed client and reach a better understanding of current practice and
effectiveness of DCP on high speed lines. To achieve this and to allow a fair appraisal
of the case presented by the Infrastructure Provider, the HSL-Zuid organisation has,
in part, commissioned this assignment. Three separate parts of the assignment were
identified. These were:
Part A An Overview of current practice and existing knowledge. Requiring

discussions with railway organisations and a search of existing
knowledge on DCP effectiveness;
Part B An estimation of effectiveness of whether DCP will work for trains

running 200-300 kph. Three configurations were identified for the
purpose of this assessment, these being:
DCP Type 1 provisions mounted between the running rails,

intended to guide the derailed train at wheel level;
DCP Type 2 provisions mounted outside of the running rails,
intended to guide the derailed train at wheel level;
DCP Type 3 provisions mounted outside the running rails intended
to guide the train at bogie level;
Part C An Assessment of the adverse effect of DCP.
The full Terms of Reference for this assignment are contained in [AD2]. This report
provides a discussion of derailments and how and why they escalate. It also reviews
several significant derailments that have occurred in Germany, United Kingdom,
Australia, and the United States to determine what may be learned from these real
world occurrences. This report also provides our findings on each of the three parts
of the assignment and draws conclusions and recommendations, based on our
findings and analysis.
1.3 Methodology and Approach
The methodology adopted to approach this assignment was founded on the three
separate parts of the assignment. It was considered both prudent and practical to
address Part C of the assignment (detrimental effects) as an integral part in the
execution of Parts A and B.
Page 7 of 92

During Part A of the assignment it became apparent that organisations and projects
were unwilling or unable to discuss the sensitive issue of derailments and DCP due
to a variety of reasons, e.g. lack of data collection processes, commercial exposure. In
addition, the numbers of derailments above 200 km/h are thankfully small. As a
result the lack of data throughout the world, the original quantified approach which
included estimating the probability of escalation, given initial derailment for the
determination of DCP effectiveness as detailed in the assignment Terms of Reference
[AD2] was un-achievable.
The approach to the assignment was modified in agreement with the client (AD3) to
a qualitative approach. The revised methodology and approach relied heavily on
interpretation by an expert panel, which had a wide range of expertise, covering
technical, operational, research, design, construction, maintenance and investigation
skills. The revised approach and methodology is detailed below in Table 1 to Table 4
and is illustrated in Figure1.
Part A Overview of current practice and existing knowledge

Task Activity performed
A1. Compiled a list of projects and organisations to be interviewed. The list was agreed
with the HSL-Zuid organisation.
The list identified projects and organisations worldwide but concentrated on
European countries. The basis for selection was relevant experience associated with
derailment and / or high-speed line operation.
A2. Developed a standard questionnaire which ensured a consistent approach to the
interviews and focused the interviewer on gaining appropriate and relevant data
for Parts B & C of the assignment. The questionnaire was agreed with the HSL-Zuid
organisation and is included as Appendix A to this report. The questionnaire was
distributed to selected organisations.
A3. Conducted face to face interviews with associated organisations and projects which
had similarities with HSL-Zuid, to ensure a full understanding of the issues and
implications.
A4. Conducted telephone interviews where face-to-face interviews were not possible.
A5. Undertook a worldwide overview of existing knowledge about effectiveness and
possible adverse effects of DCP. This overview covered academic sources as well as
sources from within the railway industry and involved:
• Literary review;
• Internet search;
• Contacting organisations for data;
• Internal sources within Booz Allen.
A6. Collation and tabulation of the captured data to enable the information to be
analysed. Any discrepancies in the captured data or white spots were identified and
supplementary interviews, investigations were conducted. The results of this
tabulation are included as Appendix B to this report.
Table 1: Activities for Part A
Page 8 of 92

Part B Assessment whether DCP will work for trains running 200-300 kph
B1. The expert panel:
a) Analysed the common trends, operational similarities and key
characteristics of the information gathered in Part A.
Pre Expert Panel Session
b) Reviewed the Questionnaires obtained from High Speed

Operators.
c) Identified and justified accidents for the Expert Panel review.
B2. Reviewed the accident synopsis, which provided factual information on

each derailment under review.
B3. Identified a comprehensive list of the causal factors based on the synopsis
from the selected accidents.
Table 2: Activities for Part B prior to Expert Panel Session
The expert panel meeting was a discussion of a number of accidents and expert opinion on how the
outcome of the accident would vary based on the application of Derailment Containment Provisions
(DCPs). The expert panel reviewed each accident and attempted to draw conclusions about DCPs.
Steps B4 to B7 were executed for each of the accidents. Drawing on the results of this analysis,
activities B8 to B 10 were concluded.
B4. Discussed escalation of the Accidents covering:
a) Did Escalation Occur?
b) When did Escalation Occur?
• At the time of the Derailment
• Some time after the derailment
During the Expert Panel Session
c) What caused the derailment Escalation to Occur?

B5 Reviewed if DCPs were in place at the derailment site?
Yes – DCPs were in place in the area of the derailment
• What type of DCPs?
• What was the effectiveness of the DCP?
No – DCPs were not in place in the area of the derailment

• How would application of DCP Type 1, DCP Type 2, DCP Type 3
have affected the accident including speeds at 200-300 Km/h?
B6. Identified steps that could have been taken to prevent escalation of the
derailment:
a) Infrastructure
b) Vehicles
c) Operation
d) Human factors
e) Others
Table 3: Activities for Part B during the Expert Panel Session
Page 9 of 92

Task Activity Identified steps that could have been taken to prevent the derailment:
performed a) Infrastructure
b) Vehicles
c) Operation
d) Human factors
e) Others
B8. Expert panel identified derailment causal factors
B9. Identified additional risks that may be associated with each type of
DCPs.
B10. Expert panel key findings
Table 3 Continued: Activities for Part B during the Expert Panel Session
Part C Assessment of potential adverse effects of DCP

General Determined and assessed the adverse effects of DCP in parallel with Part A
and Part B of the assignment. Whilst undertaking the Part A and Part B of the
assignment, the data collected and analysis performed addressed the potential
for any extra risk that could be introduced as a result of installing DCP.
The assessment considered direct impacts, such as dragging train equipment,

interaction with derailment provisions that may not be continuous, but also
indirect impacts, such as issues with maintainability, cost or system
availability.
Interaction Interaction with Part A included:
in Part A. • Inclusion of specific questions relating o adverse effects in the data
collection questionnaire;
• Addressing adverse effects explicitly in the interviews;
• Searching on adverse effects during the worldwide overview of
existing knowledge;
Identification and isolation of adverse effects in the collation and tabulation of
the available data.
Interaction Interaction with Part B included:
in Part B • Evaluation of negative effects of DCP on test derailment examples,
• Determination of negative influences on overall effectiveness,
• Identification of additional risks associated with DCP,
• Consideration of negative effects in determination of the way forward.
Table 4: Activities for Part C
Page 10 of 92

A1. Complied a list of

organisations & Projects
A2 & C1.
Produced Questionnaire
PART A
Data
Collection
& PART C, A3. & C2. A4. & C3.
A5. & C4.
Adverse Effects Conducedt Conducted
Worldwide
Data face to face telephone
media search
interviews interviews
A6.& C5. Collated Information

Reported on the
findings of Part A &
Part C of the
assignment
B1. Analysed Information
B2. Expert panel reviewed accident scenarios
B3. Identified causal factors
B4. Analysed escalation
PART B
Technical B5. Assessed effectiveness of DCP in
Analysis Including accident scenarios
Part C Technical
Analysis of
Adverse Effects B6. Identified steps to prevent escalation
B7. Identified steps to prevent derailment
B8. Expert Panel identified derailment causal

factors.
B9. Identified risks introduced by DCP Reported on the

findings of Part B &
Part C of the
B10. Expert Panel Key Findings assignment. The
expert panel
assessment on the
effectiveness of DCP
Figure 1 - Methodology Flow Chart
Page 11 of 92

1.4 Report Structure
This report is structured as follows:
Section 1 Introduction
Provides a background to the assignment, details of the assignment
itself and the methodology used for the execution of the assignment.
Section 2 Results
Provides the results of the interviews and process’s of the expert panel
activities. These results are then tabulated and inferences drawn from
them
Section 3 Conclusions
Details the conclusions drawn from the analysis.
Section 4 Recommendations
Provides recommendations coming from the assignment with cross
reference to the appropriate section of the report with the
substantiating data.
The base data from the interviews and the expert panel is provided in the appendices
and references.
Page 12 of 92

Section
2 Technical
2.1 Part A: Overview of Current practices and existing

knowledge
Part A of the assignment consisted of gathering information in order to achieve an
overview of current practices and existing knowledge in DCP. This was achieved
using a series of interviews and a media search. The results for these activities are
detailed below.
2.1.1 Interviews
The investigation included interviews with projects and organisations in Europe,
North America, the Far East and Australia and involved face-to-face interviews and
telephone interviews. The projects and organisations that were contacted are listed in
the following tables. Those organisations with which were conducted face to face
interviews are shown in Table 5 whilst those with whom telephone interviews were
conducted are shown in Table 6. The interviewees were motivated by supporting
the open exchange of information to benefit the rail industry though out the world.
An in this culture, HSL-Zuid offered to share the final report with each of the projects
or organisations providing input information.
Railway / Organisation Contact

Deutsche Bahn AG, Germany Dr. Gunnar Baumann
Director of track techniques
Société Nationale des Chemins Mr. Johan Verschaeve, Safety Manager (R+D), B-
de Fer Belges, Belgium Rail (SNCB)
Øresund Link, Johnny Restrup-Sørensen Director of Infrastructure
Denmark/Sweden
Central Japan Railway Takeshi Inagawa, manager, representing Central
Company Japan Railway Company
Société Nationale des Chemins No response
de Fer Français, France
High Speed Lines France R Dayez, French Ministry of Transport
Channel Tunnel Rail Link, UK Lorna Small Rail Safety Manager & Ken Harvey
Risk Manager
Table 5: Railways/Organisations for Face- to-Face Interviews
Page 13 of 92

Railway / Organisation Contact

Taiwan High Speed Rail Mike Baxter, Head of Track Design
Corporation
U.S. Department of Robert M. Dorer
Transportation Acting Deputy Director, Office of Safety and Security
Volpe National Transportation Systems Center, Chief of
the High-Speed, Ground Transportation Division
Research and Special Programs Administration
Chairman of a TRB committee (High Speed Ground
Transportation technical topics)
Massachusetts Institute of Professor Joseph M. Sussman
Technology Center for Professor of Civil and Environmental Engineering and
Transportation Studies Engineering Systems
Information on Japan
Great Belt (Storebælt) Jørgen Kampmann Senior Engineer, Risk Analysis COWI
Railway Link Denmark A/S
LTF (Lyon Turin high Lorenzo Brino, Assistant Director Technical Studies,
speed link), France/Italy Lyon Turin FerroViaire & Eddi Verbesselt, Risk Analyst
Italy Ing. Natoni, Chief Engineer permanent way at Italferr up
to last year, a renowned expert in Italy
Table 6: Railways/Organisations for Telephone Interviews
Results
Whilst undertaking the interviews, some organisations were willing to provide an

overview of DCP, however they would not complete the questionnaire. The
completed questionnaires are summarised in Appendix B and the original completed
questionnaires are contained in [RD1]. The major findings of the interviews are
listed below:
DEUTSCHE BAHN
The Eschede accident is a major incident in German rail history and is

covered in detail in Part B of the report.
Track standards require that guard rails are needed only on lines with mixed
traffic under certain conditions. They are not used on the Cologne –
Frankfurt/Main route.
There have been two instances where DCP Type 1 has been successful. In
1992 a train involved in a collision on a bridge was prevented from falling off
of the bridge by guard rails, and in 1996, impact to an abutment of a road
bridge was prevented by guard rails. In addition to guard rails, passive
protection systems along roads and on road bridges are used to prevent the
deflection of vehicles onto the track.
Page 14 of 92

SNCB
DCP Type 1 protection is used especially on high embankments and on

bridges. In specific areas where there is a possibility of subsidence, then DCP
Type 3 is used, i.e. concrete platforms.
SNCB are proactive in trying to prevent derailments and undertake various

preventative measures. These include:
• Hot axle box detectors used as a preventative measure.

• Each morning the lines are declared secure for operation after a ride of an
opening train (train balai according to the French practice). This opening
train can be a commercial or non-commercial ride of any train at 160 km/h
maximum. These opening rides are very helpful to avoid unexpected
obstacles when the rides at 300 km/h start. SNCB stated that with regards
to the opening train, “It is considered as a major contribution to our DCP-strategy”.
• On high speed lines, the escalation of the derailment can be reduced by
the use of a special rolling stock (high speed trains with bogies between
the cars which assist in keeping the train as one).
• The formation of a specific multi-disciplinary team for maintaining the
DCP on the high speed line.
GREAT BELT (STOREBÆLT) RAILWAY LINK
DCP Type 1 is used on the west bridge and within tunnels DCP Type 3 is
used in the form of concrete walkways. Furthermore, there are derailment
preventative measures employed. These include hot axle box detectors and
gauge control facilities at the entrance to the crossing.
It is noted that in Denmark there has been one derailment on a bridge. In this
case DCP Type 1 served its purpose. In 1978 there was a derailment on the
Storstrøms Bridge which carries a single track railway and a two lane
highway between Zealand (Sjælland) and the island of Falster, south of
Zealand. A lamp post fell across the track which was subsequently hit by a
locomotive. The locomotive derailed and was caught by the DCP.
ØRESUND LINK, DENMARK/SWEDEN
DCP Type 3 in the form of concrete and steel “side walls” or guiding
devices are used. These were part of the bridge/tunnel construction, so their
instalment did not trigger additional investments. The concrete option is
used in the tunnel and on the approach bridges, where as steel is used on
the span bridge due to constructional reasons (weight). Prior to the
entrances to tunnels and the crossing there are dragging equipment
detectors and on the trains there are rail break detectors.
Page 15 of 92

JAPAN
DCP Type 1 is used at bridges, in other specific high risk areas DCP Type 3
is used. It is of significance to note the response from Japan Eastern:
“In the case of the SHINKANSEN, a derailment accident in a high-speed run

has not occurred for 40 years since we started service in 1964. We have
confidence that the SHINKANSEN will not be derailed in a high-speed run by
internal factors in JR East, because we have done strict maintenance of wheels
and rails, and we have past results which never have been occurred. The
SHINKANSEN basically runs through a exclusive track which is surrounded
with soundproofing walls, consequently the possibility of collision with other
structures after a train was derailed to the truck outside such as the accident in
Eschede will be extremely low. We think that it is important to detect accident as
early as possible and to limit speed of trains when we have some troubles in the
track and an earthquake occurs in Japan.”
The cultural perspective of the Shinkansen is interesting. Drivers and

maintainers etc. comply with the rules and regulations rigorously. The level
of training is extremely high and is, of course, based on the foundation of an
extremely disciplined society.
In addition to this, the management of the infrastructure is such that

passenger services are complete by midnight and do not commence till 6am
the following morning. This allows a regularly available window in which
maintenance work can be carried out. Within this maintenance regime a
special train (Dr Yellow) checks all aspects of the lines and overhead power
cables at normal Shinkansen operating speeds.
This response demonstrates alternatives available an in operation that can

allow railways to operate at high safety levels with limited DCP. We
consider this to be an important indication when making informed decision
concerning DCP.
Page 16 of 92

HIGH SPEED LINES, FRANCE
DCP Type 1 is applied on bridges i.e. the bridge over the river Rhone
and at transition curves where the alignment changes. Mr R Dayez
from the French Ministry of Transport stated:
“In the few derailments that have occurred, as on HS lines as on

conventional lines the TGV has followed the track and stayed upright.
This is mainly thanks to the characteristics of the fixed train set heavy
locomotive (motor coach) bogies situated at the connections between the
coaches (jacobsdraaistellen) which leads to stiff connections between
coaches.”
At over passes, columns supporting the road or rail that cross the line
are situated at some distance from the rail track. The running rail
normally limits the lateral displacement of a derailed train and
therefore the train does not hit the columns. And on elevated sections
of track, the embankment is made wide enough to support a derailed
train caught by the running rail.
Therefore in France a combination of DCP Type 1 and construction

design is used to reduce derailment escalation. This is also supported
by concentrating on the causal events and they believe the main risk
is motor vehicles falling from overpasses. In order to mitigate this
causal event, detection devices are attached to viaducts. Therefore
when a motor vehicle is detected as a potential derailment obstacle,
trains within the relevant area are stopped automatically through the
signalling system.
CTRL
CTRL produced a Derailment Containment Strategy (ALARP Risk

assessment) which recognised that no DCP is 100% effective and that the
attention should be focused on preventing the causes of derailments and
rather than managing the consequence of a derailment.
The approach to DCP was based on their Train Accident Model which is
an averaging model in the sense that it takes and average embankment
etc.
There was no justification for the provision of DCP if CTRL was to have
been a passenger only line. The additional risk introduced by mixed
traffic required DCP to by installed at strategic and high risk locations.
Due to the absence derailment data and evidence of DCP

effectiveness, the risk analysis used in assumptions and estimated
ranges and considered common derailments. The report identified
that check rails (DCP Type 1) and guard rails (DCP Type 2) were
not effective at high speed.
Page 17 of 92

CTRL - CONTINUED
CTRL undertook an approach which was a variation on UIC Leaflet 777-2

‘Structures built over Railway lines- Construction requirements in the
track zone’ [RD5]. In addition the positioning of the DCP Type 3 is
designed so that during a derailment the concrete kerb acts and the
running rail act as a barrier to the laterally displaced wheelset.
A key lesson from CTRL is that the theoretical approach to DCP as a

result of analysis has to be technically feasible and practicable and
compatible with the track bed design.
TAIWAN, THSRC
On the THRSC there are a

significant number of tunnels,
viaducts and bridges. THRSC has a
walkway/cable duct that serves the
purpose of the DCP, which runs
through the tunnels, viaducts and
bridges and is almost continuous.
The implementation of this type of
DCP was not based on an analytical
study but on expert judgment.
THSRC Construction showing

walkway extreme left
LYON TURIN HIGH SPEED LINK
This project is at the Functional Specification stage. The approach to DCP is

based on the approach taken on Channel Tunnel, and will adopt DCP Type 3
in the form of sidewalks on each side as was the case in Channel Tunnel. It is
worth noting that LTF carries freight and passengers.
ITALY
The approach taken on Italian high-speed lines compared to their traditional

lines is significantly different to other countries. The distance of the axes of
parallel lines is increased to lower the probability that a derailed train enters
the free space of an adjacent track, and the width of the ballast at each side
of the line is increased, doubling the effect of containing the wheels of the
derailed train and of dampening the impact of the wheels on the ground.
This approach to DCP is noticeably different to other countries in one
particular aspect. Worldwide, DCP Types 1,2 and 3 are generally only used
at areas of high risk, i.e. elevated lines and tunnels and switches, where as
the Italian approach to DCP which, although not one on the defined types in
this study, is used almost continuously along their high speed lines.
Page 18 of 92

2.1.2 Media Search

The approach to the media was to undertake an extensive search of the internet to
ensure that we covered the breadth of the subject area. However this approach did
not provide substantial detailed information. In order to capture detailed
information, we utilised individual experience and contacts within the rail industry
worldwide to identify and obtain detailed information. We also approached a non-
railway academic body to broaden the search.
Results
The worldwide media search data successfully identified additional material to that
provided in the interviews and questionnaire. A summary of the media search
information is shown in Appendix C.
The media search demonstrated that throughout the world derailment is a major risk
for transport systems. This risk is extended further than the safety implications, for
example the recent derailment in Sakarya, Turkey where 38 people were killed. The
State Railways Authority (TCDD) was found to be at fault in addition to the chief
machine operator and his deputy. This derailment during has resulted in significant
political changes.
In undertaking the media study the results indicate that the derailment provision
was not designed specifically for high speed lines, and generally there has been no
robust analysis to demonstrate the effectiveness of DCP on high speed lines.
In support of the change of methodology and approach to this assignment, the only
statistics we found on derailments were published by the Independent Transport
Safety & Reliability Regulator New South Wales in Australia (RD2). The statistics
confirmed that data is only available for low speed, or depot derailments.
Examining events in the Netherlands, the recommendations from the Baarn

derailment [RD3] look towards management actions in preventing derailments
rather than the provision of additional infrastructure as an effective way forward.
This is of significant relevance to HSL-Zuid who will be required to support this
recommendation.
The TTAC ‘ The review of Derailment Risk for the HSL-Zuid Railway’ [RD15]
identified that only three high-speed derailments have occurred involving trains
operating between 200 to 300 km/h. All three occurred in France and involved TGV
or Eurostar Services. The TTAC report determined the effectiveness of DCP in a
manner which included several caveats and limitations.
In the UK there is no common practice for DCP. The risk of derailments is assessed
on a case by case basis [RD4]. For example, the Health and Safety Executive required
the Heathrow Express Airport Tunnel project to incorporate derailment containment
throughout the lengths of the tunnels, to ensure that any injuries or losses caused by
train derailment would be minimised. The project implemented a central concrete
Page 19 of 92

up-stand in the four-foot area, which is stopped when equipment is required in the
four foot area. [RD6]
There is considerable work generally being led by the European Commission on
derailments. The media search has identified some of these initiatives below:
TRAINSAFE
European Commission is supporting various activities in response to the following
issues:
• the move towards separation of infrastructure management activities from those
of train operation, and the opening up of the network to new entrants;
• the increase target increases in capacity whilst simultaneously delivering safety
improvements;
• interoperability, intermodality and the harmonisation of standards.
The TRAINSAFE thematic network provides a mechanism for addressing the passive
safety aspects of the above issues. Their review into DCP concluded that in high risk
areas, (i.e. switch and crossing layouts, on bridges and during construction work),
check rails can be installed. More importantly, the review identified the adverse
effects of using long sections of checkrail is that if the rail vehicle did become
derailed, the presence of the checkrail can impair deceleration compared to running
on ballast, [RD7].
Standards
The European Commission commissioned a study of the obstacles to the completion
of the internal market for rail mass transit systems. The report identifies standards in
Member states that relate to derailment. The report demonstrates that across the
member states there is no uniform approach to derailment. The report concluded that
standards aimed at reducing derailments would be highly beneficial for the railway
industry and was given a medium priority [RD8].
Safety In Tunnels
Economic Commission for Europe Inland Transport Committee invited a group of
experts throughout Europe to consider safety in tunnels, their recommendation to
the Inland Transport Committee was that derailment containment measures should
be provided in all tunnels, [RD9].
In Australia the approach is to use guard rails on bridges and, more significantly,
under many bridges and other structures to prevent a derailed vehicle straying too
far and into the bridge columns/supports. Other preventative measures used are hot
axle box detectors used trackside in order to prevent bearing failures that can lead to
derailment, and devices such as catch points and 'derailment devices' blocks to stop
runaways from entering mainlines.
The information from the United States of America brought a different view
towards DCP. Amtrak has not implemented DCP and considers there to be
adverse affects with its implementation and its associated maintenance [RD14].
Page 20 of 92

In terms of the information available from the academic world, the Federal Railroad
Administration has undertaken various studies. These include “Shared Right-of-Way
Safety Issues”[RD10] which look at protecting operating envelopes through the
means of ditches and barriers etc. Of more significance is the “Intrusion Barrier
Design Study” [RD11], which modelled intrusion barrier methods of DCP. Their
conclusion was that with an intrusion barrier on both sides of the running lines, the
consequences of the derailment were more severe due to trains getting wedged
between the two barriers. In instances where the intrusion barrier was on only one
side of the running rails, the lower speed trains stayed in contact with the barrier
longer and came to rest in a more serve zig-zag, which could result in more severe
consequences.
Within the United Kingdom there have been several accidents involving derailments.
As a result the Rail Safety & Standards Board has initiated several research
assignments. (Although RSSB is not an academic organisation, it does sponsor
research topics undertaken by academic and research organisations etc).
These papers, Rail Safety & Standards Board Report “Engineering Overhead line
structure design to cater for collision” [RD12] and Rail Safety & Standards Board
Report “Engineering Derailment mitigation – categorisation of past derailments”
[RD13], conclude that:
• When there is an impact with overhead line side structure, there may be a
significant amount of damage to the train. The severity is dependent on
which part of the train absorbs the impact, however line side structures can
have a beneficial containment effect.
• Statistics show that most derailments do not affect passenger services and just
under half of the derailments occurred when the train was travelling at less
than 16km/h.
2.1.3 Part A Summary

DCP is a sensitive topic amongst rail organisations and there is concern that
information provided could expose them to commercial issues.
The questionnaires and interview responses indicate that there is a general (although
not universal) acknowledgment that DCP is a positive approach to minimising.
Although there is:

• No standard used in applying DCP and consequently
• No definite opinion on the effectiveness of DCP
• No DCP is 100% effective.
In the majority of cases, it was provided because it was considered to be a ‘good

idea’ e.g. Taiwan, or due to an earlier accident. The exception was CRTL where
their approach was based on a quantified risk assessment. Due to the lack of data
Page 21 of 92

concerning derailments this risk assessment was used assumptions and ranges of
values and only considered ‘common’ derailments. The risk analysis concluded
that check rails (DCP Type 1) and guard rails (DCP Type 2) were not effective at
high speed. A key lesson from CTRL is that the theoretical approach to DCP has
to be feasible and practicable and the design of the DCP Type 3 took into
considered the track layout to improve DCP effectiveness.
The widespread findings are that most authorities looked to use DCP Type 1 or DCP
Type 3 at high risk locations. These being where there are potential obstructions
(over bridges), escalating track geometry (curves) or structure transitions (into
tunnels). However, DCPs are not used as a matter of course throughout the system.
Most agreed that the discontinuity of the DCP could in itself contribute to the
derailment escalation but offered no solution or analysis to support this concern.
The safety record of some high-speed lines has been remarkable. The Japanese
Shinkansen has never experienced a passenger fatality or had a high-speed
derailment during 40 years of operation. This is a tribute to the design, inspection,
and maintenance of both track and vehicle and includes the inspection of the
Shinkansen track every day.
Other high-speed operations have not maintained as good a safety record as the
Shinkansen. Several derailments of the French TGV and the fatal German ICE
accident at Eschede Germany mar an otherwise enviable high-speed rail safety
record. It is important to acknowledge these accidents to better understand the
issues involved in high-speed rail safety.
From the interviews and media search, the greatest contributory factors associated
with escalation are:
• A derailed train hitting a second train where one or more are passenger trains
as demonstrated by several accidents in the UK (Great Heck derailment
caused by errant motor vehicle causing the initial derailment which escalated
when a second train was hit),
• The impact of a derailed train with lineside infrastructure (e.g. Hatfield, UK
and Eschede, Germany),
• Trains overturning.
The information collected through the media study supported the trends identified
in the interview and questionnaire responses.
Adverse effects of DCP include cost, maintenance and potential to escalate

derailments in certain circumstances, however these do not appear to have been
quantitively demonstrated. The media research provided some information on the
adverse effects of DCP as discussed in the Intrusion Barrier Design Study by the US
Federal Railroad Administration and from the approach taken by Amtrak in the US.
The media research identified several European Commission initiatives, which are
looking to the future and promoting a standardisation of the approach to controlling
derailment. However, they did highlight the current fragmented approach to
Page 22 of 92

derailment across Europe. It is of interest that interviewees did not discuss the
initiatives being led by the European Commission.
Although we have been able to find examples of train derailments, we have found
that high-speed derailments above 200 km/h are not common. More frequent are
derailments that occur when trains are operating at slow speed, often in storage
yards and staging areas. The track in these slow speed areas usually contains
specialised track work – crossovers, turnouts, switches etc. Since the yards are low
speed operation, derailments are usually minor in nature with minimal damage to
equipment or long-term implications.
Mainline derailments are less frequent than low speed yard derailments, however.
trains travelling at higher speeds have greater kinetic energy. The kinetic energy
increases with the square of the train speed so a small increase in train speed can
have significant effect on the severity of the accident or the damage to the rolling
stock, track and infrastructure.
The low number of high-speed derailments is possibly due to the close attention paid
to inspecting and maintaining track and equipment and the less challenging track
structure used on high speed lines. High-speed lines typically have fewer curves,
long straight-aways, and fewer turnouts or special track work.
A high level summary of the findings of the interviews and questionnaires is
contained in Appendix C.
2.2 Part B: Effectiveness & Part C Adverse Effects
Valid data on the effectiveness of DCPs is as rare as data on high-speed derailments.

There is little information and no real world statistics on DCP effectiveness. An
appropriate methodology to assess DCPs in this situation is for an expert panel
consisting of rolling stock, permanent way, derailment investigators and risk experts
from around the world to review the characteristics of individual derailments and
provide guidance on how effective DCP may be.
In determining the causal factors, the expert panel considered it prudent to extend
the scope of the analysis beyond the information determined in Part A media search
and interviews. The panel considered it necessary to not only examine the forensics
of general concept of derailments, but also to review the generic relationship between
derailment casual and escalation behaviour. Finally, prior to the commencement of
the expert panel, a review of the DCP Type definition and summary of
implementation was conducted. This was designed to ensure all parties would be
thoroughly conversant with the framework of the assignment.
Page 23 of 92

2.2.1 Causes of Derailments

A comprehension of the different types of derailments helps in understanding how
factors may contribute to high-speed derailments, since the effectiveness of DCP is
linked to the causal event of the derailment.
When design, inspection, and maintenance activities fail to fully address a defect or
hazard, derailments and other accidents can occur. Specifically, there are many
conditions of track and vehicles that can affect the safety critical wheel rail interface.
Derailments can result from failures of the track, track structure, special track work,
bridges, or other static structures. Derailments can also result from failures of
mechanical components related to the vehicle wheels and suspension, or from
human error or operations related problems. The causes of derailments are described
in more detail in Appendix D.
2.2.2 Relationship of Derailment Cause and Escalation

Understanding Derailment Escalation
Some types of derailments may escalate quickly (e.g. Potters Bar, UK accident).
These most usually result from derailments involving the following:
• Track Buckle
• Defective Switch
• Washout
• Overspeed
• Collision
• Elevated Track or Track on an embankment
• Structures i.e. Over Head Wire
• Geometry of curved track
• Location of curves
• Signalling/turnouts/conflicting movements
Other types of derailments may not escalate immediately (e.g. Eschede, Germany
accident). These may involve the following:
• Broken wheel
• Wheel lift
• Wheel climb
• Wheel tread defects
• Bogie defects
The dynamics of each derailment is different. Some derailments immediately

become major derailments at the point of initial derailment. The derailments that
escalate immediately often involve a collision or the catastrophic failure of critical
vehicle component or catastrophic failure and damage to the track structure. In these
cases the location of the initial point of derailment and the location of the general
derailment are in close proximity. The over speed derailment that occurred at
Waterfall Australia is an accident that escalated at the initial point of derailment.
Page 24 of 92

Other derailments may not escalate until the train has travelled kilometres from the
initial point of derailment. These types of derailments involve problems with wheels,
track, or bogies but are not so catastrophic as to result in an immediate pile up of
equipment. In these types of derailments, the initial point of derailment and the
location of the general derailment are far removed. The general derailment may be
triggered by track curves, elevated tracks with steep embankments, stationary object
close to the right of way, grade crossings, or special track work. Special track work
in the form of turnouts is especially efficient at locating a derailed wheel and causing
and contributing to derailment escalation.
Several of the general derailments took place at turnouts and accident records
contain numerous examples of derailed trains that have travelled for km before
escalating into a general derailment. For example, the Eschede ICE train travelled 5.6
km between the initial derailment and the point of escalation.
Some derailments never escalate. The derailment is discovered and the train is
brought to a stop with minimal damage. Once the train stops, the train is inspected
and the derailed wheel(s) is discovered. These incidents are not widely publicised so
we do not hear about many of the non-escalating accidents as the damage and
disruption is minimal. The Laval, France TGV accident is one such accident where
the train was brought to a stop with no escalation and no injuries to passengers.
2.2.3 Review of Derailment Containment Provisions
The types of DCPs considered within this assignment are:
• DCP 1
The first type of DCP mentioned consists of check rails installed between the
running rails within the track gauge. The check rails can help keep the rolling
stock in line with the roadbed. This is especially important on elevated
structures where keeping the carriage in line with the guideway is critical to
prevent the carriage from overturning or falling down an embankment.
Current practice is to apply check rails only in areas where there is a risk of
the carriage falling or overturning. They are not installed continuously
because the check rail can interfere with special track work such as turnouts.
The curve rail of the turnout occupies the same space as that needed by the
check rail.
Note: An alternative type of DCP I being proposed by Infraspeed for the

HSL-Zuid project is a tapered concrete plinth situated between the running
rails which restricts the lateral movement of the wheelset in the situation
where the wheels set has lost contact with the running rail. The configuration
of this DCP Type 1 is detailed in Appendix E.
Page 25 of 92

• DCP 2
The second type of DCP consists of guide rails installed outside of the
running rail. Outside guide rails can also help maintain a carriage upright
and in line on the right of way. Outside guide rails are not as common as
inside check rails and are applied in areas where inside check rails cannot be
used – such as in the area of special track work. However, outside guide rails
will still have application issues with turnouts and cannot be applied
continually.
• DCP 3
The third type of DCP is the use of a structure that contacts the sides of the
bogies in the event of a derailment and holds the bogies and car in alignment
with the roadbed and right of way. The third type of DCP can be effective if
it is designed to work with the expected forces generated by a train travelling
at 200 to 300 km/h. A structure this strong can exert significant forces on the
bogie and car body and may contribute to the severity of the accident.
There is no doubt that a strong enough structure could be designed to keep a

derailment from escalating. The question is how much would such a
structure cost and would it be cost effective to apply to the entire system.
Other then short barriers at the entrances to tunnels and bridges, we know of
no application of DCP Type 3 barriers exclusively designed to restrain the
movement of a derailed train. Research and testing in this area would be
required to confirm that the structure selected was strong enough for the
application.
2.2.4 Expert Panel Assessment

The expert panel was held on the in Amsterdam and was attended by:
Gerard Nakken DCP Project Manager, HSL-Zuid

Rob Houben Hoofd Integrale Safety Case + VG ontwerp, HSL-Zuid
Marten Brascamp Safety Manager, HSL-Zuid
Rene van der Vooren Manager Technical Compliance Department, HSL-Zuid
Harry Killaars Fire and Safety Department, Dutch Ministry
Robert Bos Manager Contracts, Infraspeed
Hans Risseeuw Manager of Derailment Risk Management and Mitigation
Safety Case, Infraspeed
Andy Espley Project Manager, Booz Allen Hamilton
Nicholas Bahr Safety & Risk Technical Expert, Booz Allen Hamilton
Fred Mau Track and Derailment Technical Expert, Booz Allen
Hamilton
John Boss HSL-Zuid / Booz Allen Hamilton.
The expert panel was presented with derailment accidents for which they were to
assess the effectiveness of the three different types of DCP if it had been installed at
the time of the accident.
Page 26 of 92

The expert panel undertook a review of the selected accidents in Table 7 against the
effectiveness of DCP in line with the assignment methodology. Details of these
accidents have been included in Appendix F. Appendix G contains further reading
on derailment accidents.
The results of the qualitative assessment of the effectiveness of DCP was indicated by
"Harvey balls" which provide a quick visual indicator as to how well an item being
evaluated meets the evaluation criteria.
For the expert panel evaluation in Table 10, the following nomenclature was used:
DCP Effectiveness Harvey Ball
Stops/prevents escalation
Significantly stops escalation
Somewhat affects escalation
Marginal/minor effect on escalation
No effect on escalation
Negative effect on escalation
The results of the expert panel methodology Steps B1 and B3 - B10 are detailed in
Tables 7- 13.
Accident Justification
1. Eschede, • Severity
Germany • First accident and the only accident in the history of high-speed
rail that resulted in a passenger fatality
• Vehicle failure cause event
• Combination of several causal events
• Similarity with HSL-Zuid infrastructure for example the Flyover
Den Hoek
2. Waterfall, • Operations causal event
Australia • Expert panel familiarity
3. Hatfield UK • Infrastructure causal event
• Relative high speed
4. Mobile, • Human factors causal event and involved a bridge crossing
Alabama, USA water which is relevant to the HSL-Zuid geography i.e.
Holland’s Diep bridge.
Table 7: Methodology Step B1, Accidents Identified For Expert Panel
Investigation
Page 27 of 92

Expert Panel Methodology Step

B3. Identified Generic Causal Events
The causal factors are common reasons for derailments that have occurred in the
past on all types of trains - not just high-speed trains. A review of these high-speed
derailments provides little insight into preventing other high-speed derailments
beyond the concept of design, inspection and maintenance discussed previously.
1. Track and Track Components including:
• Track buckle/Twist of Continuous Welded Rail (CWR)
• Broken rail
• Collapsed rail bed (i.e. WWI event France)
• Soft road bed (inadequate drainage, poor track geometry)
• Inadequate lateral stability (ties/fasteners)
• Track tolerances (rail wear)
• Rail geometry (cross level, gauge spread, etc.)
• Defective switch (numerous failure initiating events can occur)
2. Vehicle (locomotive and coach) Wheel and Bogie Components including:
• Broken wheels
• Worn / broken suspension
• Wheel lift
• Wheel climb
• Wheel tread defects (sharp flange, hollow tread, etc.)
• Wheel profile condition
3. Special Track Work related including:
• Switch or turnout defects
4. Bridges, Structures, Signals, and other Infrastructure including:
• Errant vehicles on level crossings
• Structural failure of lineside equipment (lamp post, signal falling onto the
running rails)
5. Operational Issues (Operating Rules and Methods) including:
• Over speed roll over
• Collision
- Train to train
- Train to highway vehicle
• Train handling
6. Human Factors
• Human error
• Training
7. Other Causal Factors
• Washout
• Maintenance tools, branches, landslip
• Flooding
Table 8: Methodology Step B3, Expert Panel Findings
Page 28 of 92


Accident
B4. Did escalation occur and if so when?
1. Escalation did occur at the points. It seems probable that a derailed wheel was running on the outside of the right-hand rail and
Eschede, on reaching the turnout it was diverted towards the slow line and onto a collision course with the bridge supports. There was
Germany lateral movement of the wheels set which escalated to the eventual impact with the bridge support. Although facing points are
essential to the smooth flow of traffic, there will be a tendency to guide a derailed wheel, running outside of the rail, in the
direction of the turnout. The guard rail (function of point design) contributed to the escalation of the derailment. The number
and location of points is an important consideration in the design stage of a project in order to minimise the consequences of
derailments.
2. The escalation occurred quickly after the derailment due to the rail break and the curvature of the track.
Hatfield,
United
Kingdom
3. The escalation was immediate.
Waterfall,
Australia
4. Mobile, The escalation was immediate.

Alabama,
USA
Table 9 Methodology Step B4, Expert Panel Findings
Page 29 of 92

B5. Was DCP in place at the derailment site and estimates the effectiveness of each type of DCPs
Yes, what type and what No
was there effectiveness? How would application of DCP Type 1, DCP Type 2, DCP Type 3 have affected the accident including
speeds at 200-300 Km/h.
Accident 1. Eschede, Accident Stage DCP Type 1 DCP Type 2 DCP Type 3
Germany Initial derailment 0 0 0
Continued travel after initial derailment 1 4
1
Escalation derailment 0 0 2or0
Accident 2. Accident Stage DCP Type 1 DCP Type 2 DCP Type 3
Hatfield, United Initial derailment 0 0 0
Kingdom
Continued travel after initial derailment 3 3 3
Escalation derailment 4 3or0 1 2
Accident 3. Accident Stage DCP Type 1 DCP Type 2 DCP Type 3

Waterfall, Australia Initial derailment 0 0 0
1
Escalation derailment 4 3or0 2
Accident 4. Mobile, Yes DCP Type 1 but Accident Stage DCP Type 1 DCP Type 2 DCP Type 3
Alabama, USA was not effective 0 Initial derailment 0 0 0
Escalation derailment 0 0 0
Table 10 Methodology Step B5, Expert Panel Findings
1 Note: Effectiveness of the DCP is directly related to which side of the track the DCP is fitted relative to which side of the track the wheel set derails.
Page 30 of 92


Accident B6. Expert Panel identifies steps that could have been taken to prevent escalation of the derailment
1. Eschede, The importance of detecting a derailment (broken wheel) before escalation takes place significantly affects the escalation consequences. The
Germany consequences of Eschede could have been minimised if the derailment was detected immediately. In the case of this accident the
identification of derailment was detected by the cutting of the LZB cable and the vehicle floor penetration. The identification of the
derailment was delayed because the power car was already located on the next LZB loop when the previous LZB loop was cut and the lack
of action of the Eschede conductor who did not stop the train after receiving reports that debris had penetrated the floor of the first
passenger coach.
Infrastructure
a) The location of a turnout immediately before a bridge is a significant escalation factor that should be avoided. They should not be
placed near highway bridges, train bridges, elevated track, tunnel entrances, or other locations that are considered excessively
hazardous if a derailment were to take place. A safety based hazard analysis should have been used to identify these high-risk
areas.
b) The lack of crash walls or other protection for the bridge support should have been identified as an issue. The use of crash walls in
the vicinity of the bridge may have been a method of reducing the hazard of having the turnouts in such close proximity to the
bridge. Again, a good hazard analysis would have identified the hazard.
c) Installation of Dragging Equipment Detectors (DED) which automatically stop the train. These detectors could be placed
specifically prior to turnouts at sufficient distances to allow the train to stop before the turn out. For example, DEDs are commonly
used by Amtrak in the US and by New South Wales in Australia. These devices are of relatively low cost and there is no known
evidence of adverse reliability.
Vehicles
d) The design of the wheel may not have been appropriate for a high-speed application. Components designed for a safety critical
function such as the wheel rail interface must be able to be adequately inspected for defects. The resilient material interfered with
the ultrasonic testing and may have masked a defect. In any case, a thorough inspection of the wheel would have indicated that the
wheel tread was below the condemning limit and should be changed out. Again, inspection and maintenance practices are a
primary method of controlling hazards of safety critical systems.
e) On board DED
f) Detectors mounted on the bogies to detect excessive movement in the horizontal plane which would, in approximately 90% of
instances, identify a derailment. (These are currently being installed on the Thalys trains which will be operated on the HSL-Zuid).

Page 31 of 92


1. Operations
Eschede, g) It is true that the brake application may have escalated the accident before reaching Eschede but the results would have certainly
Germany been more forgiving than running into a concrete bridge structure at 180 km/h.
(cont) h) If the installation location on the infrastructure of the DED is restricted then it may have to be moved closer to the risk area. In such
instances, a Permanent Speed restriction may be necessary to allow the train to be brought to a halt before the turnout.
i) Loop 9 of the LZB system was ruptured. If this information could have been communicated to the controlling power car, then the
train could have been stopped before it reached the bridge at Eschede. The power car finally received the information and applied
emergency brakes after loop 11 was ruptured by the general derailment.
j) Operational rules that instruct the driver to slow down to 80kph if he receives a warning signal that the train is not operating in an
acceptable manner (This rule is being proposed on the HSL-Zuid line).
k) Several trains may traverse the cause of a derailment as in the case of the TGV derailment at Lille in France, where it was the 3rd train
to pass the site which derailed. Therefore train drivers and train crew should be trained in responding to abnormal instances and
have the means of communicating information to the relevant persons (i.e., signallers).
Human Factors
l) Driver and crew training to recognise and respond to abnormal events.
Others
m) Trip wire at free space envelope as used on the Washington Metro and similar to the installation used on high speed lines in France
as identified during the interviews in Part A of this assignment which detects the presence of automobiles etc.
Table 11 Continued: Methodology Step B6, Expert Panel Findings
Page 32 of 92

2. Infrastructure
Hatfield a) Installation of DCP Type 3 would have had a positive effect on the escalation of the derailment. In terms of DCP Type 3 the accident
United occurred along a curved section of route in which there were multiple running lines. The derailed train was operating on one of the
Kingdom. inner lines. For DCP Type 3 to have been effective, it would require installation along side all the lines on the curves section of route.
In order for this to take place, the original track scheme would need to have considered this requirement and made sufficient space
provision. To install DCP Type 3 on routes containing multiple lines may be difficult due to space limitations. Furthermore, as
identified in the media study, the effect of having DCP Type 3 on both sides of a line may have an adverse effect, with the derailed
train experiencing becoming wedged between the DCP barriers, while following coaches continue to impact into the jammed train.
The effectiveness of DCP Type 3 is difficult to quantify, however modelling work undertaken by Infraspeed has indicated that at instances
where there is greater than 450Kn of force on the bogie it will separate from the coach body.
The Greenhout tunnel will be installed with DCP Type 3 in the form of walkways which have been designed to act as DCP.
b) The cant deficiency on the track should be minimised as far as possible to reduce the track forces.
c) The design and location of line side structures should be such that they minimise the affect of escalation
d) Cross section of the track design to maintain the derailed train upright.
e) DCP Type 1 would have had some effect as it would have not been subject to effects of track forces and so would not have broken
when the rail break occurred. It then would have guided the derailed wheelset.
f) Detection of the broken rail
Vehicles
g) Detection devices attached to Bogies to detect derailment
h) The use of articulated bogies
Operations
i) Adequate inspection, testing and maintenance regimes
Human factors
j) No significant steps identified
Others
k) No significant steps identified
Page 33 of 92

3. In addition to steps identified in the previous accident, the following steps were identified:
Waterfall, • Detection of overspeed and the application of automatic braking as appropriate (ATP)
Australia • Design of deadmans handle to ensure that fatalities will not continue to operate deadmans handle.
4. None of the 3 different types of DCP would have prevented the escalation of the accident. Items identified in this case related to operation of
Mobile, the river barge transport and infrastructure elements on and around the bridge, including bridge impact detectors.
Alabama,
United
States
America.
Table 11 Continued: Methodology Step B6, Expert Panel Findings
Page 34 of 92

Accident B7. Expert Panel identifies steps that could have been taken to prevent the derailment
1. Eschede, Germany Preventing the wheel departing the rail.
Infrastructure
a) Installation of dynamic wheel impact load detectors, although this equipment is considered costly.
Vehicles
b) Maintenance and Inspection
c) Component Integrity (design, constraints)
d) Quality (control assurance, suppliers)
Operations
e) Detection
f) Communication
g) Failsafe
h) Train handling
Human Factors
i) No significant steps identified
Others
j) No significant steps identified
The panel noted that the DCP does not affect the cause of the broken wheel.
2. Hatfield, United It is very difficult to ensure there are no rail breaks. The use of suppliers with high standards of Quality Assurance and Quality
Kingdom control will minimise rail breaks. Regular maintenance and inspection of the running rails can identify some potential failures. It
was noted that ultrasonic testing would not identify all instances of potential failure.
On the Shinkansen railway in Japan in addition to the maintenance and inspection regime, the running rails are replaced at
regular frequencies.
3. Waterfall, Fencing to prevent obstacles on the track.
Australia Prevent overspeeding by the use of ATP.
4. Mobile, Alabama, No significant steps identified
USA

Page 35 of 92

Accident B8. Expert Panel identified derailment causal factors.

1. Eschede, a) Broken wheel
Germany
b) Warning signal not received from LZB system
c) The guard rail2 used in the configuration of the points may have contributed to the escalation.
d) Speed at turnout
e) Deflection of wheel set
f) Train struck bridge support
g) Bridge collapse (consequences)
2. Hatfield, a) Rail break (increased probability due to high cant deficiency) Note: on HSL-Zuid cant deficiency is 180mm.
United
b) Track geometry (curvature of the track)
Kingdom
c) Hits catenary
3. a) Over speed.
Waterfall,
b) Curvature of the track.
Australia
c) Hitting fallen rocks as a result of the initial derailment
d) Human Factors, the design of the deadmans handle was deficient and required a constantly maintained pressure to ensure the
brakes were not automatically applied.
e) The guard on the train did not recognise he was in danger.
f) The culture was that guards did not stop trains since this was the responsibility of the drivers. The divers also had a different
culture to maintainers.
4. Mobile, Bogies were derailed as a result of running rail bridge and running rail displacement.
Alabama,
USA
2 Guard Rails are generally used in as part of the basic design of points. These guard rails are designed to guide the wheelset, as opposed to the Guide rails described in DCP Type 1, which
are more robust and designed to prevent Derailment. Guard rails are found in two locations within points: a) guard rail (turnout ) A rail or other device laid parallel to the running rail
opposite a frog, b) guard rail (switch) A rail or other device laid parallel to the running rail ahead of a split switch
Page 36 of 92


B9. Expert Panel identified any additional risk that may be associated with each type of DCP
General
The transition along a line from no DCP to DCP presents an additional opportunity to escalate the derailed train. Furthermore there is also additional risk to
maintenance workers due to additional equipment on track which may require more workers on the track and the possibility of a more complicated
working environment.
DCP Type 1:
Debris between the check rail and the running rail could cause wheel climb. Debris can be from a variety of sources i.e. maintenance tools left behind after
works, fragmentation of a running rail, ballast or foreign object on the track.
Deutsche Bahn have indicated that they limit the use of Type I DCPs because of the difference in passenger carriage bogie and axle designs. For example,
the axle design for passenger carriages include a brake disc that can contact and ride along the guide rail causing damage to both the guide rail and the
brake disc that could escalate a simple derailment. Another consideration is that the under car arrangement of brake rigging and traction equipment may
also interfere with the guide rails and cause a more dangerous situation after a derailment.
Guide rails were originally designed for freight cars, which are not equipped with the axle or bogie mounted equipment mentioned above. The concern at
DB was sufficient to limit the use of guide rails to only the most significant areas.
DCP Type 2:
Again debris between the guard rail and the running rail may result in DCP being ineffective. Under some circumstances, outside guardrails may
contribute to the overturning of the carriage.
DCP Type 3:
The height of DCP Type 3 in relation to double decker trains could significantly cause an increases in the escalation derailment risk, since the DCP Type 3
barrier could cut through the passenger areas on a double decker train. Therefore the effectiveness of DCP is influenced by the fleets of trains which may be
expected to operate on HSL Zuid?
The implementation on DCP Type 3 near switches and points can act as a physical barrier to the maintenance activities which are so critical to preventing
the causal factors.
Page 37 of 92

2.2.5 Expert Panel Commentary & Methodology Step 10
The expert panel provided a solid judgement of DCP effectiveness based on the
combination of the experience of individuals and the latest state of the art knowledge.
The panel reviewed the effectiveness of DCP on four accidents and in accordance with
the Methodology defined in Section 1 the results of which are detailed in tables 9-13.
It became apparent that there are complex variable factors to be considered when trying
to evaluate the effectiveness of DCP, and each application of DCP is specific to the
operational environment. Therefore the effectiveness of DCP cannot be quantified,
[Key Finding 1].
Step Zero
The expert panel identified that the most effective approach to DCP is actually at “Step
Zero” in the design and implementation phase and subsequent maintenance stage.
The early detection of abnormal events can reduce escalation, [Key Finding 2].
During system design there needs to be a realistic expectation regarding the

maintainability of the equipment. The designer should be careful about making design
assumptions which can have long-term consequences.
When a change to the system is proposed, the custodian of the system should evaluate
the effect of the change across the whole system, [Key Finding 3].
During the life of the system, specifically the infrastructure, there may be an increase in
track forces due to an increase in:
• Density of rolling stock,

• Heavier rolling stock (longer trains, more auxiliary systems),
• Faster rolling stock (accelerate faster and therefore have a higher average speed
which means there is an increase in track forces around switches and crossings in
close proximity to stations).
This increase in track forces may result in an acceleration in the degradation of the track
quality and an increase in serious faults.
For example, a change in the area of operations such as increased train paths in the
timetable can result in accelerated rolling contact fatigue (RCF). Therefore the
maintenance and inspection aspects of the system would require evaluation. This was
discussed as part of the Hatfield accident where high cant deficient track and track
geometry were important contributors to the risk of derailment and its escalation. High
cant deficient tracks with head hardened rails should be subject to regular inspection
and maintenance (rail grinding) to identify any potential RCF which could result in
gauge corner cracking and ultimately a rail break, the cause of the Hatfield accident.
Page 38 of 92

For example, in the case of the Eschede accident, to minimise the escalation of
derailment, the bridge structure should have been constructed to withstand an impact
from the train. However, this may have been the part of the original design
requirements, but changes to the system may have taken place without evaluating the
effect of the proposed change, i.e. linespeed may have increased.
Mitigating Escalation
Operations
In the case of Eschede, the passengers heard noise and felt vibration for approximately
two minutes prior to the impact with the bridge structure. During this time, actions
could have been taken to reduce the consequence of the derailment. A further example
of this is the Central Line underground train in the UK. It did not respond to the
abnormal noise of the motor shearing away from securing bolts, hitting the track and
bouncing along the track hitting the underside of the train and subsequently derailing
the train as it entered Chancery Lane station.
It was suggested that detection of a derailment needs to be identified as quickly as

possible and this facility should be incorporated into the Railway System,
[Key Finding 4].
This can be achieved either through a technical or operational approach, i.e. training
drivers and train crew to recognise the derailment.
The risk of derailing or minimising escalation can be achieved by stopping rail traffic as
quickly as possible after an appropriate safety critical event, [Key Finding 5].
The Great Heck accident (as detailed in Appendix G) would have been less serve if both
a passenger and freight train had not interacted so disastrously when the automobile
landed on the line.
As the mitigation of derailment is placed on the maintenance and inspection of the

vehicle and the infrastructure, an increase in the maintenance and inspection periods is
required, irrespective of the pressure to maximise operations.
Infrastructure
As identified in the interviews, high-speed lines in France use ballast to slow down a
derailed train and provide sufficient free space on bridges to allow this to occur.
However, on the HSL-Zuid line there is a combination of ballasted track and Rheda slab
from track. In the instance of a derailment on Rheda track there is no ballast to act as
resistance to slow the train down. Furthermore the structure of the Rheda track, such as
the concrete supports to the rails, could accelerate the lateral movement of a wheelset
and subsequently escalate the derailment.
Page 39 of 92

In addition, the Rheda track is being backfilled with ballast to a level such that the
distance between the rail and the ballast is 240mm. This may increase the probability of
derailment escalation since the wheel set could “fall off” the Rheda track bed slab.
provision, though it should be designed so as to ensure the outside derailed wheel
remains securely upon the track slab. In Part A of the assignment, DB reported that the
interference between the brake disc and the Type 1 DCP was a main concern for limiting
the use of DCPs. Therefore the expert panel was concerned with the interface design on
HSL-Zuid between the vehicle and the DCP, where the axle mounted brake disc may
ride on top of the concrete plinth DCP and cause potentially serious consequences,
[Key Finding 6].
The effectiveness for track systems with proposed derailment provisions verses
scenarios / causes has already been subject to an earlier expert panel. This report does
not seek to duplicate work, therefore further information is detailed in ‘ Expert Panel
Meeting on Effectiveness of different derailment provisions ’ [RD.14].
One of the mitigation actions to reduce escalation would be to add deflector plates, to
lineside structures to assist in reducing escalation dynamics during derailment. The
installation deflector plates could be fitted retrospectively to lineside infrastructure and
would not incur a significant cost. This approach would require further investigation
and an appropriate cost benefit analysis.
Management Systems
A deficiency in the competence of staff (experience, knowledge, training) may result in

the failure to identify the fault during inspection and maintenance. There may be
insufficient staff to rectify the fault in an acceptable timeframe, and there may be
insufficient competent staff to supervise the work undertaken, [Key Finding 7].
Japan’s remarkable safety record is due to strong in-depth inspection (100% every 24hrs)
and maintenance accompanied by reduced wartime of the system components. In order
to support this approach, asset management plans are a contributory factor to ensuring a
coordinated approach, [Key Finding 8].
DCP Effectiveness
There are complex variable factors to be considered when trying to evaluate the
effectiveness of DCP. Every accident is unique and even where it is possible to break
down the accident into phases (initial derailment, continued travel, escalation
derailment), as in the case of Eschede, it was difficult to establish consistent causal
factors which could be utilised in other derailment accident analysis. Therefore the
effectiveness of DCP cannot be quantified, [Key Finding 9].
Page 40 of 92

The expert panel agreed that the greater the lateral movement generally the greater
escalation of the derailment. In all the accidents it was considered that no form of DCP
could have prevented the derailment. The mechanism causing the derailment was
independent to the control mitigating DCP, [Key Finding 10].
DCP Types
DCP Type 1 is effective at mitigating escalation in certain scenarios, although it cannot
be used in points and crossings and, as shown by the approach by DB, there are adverse
effects to its implementation, [Key Finding 11].
During the expert panel, the meeting concluded that the effectiveness of DCP Type 2 is
less than DCP Type 1, since DCP Type 2 would have to be applied at the end of the
structure or the ends of the sleeper where the track structure has less strength, [Key
Finding 12]. This supports the trends observed in Part A of the assignment, where
worldwide DCP Type 1 and DCP Type 3 are more commonly used.
The panel discussed that at points and crossings DCP Type 3 could be installed which
would reduce the escalation of the derailments. The vertical profile of this would be
ramped reaching a maximum height at switches and crossings, [Key Finding 13].
When examining the implementation of DCP Type 3, it is not always clear if the
structure is initially designed as DCP or for another function, such as a walkway. If
designed as DCP, then the structure should have performance standards or
specifications that relate the design to controlling train derailments. There is a
possibility that because a structure is located either side of the right-of-way, it will be
assumed to be an effective DCP. This is not necessarily true especially at high speeds,
Key Finding 14].
DCP Application
Through switches and points, the ability to implement effective DCP is minimal.
The effectiveness of DCP can be enhanced through mixing types. The expert panel
identified the mix of Type 3 DCP at switches in conjunction with Type 1 to be potentially
effective for a derailment in the trailing direction with the bogie off set away from the
direction of the approaching turnout, [Key Finding 15].
Implement DCP Type 1 on open track at high risk areas (i.e. urban) and elevated
sections subject to cost benefit analysis and DCP type 3 in tunnels and around turnouts,
[Key Finding 16].
Since each potential application of DCP is unique, there is a danger that the
organisations or projects may adopt what is seen as best practice. This approach should
be used with caution since it is critical each organisation should understand the failure
modes of their system before implementing DCP, [Key Finding 17].
Page 41 of 92

Adverse effects of DCP
The installation of a DCP Type 3 barrier in close proximity to switches and crossings
may act as a physical barrier to inspection and maintenance regimes which are designed
to mitigate the causal factors of derailment. Therefore although DCP Type 3 may
mitigate escalation, it may be contributing to the causal event in the first instance,
Key Finding 18].
The height of DCP Type 3 in relation to double decker trains could significantly cause an
increases in the escalation derailment risk, [Key Finding 19].
There may be adverse effects of installing DCP Type 1 where foreign objects such as
previously sucked up ballast or fragmented rail as a result of an infrastructure fault
could act as a ramp to derail the wheel set, [Key Finding 20]. This ramp principle is very
effective and is used in certain locations i.e. in sidings to protect main lines from
runaway rolling stock.
Under some circumstances, DCP Type 2 may contribute to the overturning of the
carriage. Again debris between the guard rail and the running rail may result in DCP
Type 2 being ineffective.
Additional Analysis
At the conclusion of the expert panel meeting a high level qualitative assessment (Table
15.) of the effectiveness of DCP Types 1,2 and 3 against the causal factors identified in
Table 8. was undertaken using the following nomenclature:
DCP Effectiveness Harvey Ball
Stops/prevents escalation
Significantly stops escalation
Somewhat affects escalation
Marginal/minor effect on escalation
No effect on escalation
Negative effect on escalation
However, whilst undertaking the qualitative analysis, it quickly became apparent that
balanced and comparative qualitative assessment was not achievable since the causal
factors were generic in some instances (i.e. human factors) and, as identified in the
expert panel meeting, there are complex relationships between factors in derailments
and many of these are variable. Therefore the results of the qualitative assessment in
Table 16 can only be regarded as an initial high level judgement at a generic level.
Page 42 of 92

They are not supported by robust justifications and therefore should be viewed as an
interpretation rather than based on robust arguments.
From Table 16 is can be seen that at a generic level DCP Type 3 is the most effective form
of DCP on high speed lines a range of high level causal events. This is supported by
Part A of the assignment, where organisations which operate high speed lines such as
CTRL and Skinkansen implement DCP Type 3.
Page 43 of 92

Derailment Causal Event DCP DCP DCP

Type 1 Type 2 Type 3
Track and Track Components including:
Track Buckle of Continuous Welded Rail (CWR)
Broken rail
Collapsed rail bed
Soft road bed (inadequate drainage) / poor track geometry
Inadequate lateral stability (ties/fasteners)
Track tolerances (rail wear)
Rail Geometry (cross level, gauge, etc.)
Defective switch (failure dependent)
Vehicle Wheel and Bogie Components including:

Worn / broken suspension
Wheel lift
Wheel tread defects (sharp flange, hollow tread, etc.)
Special Track Work related including:

Switch or turnout defects
Bridges, Structures, Signals, and other Infrastructure

Errant vehicles at level crossings
Structural failure of line side equipment (Lamp post, signal)

fall onto track
Train handling
Human Factors
Human error and training (too generic to assess) Not possible to assess
Other Causal Factors
Washout
Maintenance tools, branches, landslip
Flooding
Table 15 Continued: Qualitative Assessment of DCP’s
Page 44 of 92

Section
3 Summary
In managing the delivery of the HSL-Zuid Transportation System, the HSL-Zuid
organisation wishes to act as the informed client with regards to current practice and
effectiveness of DCP on high speed lines. To achieve this HSL-Zuid organisation has, in
part, commissioned this assignment which consists of three parts, which are as follows:
3.1 Overview of Current Practice
DCP is a sensitive topic amongst rail organisations and there is concern that information
provided could expose them to commercial issues.
The questionnaires and interview responses indicate that there is a general (although not
universal) acknowledgment that DCP is a positive approach to minimising derailments.
In spite of this there is:

• No standard used in applying DCP and consequently
• No definite opinion on the effectiveness of DCP,
• No DCP is 100% effective,
• Widespread agreement that DCPs are not used as a matter of course throughout
the system.
The findings indicate that DCP has been provided because of it was considered to be a
‘good idea’ and the approach to installation of DCP has been based on other railway
projects. The assignment identified that most projects and organisations looked to use
DCP Type 1 or DCP Type 3 at high risk locations.
The media research identified several European Commission initiatives, which are
looking to the future and promoting a standardisation of the approach to controlling
derailment.
3.2 Estimation of Effectiveness of DCP

During the assignment it became apparent that organisations and projects were
unwilling or unable to discuss the sensitive issue of derailments and DCP due to a
variety of reasons. In addition, the numbers of derailments above 200 km/h are small.
As a result the lack of data made, the original quantified approach for the determination
of DCP effectiveness un-achievable.
Page 45 of 92

Therefore the approach to the assignment was modified in agreement with the client to a
qualitative approach. The revised methodology and approach relied heavily on
interpretation by an expert panel
There is a general acknowledgment that DCP is a positive approach to minimising

derailment escalation under certain circumstances. However, there is insufficient
knowledge to make a quantified decision on the effectiveness of DCPs to confidently
predict how a train will behave in a derailment at speeds above 200 km/h. There are
complex variable factors to be considered when trying to evaluate the effectiveness of
DCP. Every accident is unique, and even where it is possible to break down the accident
into phases (initial derailment, continued travel, escalation derailment), is difficult to
establish consistent causal factors which could be utilised in other derailment accident
analysis. Therefore the effectiveness of DCP cannot be quantified.
The effectiveness of DCPs is strongly dependant on the cause and configuration of the
initial derailment. Effectiveness is very limited when derailments escalate immediately.
When escalation is not immediate, DCPs may help keep some trains upright and in line
but a derailment escalation may still occur if the derailment is not detected quickly and
appropriate emergency systems come into effect, e.g. emergency stop.
At a generic level DCP Type 3 is the most effective form of DCP on high speed lines
based on high level causal events. This is supported by organisations which operate
high speed lines such as CTRL and Skinkansen who implement DCP Type 3.
The effectiveness of DCP can be enhanced through mixing types. The expert panel
identified the mix of Type 3 DCP at switches in conjunction with Type 1 to be potentially
effective for a derailment in the trailing direction with the bogie off set away from the
direction of the approaching turnout.
One of the key outputs from the expert panel was the identification of “Step Zero”,
where the prevention of the causal events was the area on which to focus attention.
Many of the causal factors associated with derailment could be mitigated to some degree
by detection or prevention of the faults on the vehicle or the infrastructure. The longer
these faults remain undetected the greater the potential risk of derailment. An increase
in the frequency of inspection would reduce the time that track, or train faults exist, and
therefore reduce some of the causal factors in derailments by proactively controlling
derailment risk.
During system design there needs to be a realistic expectation regarding the

maintainability of the equipment. The designer should be careful about making design
assumptions which can have long term consequences. When a change to the system is
proposed, the custodian of the system should evaluate the effect of the change across the
whole system environment.
Page 46 of 92

A logical approach to addressing the derailment hazard would be to:
1. Prevent as many derailments as possible through superior design, inspection and

maintenance practices.
2. Conduct regular risk assessments to detect and correct hazardous conditions.
3. Reduce the number of turnouts and special track work along the right of way.
4. Use technology to detect derailments.
5. Only install the appropriate type of DCP or mixture of DCP’s in the relevant
areas.
3.3 Adverse Effects of DCP
The overall continuity of the DCP has been identified by most projects and organisations
as a factor in the effectiveness of the DCP provision. Discontinuity of DCP can be the
cause of escalation and is considered to be a significant adverse effect of DCP. Other
adverse effects relate mainly to the additional capital costs and maintenance elements
involved (Amtrak does not use DCP due to maintenance issues).
Specifically DB indicated that they limit the use of Type 1 DCP because of the difference
in passenger carriage bogie and axle designs. Another consideration is that the under
coach arrangement of brake rigging and traction equipment may also interfere with the
guide rails and cause a more dangerous situation after a derailment.
Under some circumstances, DCP Type 2 may contribute to the overturning of the
carriage.
The installation of a DCP Type 3 barrier in close proximity to switches and crossings
may act as a physical barrier to inspection and maintenance regimes which are designed
to mitigate the causal factors of derailment. The expert panel identified a potential
increase in escalation with DCP Type 3, since the barrier could cut through the
passenger areas on a double-decker train.
3.4 HSL-Zuid
On the HSL-Zuid line of route there are certain locations i.e. Hollandsch Diep bridge,
Flyover Van Hoek and Viaduct Bleiswijk, which could be considered as high risk, where
as a result of a derailment the consequences could be severe.
The expert panel discussed the derailment escalation in relationship to the presence of
structural discontinuities in the near vicinity of the rails and to the track bed
constructions on the HSL-Zuid line, where there is a combination of ballasted track and
Rheda slab form track.
In the instance of a derailment on Rheda slab form track there is no ballast to act as
resistance to slow the train down. Furthermore the structure of the Rheda slab form
Page 47 of 92

track, such as the concrete boots for the base plates could accelerate the escalation of the
derailment. In addition, the Rheda slab form track is being backfilled with ballast to a
level such that the distance between the rail and the ballast is 240mm. This may increase
the probability of derailment escalation since the wheel set could “fall off” the Rheda
track bed slab.
provision, though it should be designed so as to ensure the outside derailed wheel
remains securely upon the track slab. In Part A of the assignment, DB reported that the
interference between the brake disc and the Type 1 DCP was a main concern for limiting
the use of DCPs. Therefore the expert panel was concerned with the interface design on
HSL-Zuid between the vehicle and the DCP, since the brake disc may ride on top of the
concrete plinth DCP and cause potentially serious consequences.
3.5 Recommendations
1. Review the applicability of design inputs that have taken the “best practice” on
DCP from other railways. Differences in operations, geometry, track speed and
track bed profile could invalidate the original reasoning behind the application
and continue to review during changes in the operation of the HSL-Zuid or the
surrounding environment.
2. Review the rolling stock design input into the design of the HSL-Zuid DCP to
ensure the risks associated with brake discs and other train characteristics on the
Thalys trains were addressed throughout the design process.
3. Ensure the risk analysis for the determination of the high risk areas has
addressed the characteristics of the Rheda track with respect to derailment
escalation in the absence of any DCP.
4. Establish a management process to prioritise the rectification of faults which

could lead to derailments.
5. Ensure the management system accurately captures the reporting of derailment

incidents on HSL-Zuid. This will improve the understanding of derailment
incidents and their associated causes and will result in the appropriate
modification of derailment containment provision strategies, assisting in
mitigating new issues as they develop. Note that the project should monitor
derailment mechanism from around the world to ensure that any potential issues
are managed.
6. Ensure there are appropriate assurance regimes in place in order to ensure that
installation, inspection and maintenance are undertaken as prescribed in the
relevant documentation over time (rolling stock and infrastructure).
Page 48 of 92

7. Review the safety management system to ensure it is effective, as far as

reasonably practicable, in eliminating derailments since this is the most effective
solution to reduce escalation. The safety management system should
demonstrate the following steps:
a) Identify and understand derailment causal factors

b) Determine individual risk of each causal factor
c) Rank and prioritise:
• Design
• Operations
• Maintenance
d) Determine controls for each risk
e) Re-evaluate risk
f) Develop robust verifiable monitoring systems
g) Determine lagging and leading indicators
h) Adjust strategy as necessary
8. Assess the derailment resistance of vehicles operating on HSL-Zuid and the track
forces they generate on a periodic basis.
9. Review the possibilities for the application of a mix of different DCP types to
minimise the escalation of derailments.
10. Review the possibilities for other (non DCP) infrastructure equipment to assist in
reducing escalation of derailment, or derailments occurring in the first place, e.g.
dragging equipment detectors, Hot box detectors, wheel condition monitors,
bridge movement detectors.
11. Review the rail change out policy against the risk of undetected rail flaws.
12. Ensure derailment issues are taken into account in operations, training and
procedure definition. This requires specific attention to all steps along the
derailment event to ensure appropriate behaviour to limit escalation.
13. Develop a culture of active intervention upon recognition of “abnormal events “

which could escalate with serious consequences.
Page 49 of 92

4 References
Applicable Documents
The following documents identify dependencies, interfaces and/or other requirements

with which this document or activities described therein must comply:
AD1 HSL-Zuid Remit letter ref HAVL/517575.

AD2 HSL-Zuid Terms of Reference 16th April 2004.
AD3 Email from HSL-Zuid confirming revised Methodology and Approach.
Reference Documents
The following documents provide supporting information:
RD1. Original Interview Data

RD2. Independent Transport Safety & Reliability Regulator New South Wales in
Australia Rail Derailments Statistics
http://www.transportregulator.nsw.gov.au/stats/rail-derailments.html
RD3. Council For transport safety Netherlands, Baarn derailment.
http://www.rvtv.nl/english/data/985865498.pdf
RD4. Principles of Railway Derailment Management R-0014-00-HSL-J2.
RD5. UIC Leaflet 777-2 ‘Structures built over Railway lines- Construction
requirements in the track zone’.
RD6. Pandrol Case Study Heathrow Airport Rail link.
http://www.pandrol.com/cstudies/c05_bdy.htm
RD7. TRAINSAFE Thematic Network Passive Safety in the Railways: A State of the
Art Review December 2003.
http://www.arrc.ac.uk/docs/trainsafe_state_of_the_art_report.pdf
RD8. Obstacles to Internal Market in Rail Mass Transit.
http://europa.eu.int/comm/enterprise/rail_guided_transport/rail_mass_tra
nsit.pdf
RD9. Economic commission for Europe Trans/ac.9/91 December 2003
Recommendations of the Multidisciplinary group of experts on safety in
tunnels (rail).
http://www.unece.org/trans/doc/2003/ac9/TRANS-AC9-09e.pdf
RD10. Federal Railroad Administration, Safety of High Speed Guided Ground
Transportation Systems Shared Right-of-Way Safety Issues.
RD11. Federal Railroad Administration, Safety of High Speed Guided Ground
Transportation Systems, Intrusion barrier Design Study.
Page 50 of 92

RD12. Rail Safety & Standards Board Report “Engineering Overhead line structure
design to cater for collision.”
http://www.rssb.co.uk/pdf/reports/research/Overhead line structure
design to cater for collision (T177).pdf
RD13. Rail Safety & Standards Board Report “Engineering Derailment mitigation –
catergorisation of past derailments.
www.rssb.co.uk/pdf/reports/Research/Overhead%20line%20structure%20d
esign%20to%20cater%20for%20collision%20(T177).pdf
RD14. Infraspeed Expert Panel meeting on Derailment Provisions IDE(TRK+T&CDD
# 000006) Revision A November 2003.
RD15. Review of Derailment Risk for the HSL-Zuid Railway February 2004, Issue 1.
5 Glossary & Abbreviations
Check A guiding rail, located between the two running rails, and set close to
Rail one of the running rails to make contact with the back of a flange.
Normally used to prevent the opposite flange from making hard
contact with the running rail on a sharp curve; or to prevent the
opposite flange from taking the wrong route at a rail crossing.
DB Deutsche Bahn
DCP Derailment Containment Provision
DED Derailment Equipment Detection
Derailment Generic Term:
“A derailment is the action of one or more flanged railway
wheels departing from the rail”
HSL-Zuid RIA – Restated Implementation Agreement (contract with

Infraspeed) Term:
"A derailment on the HSL assets occurs when one or more wheels lose
contact with the rail"
Derailment The situation where the consequences of the accident are severely
Escalation aggravated, because the derailed train leaves the track, enters an
adjacent track, turns over, falls to a lower level, or jack-knives, etc.
Guard Rail A longitudinal rail running alongside a railway track and raised in
height above the running rails. Guard rails are sometimes found on
bridges and are intended to restrain the lateral movement of vehicles
which may become derailed.
HSL-Zuid High Speed Line - South
Page 51 of 92

R+D Research + Development

RCF Rolling Contact Fatigue
Robust Generally a reinforced concrete up-stand located adjacent to the track
Kerb at varying heights usually to restrain lateral movement following a
derailment on a rail bridge or to protect bridge piers.
SNCB Société Nationale des Chemins de Fer Belges

SNCF Société Nationale des Chemins de fer Français
TGV Train à Grande Vitesse
Page 52 of 92

Appendices
Appendices
C Summary of Questionnaires, Interviews & Media Search
D Causes of Derailments
F Accidents for Expert Panel Assessment
Page 53 of 92

Appendix
DERAILMENT CONTAINMENT
PROVISION QUESTIONNAIRE
HSL-Zuid
Location_______
Date____________________
This report is confidential and intended solely for the use and
information of the company to whom it is addressed.
Page 54 of 92

Interviewee Name Date of Interview
Project / Role on the Project

Organisation or in the
discussed organisation
1.0 High Speed Rail (HSR) Service

1.1 Where does the HSR line run?
1.2 How many miles/km of HSR service is provided?
1.3 What is the maximum line speed?
1.4 What is the average train speed?

1.5 How is the train powered? (3rd rail, diesel engine, overhead catenary)
1.6 If electric, what is the operating voltage?
1.7 Is the HSR service on exclusive right of way?
1.8 Is the HSR service for passenger service or for both freight and
passenger service?
1.9 How often do HSR trains operate in each direction?
1.10 What are the minimum headways?
1.11 What is the total length (miles/km) of single track?
1.12 What is the total length (miles/km) of double track?
1.13a What is the minimum vertical curvature of the track?
1.13b What is the percentage of track in a curve?
1.14 What is the minimum horizontal curvature of the track?
1.15 What is the maximum track cant?
1.16a What are the typical intervals between mainline turnouts?
1.16b What are the typical intervals between switches?
1.17 What provisions are made to keep road vehicles, people and animals
away from the right of way (fencing)?
1.18 How is broken rail detected?
1.19 What types of rail fasteners are used?
1.20 What types of pads are used under tie plates?
2 Rolling Stock
2.1 What type of rolling stock is used?
2.2 What is the length of a standard HSR train?
2.3 How many train sets support the HSR service?
2.4 Are they tilting train sets?
2.5 How many passenger cars?
2.6 How many power cars (or other cars)?
2.7 How many passengers can a train set carry?
2.8 What is the annual ridership and the distance operated?
3 Infrastructure
Page 55 of 92

3.1 How much track (by percentage or mile/km) is installed:

• in tunnels?
• on viaducts?
• on filled embankments?
• at grade?
3.2 How many highway rail grade crossings are included in the HSR
network? (Typical intervals 1 or 2 or 3 km)
3.3 How many highway overpasses are included in the HSR network?
(Typical intervals 1 or 2 or 3 km)
3.4 How many bridges are included in the system?
3.5 How many tunnels are included in the system?
3.6 How many mainline turnouts are installed in the track?
3.7 How is the track structure supported (sleepers on ballast, concrete
sleepers, concrete slab, etc.)?
3.8 Are train speeds reduced over certain types of track or infrastructure
features?
3.9 Who takes responsibility for maintaining the DCP?
4 Track Standards
4.1 What track standards are applied to the HSR line?
4.2 Do the track standards include any derailment containment strategies
(for example, do you require guard rails on bridges)?
4.3 If so, what types of derailment containment strategies are used?
• DCP Type 1 – Guide Rails inside the gauge
• DCP Type 2 – Guide Rails outside the gauge
• DCP Type 3 – Guide mechanisms that work on the bogie frame
4.4 How did you make the decision to use a specific derailment
containment strategy?
4.5 What type of information was available to help you make a derailment
containment strategy decision?
4.6 Have you considered and rejected other derailment containment
strategies? Why?
4.7 What are/ were the expectations of Derailment Containment
Provision?
4.8 If you use derailment containment strategies, have the results been
positive? What if any were the adverse effects?
5 Derailments
5.1 Have you experienced any derailments on your HSR operation?
5.2 If so, under what circumstances did they occur?
5.3 What was the cause of each derailment?
5.4 Do you use any specific technology to detect a derailment (dragging
equipment detectors, impact sensors, etc.)?
5.5 Do you consider preventive measures in the rolling stock (Hot Boxes
etc.)?
Page 56 of 92

5.6 Did the derailment escalate (per HSL definitions)?

• Train failed to follow track
• Train failed to remain upright
• Train entered into the clearance envelope for a train
potentially travelling in the opposite direction
• Other
5.7 Did derailment containment strategies adopted for your operation
play a role in controlling the derailment?
5.8 Could the application of a derailment containment strategy have
reduced the consequences of the derailment?
6.0 Other questions
6.1 What derailment scenarios do you worry about most when
considering derailments on your HSR system?
6.2 Are there specific facilities or structures located along the right of way
that could escalate a derailment?
6.3 Are there natural factors or considerations that could cause or escalate
a derailment (earthquakes, floods, brush fires, etc.)?
6.4 Do you have any experience with derailments being caused or
escalated due to one of these natural phenomena?
6.5 In your opinion, what differentiators between a low-speed derailment
and a high-speed derailment should be considered?
6.6 In your experience, what are the most important factors that cause a
simple derailment to escalate to a more serious accident?
6.7 What do you think can be done to control these escalation factors?
6.8 What would you like to see from the DCP in the future?
6.9 Which specific prevention measures are taken (i.e. gates, (more)
maintenance, nets near viaducts) and how effective are these
measures?
7 Any Further Information?
Page 57 of 92

Appendix

Note 1 The information provided by Japan Central Railway was for the sole use of the Client and Booz Allen. Since this report may be released to other parties, the information regarding Japan Central Railway
has been filtered.
DB SNCB Great Belt (Storebælt) railway link Oresund Link Japan
Denmark Sweden See note 1.
1.0 High Speed Rail (HSR) Service
1.1 Where does the HSR line run? 1 Hannover – Würzburg In Belgium, in 2004, the HSR lines run: Between the Danish islands of Funen (Fyn) and Zealand On the tunnel and bridge Tokyo - Shin-
2 Mannheim – Stuttgart - Between Bruxelles and the French border towards Paris (Sjælland) across the Great Belt (Storebælt) between Denmark and Osaka, Japan
3 Oebisfelde – Berlin (line 1) and From Funen to the small island of Sprogø in the middle of Sweden (fixed Oeresund
4 Cologne – Frankfurt/Main - Between Bruxelles and Liège towards the German the Great Belt, the railway is on the 6.6 km long West Bridge, link)
border and Köln (line 2). which carries both road and railway. From Sprogø to
Two other HSR lines are under construction: Zealand the railway passes through the 8 km long East
- line 3 between Liège and the German border and Tunnel.
- line 4 between Antwerpen Centraal and the Dutch
border
1.2 How many miles/km of HSR 1 327 km Line 1 : Bruxelles to the French border : 91 km with a section The part of the project covered by the requirements for risk 16 km 515.4km
service is provided? 2 99 km of 71 km (Lembeek to the French border) up to 300 km/h management and risk acceptance criteria includes 19.4 km of
3 159 km (design speed 320 km/h) double track railway.
4 164 km Line 2 : Bruxelles to Liège: 100 km with a section of 64 km
(Leuven - Ans) up to 300 km/h.
1.3 What is the maximum line 1 280 km/h 2 280 km/h The maximum line speed is 300 km/h. 180 km/hr (in some design aspects 200 km/hr has been 200 km/h (120 km/h for 270km/h
speed? 3 280 km/h 4 330 km/h Line 1 has been tested for a future maximum speed up to 320 used) freight trains)
km/h
1.4 What is the average train speed? 1 250 km/h Line 1 : 245 km/h In the update of the risk analyses just completed the ~ 180 km/h Information not
2 250 km/h Line 2 : 211 km/h following average speeds for passenger trains were assumed: for Public domain
3 250 km/h - West Bridge and Sprogø: 180 km/hr.
4 300 km/h East Tunnel: 131 km/hr.
1.5 How is the train powered? (3rd all overhead catenary The trains are powered by overhead catenary Diesel and overhead catenary Overhead catenary Overhead
rail, diesel engine, overhead catenary
catenary)
1.6 If electric, what is the operating 15.000 V/ 16,7 Hz 25.000 Volts 50 Hz 25 KV AC 25kV AC
voltage?
1.7 Is the HSR service on exclusive 1 no On line 1, the HSR service is exclusive right of way yes Exclusive
right of way? 2 no (High speed trains PBKA and Eurostar only for passengers
3 no up to 300 km/h)
4 yes On line 2, the HSR service is exclusive right of way for
passengers:
- up to 300 km/h for high speed trains PBKA
- up to 200 km/h for intercity trains (Ostende - Eupen)
push-pull formation
1.8 Is the HSR service for passenger 1, 2, 3 for both passenger and freight The HSR services are only for passenger services (see above) In 2015 (basis year for the risk analysis update) it is expected Both Passenger only
service or for both freight and service that there will be 51.650 passenger trains (31% electric and
passenger service? 4 only for passenger service 69% diesel) and 18.168 goods trains.
1.9 How often do HSR trains operate 1 4 trains/per hour L1: 533 trains per week in both directions With very few exceptions, all passenger trains are able to run Four trains per hour and Information not
in each direction? 3 3 trains/per hour L2: 7 trains per day and per direction 180 km/hr, such that the number of HSR trains per hour can direction (160 trains/day for Public domain
4 3 trains/per hour be estimated on the basis of the above 51.650 passenger thereof 20 freight trains)
trains in 2015.
Page 58 of 92


1.10 What are the minimum At operation with the LZB (line-type Minimal distance between “signals” = 1500 m 1.5 min Information not
headways? inductive train control system) it Minimal distance between trains = 9000 m (typical) for Public domain
depends on the braking capacity of both
successive trains
1.11 What is the total length none Line 1 : None 0 km
(miles/km) of single track? - One single track of 1.610 m with a maximum speed of
160 km/h for the junction at Antoing between an intern
Belgium railway line (L78) and the line 1, for the service
"high speed train" from Namur to Paris.
- One single track of 1.372 m with a maximum speed of
160 km/h for the junction at Silly between an intern
Belgium railway line (line 94) and the line 1.
- Line 2 : no single track
1.12 What is the total length 327 + 99 + 159 + 164 = 749 km Line 1 : 71 km of double track (300 km/h) 19.4 km as explained in 1.2 16 km
(miles/km) of double track? Line 2 : 64 km of double track (300 km/h)
1.13 What is the minimum vertical 4 14000 m We suppose the minimum vertical curvature of the track: 5000 m
a curvature of the track? Line 1 : 19.585 m
Line 2 : 15.000 m
1.13 What is the percentage of track We suppose the percentage of track in a horizontal curve. ~ 60 %
b in a curve? Line 1 : 58 %, Line 2 : 52 %
1.14 What is the minimum 4 3350 m We suppose the minimum horizontal curve of the track: 11.63 mm
horizontal curvature of the Line 1: 4.806 m, Line 2 : 4.650 m
track?
1.15 What is the maximum track 4 170 mm Line 1 : 150 mm 65 mm
cant? Line 2 : 130 mm (mixed traffic)
1.16 What are the typical intervals no typical intervals Line 1 : 29 km/20 km/30 km None. -
a between mainline turnouts? Line 2 : 21 km/29 km/14 km
1.16 What are the typical intervals General requirement: every 25 km a place There is no typical interval between switches. Switches are only located on the island of Sprogø. None on -
b between switches to overhaul trains On line 1 and 2 : an interval of 100 m between the ends of the the West Bridge and in the East Tunnel
switch blades. This distance can be reduced.
1.17 What provisions are made to General no special provisions, but most For people and animals, everywhere fenced. The road vehicles are kept on the motorway by means of - Fencing
keep road vehicles, people, of the line is fenced by noise protection For road vehicles, several dispositions: crash barriers. There is no public access to the islands of Any trespassers
animals away from the right of walls, wind protection walls and other - or embankment Sprogø. are banned by
way (fencing)? earthworks - or new jersey national law
- or concrete walls (can be noise barriers)
1.18 How is broken rail detected? No special detection By failing of the track circuit Electronic testing (every Information not
two years) for Public domain
1.19 What types of rail fasteners are Vossloh, different types for slab track and Pandrol system (elastic fastening clips) Pandrol fast clips
used? ballasted track
1.20 What types of pads are used ZW 700 and ZW 900 Elastic pads, 10 mm thick Pandrol
under tie plates?
2 Rolling Stock
2.1 What type of rolling stock is ICE 1, ICE 2, ICE 3, ICE-T TMST-sets (Transmanche) The diesel passenger trains at present are IC3 trains Information not
used? PBKA en PBA-sets (Paris-Bruxelles-Köln-Amsterdam) consisting of up to 5 sets of carriages. Each set consists of Oeresund-Trains for Public domain
TGV-Réseau three carriages, the one in the middle has no engine and the (trainsets)
two others have diesel engines.
The electrical passenger trains are mostly ER trains
consisting of up to five sets of carriages. Each set consists of
four carriages, the two in the middle have no engine and the
two others have electrical engines.
2.2 What is the length of a standard ICE 1: 360 m; ICE 2: 205 m (Double 410 PBKA, PBA, TGV-Réseau = 200 m During design a 300 m long IC3 train with 5 sets of carriages
HSR train? m); ICE 3: 200 m (Double 400 m); ICE-T: TMST = 394 m carrying 720 passengers was used as basis. 78,90 m
(7) 185 m (5) 133 m
Page 59 of 92


2.3 How many train sets support the ICE 1: 59; ICE 2: 44; ICE 3: 63; ICE-T: 43 Eurostar = 10 + reserve The trains are the trains generally used by the main Danish n.a.
HSR service? Thalys = 35 operator, DSB, in Denmark. See www.dsb.dk
TGV Bruxelles-France (also called TGV-Interconnexion) = +/-
8 + reserve
2.4 Are they tilting train sets? yes, ICE-T No No No Information not
2.5 How many passenger cars? ICE 1: 12; ICE 2: 7; ICE 3: 9, ICE-T: 7 or TMST = 18 cars / set See 2.1 for Public domain
5 PBKA, PBA, TGV-Réseau = 8 cars/set n.a.
2.6 How many power cars (or other ICE 1: 2 (Loco); ICE 2: 1 (Loco), ICE 3: 4; 2 powercars per unit (TMST, PBKA,…) See 2.1.
cars)? ICE-T: 1 n.a.
2.7 How many passengers can a train ICE 1: 645; ICE 2: 370; ICE 3: 415; ICE-T: TMST = 794 passengers See 2.2
set carry? (5) 250, (7) 381 PBKA, PBA, TGV-Réseau = 375 passengers 237
2.8 What is the annual ridership and ICE 1: 500 000 km/a/train; ICE 2: 515 L1 : 533 rides/week x 71 km x 52 1/7 weeks per year, L2 : 98 In 2015 the number of passengers is assumed to be 9 million.
the distance operated? 000 km/a/train; ICE 3: 520 000 rides/week x 64 km x 52 1/7 weeks per year Distance: See 1.2. n.a.
km/a/train; ICE-T: 400 000 km/a/train
3 Infrastructure
3.1 How much track (by percentage 1 Tunnel 122,3 km; Viaducts 32,6 km; Line 1: 6.6 km on the West Bridge Tunnel: 3.510 km Information not
or mile/km) is installed: Embankments/ at grade 172,5 km Tunnel: 1.177 m or 1.65 % 8 km in the East Tunnel Bridge: 7.845 km for Public domain
• in tunnels? 2 Tunnel 30,3 km; Viaducts 5,6 km; Viaducts : 4.142 m or 5.83 % The remaining km are at filled embankments leading to or
• on viaducts? Embankments/ at grade 63,4 km On filled embankments:20.600 m or 29 % from the West Bridge or at ramps leading to or from the East
• on filled embankments? 3 Tunnel 0 km; Viaducts 3,1 km; At grade (of the ground) : 12.600 m or 17.7 % Tunnel.
• at grade? Embankments/ at grade 156,0 km Line 2:
4 Tunnel 40,5 km; Viaducts 8,1 km; Tunnel: 757 m (Bierbeek) or 1.18 %
Embankments/ at grade 115,8 km Viaducts (on ground): 3.266 m or 5.1 %
On filled embankments: 20.800 m or 32.5 %
At grade (of the ground) : 11.746 m or 18.35 %
3.2 How many highway rail grade none No grade crossings. None -
crossings are included in the HSR There are no grade crossings authorised at a speed higher
network? than 160 km/h
(Typical intervals 1 or 2 or 3 km)
3.3 How many highway overpasses unknown Highway overpasses and underpasses None within the 19.4 km.
are included in the HSR network? Line 1 : 11 bridges (over or under) on 71 km - A bridge, every -
(Typical intervals 1 or 2 or 3 km) 6 to 7 km
(N508; N507; N52, A16, N50, N60, N56, N57, N55, N6, and
N203)
Line 2 : 8 bridges (over or under) on 64 km - A bridge every
to 8 km
(N3, N25, E40, N29, N64, N80, N69 and N614)
3.4 How many bridges are included Railway bridges Bridges = viaducts + road overpasses and underpasses + 6.6 km long West Bridge 1
in the system? 1 206 river overpasses +footbridges + aqueducts (concrete bridges
2 39 and pipes)
3 73 Line 1: 79 bridges and 94 aqueducts
4 101 Line 2: 76 bridges and 87 aqueducts
3.5 How many tunnels are included 1 62 Line 1 : 8 km East Tunnel 1
in the system? 2 14 Tunnel of Halle: 551 m
3 0 Tunnel of Tubize : 270 m
4 31 Tunnel of Antoing: 356 m
Line 2:
Tunnel of Bierbeek : 757 m
3.6 How many mainline turnouts are Line 1: 14 mainline turnouts None 8
installed in the track? unknown Line 2: 10 mainline turnouts
Page 60 of 92


3.7 How is the track structure ballastless track (slab track) The Track structure is: Ballast track Tunnel: concrete slab Information not
supported (sleepers on ballast, ballasted track with concrete sleepers - rail UIC 60 kg/m (long welded rails) track for Public domain
concrete sleepers, concrete slab, - prestressed monobloc concrete sleeper Bridge: ballasted on
etc.)? - elastic fastening system pandrol monoblock sleepers
- elastic pads 10 mm thick
- ballast 35 cm minimum thick under the sleeper
- mainline turnouts in rails UIC 60 with mobile heart
3.8 Are train speeds reduced over speed restriction in tunnels on lines 1-3: There is no speed restriction in section with a authorised For some of the trains including IC3, the speed in the East No
certain types of track or 250 km/h speed of 300 km/h tunnel is limited to 140 km/hr.
infrastructure features?
3.9 Who takes responsibility for Infrastructure owner, maintenance The Belgian railway company has created special multi- The infrastructure operator is Banedanmark, Rail Net Oeresundbro Konsortiet
maintaining the DCP? departments disciplinary staff for the maintaining of DCP of high speed Denmark, www.bane.dk. (Infrastructure Manager)
line.
4 Track Standards
4.1 What track standards are slab track, by exception only concrete Technical specifications interoperability of the European These are detailed in the Company
applied to the HSR line? sleepers on ballast with UIC 60 rails. community, EN and the UIC Register of Infrastructure established
standard which
covers tracks,
civil engineering
structures,
electrical and
signalling
facilities and
rolling stocks.
This standard is
approved by the
Transportation
Ministry of Japan.
4.2 Do the track standards include Guard rails are required only on lines The track standards include derailment containment There are guard rails on the West Bridge in accordance with - Yes, concrete and steel Yes
any derailment containment with mixed traffic under certain strategies. the practice used on major Danish railway bridges for a long “side walls” or guiding
strategies (for example, do you conditions, not on Cologne – time. devices
require guard rails on bridges)? Frankfurt/Main The East Tunnel is a bored tunnel, two bores with one track - No guard rails
each. There are elevated walkways on both sides of each - The devices are part
track. These walkways are made of concrete and designed of the bridge/tunnel
for a horizontal accidental load of 100 kN per metre length at construction, so
any level and position along the tunnel. their instalment did
Providing guard rails in the tunnel was considered, but this not trigger
was abandoned as not being cost efficient in view of the additional
walkways acting as derailment containment. investments
Hot box detectors, gauge control and derailment control - Concrete is used in
facilities are installed at both sides of the Great Belt crossing the tunnel and on
reducing the probability of derailment. the approach
bridges, steel is used
on the span bridge
due to
constructional
reasons (weight)
Page 61 of 92


4.3 If so, what types of derailment DCP type 1 - Guide rails inside the gauge especially on high DCP Type 1 for the West Bridge. n.a. DCP Type 1 and
containment strategies are embankments and on bridges DCP Type 3
used? + overflowed platform detectors at the low points of the HSR
• DCP Type 1 – Guide Rails DCP Type 1 line
inside the gauge + hot boxes detectors for the rolling stock
• DCP Type 2 – Guide Rails + concrete platform on chalky and old mining areas
outside the gauge
• DCP Type 3 – Guide On single track viaducts there are two guide rails; on the
mechanisms that work on other places only one guide rail near the outside rail
the bogie frame
Stability control of the bogie
More information has to be requested from SNCF
4.4 How did you make the decision According to the Belgian railway rules for the track and the See 4.2. The general approach used for the Great Belt project Technical risk analysis Information not
to use a specific derailment Railway standard requirements local characteristics for the stability of the platform. is that risk acceptance criteria were first established. Then and a small survey for Public domain
containment strategy? The Belgian regulations are close to those of France. the risk was estimated and compared to the risk acceptance
criteria. It is found that these criteria are met, such that
additional containment provisions are not required.
4.5 What type of information was According to the Belgian railway rules for the track and the See 4.5. It is noted that in Denmark there has been one No other railway used
available to help you make a Regulations (Railway standards) and local characteristics for the stability of the platform, after derailment on a bridge. In this case the guard rails served the system when
derailment containment expert reports sound geographical, geological and geophysical campaign of their purpose: In 1978 there was a derailment on the construction began.
strategy decision? measures and studies Storstrøms Bridge which carries a single track railway and a System is a unique
two lane highway between Zealand (Sjælland) and the development, so no
island of Falster south of Zealand. On a stormy night wind information was
caused a lightly loaded lorry to turn over and hit a lamp available
post. The lamp post fell across the track. The post was hit by
a locomotive and caught beneath the cowcatcher of the
locomotive such that the locomotive derailed. The
locomotive was caught by the guard rail.
4.6 Have you considered and Attention is paid to the piles of bridges and viaducts; on See 4.2. No, as the used strategy
rejected other derailment yes, on the base of expert reports for existing bridges the piles became an extra protection against seemed to be the “most
containment strategies? Why? special projects damage. natural and logic” one
4.7 What are/ were the To prevent We have considered no possibility of derailment Question not understood. To keep trains in the
expectations of Derailment • the deflection of a car body in the guiding device
Containment Provision? clearance of the opposite track,
• the impact of a train to structures
near the track.
4.8 If you use derailment 1992: collision on a bridge was prevented No derailment since the begin in 1995 of the operation at 300 There has been no train accident on the Great Belt Link since Yes.
containment strategies, have by guard rails. km/h of the line 1 (2003 for the line 2). opening for train traffic in July 1997. Our standard is
the results been positive? What 1996: impact to an abutment of a road No experience working.
if any were the adverse effects? bridge was prevented by guard rails
5 Derailments
5.1 Have you experienced any not on the special HSR lines, but in No derailment See 4.8 No None on
derailments on your HSR Eschede. Shinkansen.
operation?
5.2 If so, under what circumstances broken tyre of wheel 6 km in front of a - n.a. Information not
did they occur? bridge, then a deflection at a switch for Public domain
directly before the bridge, impact on
bridge piles, bridge collapsed
speed 200 km/h, 100 victims
5.3 What was the cause of each break of tyre of wheel and deflection at a -
derailment? switch n.a.
Page 62 of 92


5.4 Do you use any specific Indirectly with special sensors at the No technology to detect a derailment but some technology See 4.2. Detectors on both Yes, on bogies.
technology to detect a bogies to detect extraordinary shunting to prevent a derailment (see after) and stability control of the endings of the bridge
derailment (dragging movement bogie (tunnel)
equipment detectors, impact - Cable lying on the
sensors, etc.)? sleepers as an electric
circuit
5.5 Do you consider preventive At the moment no special detectors exist Yes, hot box detectors in track See 4.2. There are rail break Information not
measures in the rolling stock to detect a derailment, but this would be detectors on the trains for Public domain
(Hot Boxes etc.)? the best solution (further information
available at DSB)
5.6 Did the derailment escalate (per Yes, after the deflection at the switch the - n.a.
HSL definitions)? train failed to follow track.
• Train failed to follow
track
• Train failed to remain
upright
• Train entered into the
clearance envelope for a
train potentially travelling in
the opposite direction
• Other
5.7 Did derailment containment No Yes, in case of derailment, objective is to keep the train Information not
strategies adopted for your upright and outside the clearance of the opposite track n.a. for Public domain
operation play a role in (for example, two guide rails inside the gauge). Ballast is
controlling the derailment? very helpful to keep the rolling stock upright.
5.8 Could the application of a Theoretically, yes - n.a.

derailment containment
strategy have reduced the
consequences of the
derailment?
6.0 Other questions
6.1 What derailment scenarios do A ground collapse in chalky and in old mining areas which Derailment on the West Bridge resulting in carriages falling A train damaging the Information not
you worry about most when can affect the stability of the platform and the tracks. In such off the bridge. bridge structure for Public domain
considering derailments on areas, we use a concrete platform which allows a gate of 4 m In the risk analysis carried out, the escalation of
your HSR system? diameter under the platform. consequences in the event of derailment at the West Bridge
was considered including:
− Train entering into the clearance envelope of trains on the
other track.
− Train driving off the bridge and falling into the sea,
which in the analysis was assumed to be fatal to all
passengers in the carriages falling off the bridge.
6.2 Are there specific facilities or No experience No Open water as described above. It is noted that the carriages Bridge structure
structures located along the may fall into the sea at one side of the railway girder. At the
right of way that could escalate other side there is a 1.35 m wide gap between the railway
a derailment? girder and the adjacent road girder for the motorway
crossing such that carriages cannot fall into the sea here.
6.3 Are there natural factors or Yes: - Collision of a ship
considerations that could cause High water for lines besides a river • The ground collapse in chalky and old mining areas with a column while
or escalate a derailment (see above) a train is running on
(earthquakes, floods, brush • the overflowing of the platform at the low points of the bridge
fires, etc.)? the HSR lines (detection devices are installed in - Most of the columns
these areas) are protected by
• earthquakes, but in a limited way islands which
(viaducts and tunnels are calculated for preventing any would let the ships
derailmentPage
of high
63speed
of 92 train by an earthquake) ground before
hitting a column

6.4 Do you have any experience No experience, only information about a ground collapse on No No.
with derailments being caused No line 3 in France near Amiens
or escalated due to one of these
natural phenomena?
6.5 In your opinion, what For the same rolling stock, it is obvious that a high speed The consequences of a high-speed derailment of course may be larger None Information not
differentiators between a low- derailment is more dangerous for the passengers and than the consequences of low-speed derailment. for Public domain
speed derailment and a high- damages the rolling stock and infrastructure. In the risk analysis carried out for the West Bridge, it was considered that
speed derailment should be But, on a high speed line, the eventual damages of a the guard rails would be less efficient in the event of a high-speed
considered? derailment have been reduced by the use of a special rolling derailment than in the event of a low-speed derailment.
stock (high speed trains with bogies between the cars which
keep the train as one)
Note that ICE-3 trains will be allowed to ride on Belgian
HSL; this rolling stock has no bogie between the cars
6.6 In your experience, what are Track irregularities like switches, No experience Speed, weight,
the most important factors that crossings and others derailment
cause a simple derailment to containment
escalate to a more serious provisions
accident?
6.7 What do you think can be done No experience Derailment containment provisions
to control these escalation
factors?
6.8 What would you like to see Entrance of tunnels need to be equipped with derailment containment
from the DCP in the future? provisions (from 500 m before the entrance) so that a derailed train
cannot collide with the walls of the tunnel entrance
6.9 Which specific prevention Passive protection systems along roads, No specific prevention, only a regular visit of the Normal preventive maintenance strategy
measures are taken (i.e. gates, on road bridges to prevent the deflection installations according to the UIC leaflet “Maintenance of
Additional inspections at the connection between slab track and ballasted
(more) maintenance, nets near of vehicles to the rail. high speed lines” (1996)
track
viaducts) and how effective are
these measures?
7 Any Further Information? Each morning the lines are declared secure for operation Information on the Great Belt project and the risk analysis work carried
after a ride of an “opening train” (train balai according to the out can be obtained from "J. Kampmann, K. Kieler and B. Kohl: Risk
French practice). Analysis of the Railway Tunnel under the Great Belt. Safety in Road and
This opening train can be commercial or non-commercial Rail Tunnels. Basel, Switzerland, Nov. 1992". It is noted that the risk
ride of any train at 160 km/h maximum. acceptance criteria for user risk presented in this document are no longer
These opening rides are very helpful to avoid unexpected valid, as the risk acceptance criteria for the Great Belt Link have been
obstacles when the rides at 300 km/h start. revised recently such that changes in the traffic across the Link, both road
It is considered as a major contribution to our DCP-strategy. and rail, can be taken into account. These new criteria have not been
published.
Page 64 of 92

Appendix Summary of Questionnaires, Interviews &

C Media Search
Notes for Appendix C:

1) Shading indicates information from media source.
2) A shaded row indicates the source material was from the reference stated in the left column.
3) Where text only is shaded then this is from the stated source stated at the end of the block of text.
4) No shading indicates text from interview.
Organisation / Project Type of DCP Why Comments Identified Adverse

Effects
Deutsche Bahn AG Type 1 Regulations (Railway standards) There is a common practice for derailment provision None identified
required that guard rails are which involves the use of a guard rail. The derailment
required only on lines with provision is not specific to high speed lines.
mixed traffic under certain Source: [RD 14]
conditions, (not on Cologne –
Frankfurt/Main) and the
decision to use DCP 1 was also
based on expert reports.
SNCB Type 1 According to the Belgian railway The Belgian regulations are similar to the French None identified
Type 3 rules for the track and the local regulations.
characteristics for the stability of
the platform, after sound
geographical, geological and
geophysical campaign of
measures and studies.
Summary of Questionnaires, Interviews & Media Search
Page 65 of 92


Effects
Great Belt Type 1 There are guard rails on the The general approach is that risk acceptance criteria None identified
(Storebælt) Railway Type 3 West Bridge in accordance with were first established. The risk was estimated and
Link Denmark the practice used on major compared to the risk acceptance criteria. It was
Danish railway bridges for a found that these criteria have been met, such that
long time. additional containment provisions are not required
The East Tunnel is a bored
tunnel, two bores with one
track each. There are elevated
walkways on both sides of each
track. These walkways are
made of concrete and designed
for a horizontal accidental load
of 100 kN per metre length at
any level and position along the
tunnel. The walkways act as
derailment containment.
Oresund Link Type 3 The system is a unique None identified
Sweden/ Denmark development, so no information
was available at the time of
construction, therefore a
technical risk analysis was
undertaken.
Page 66 of 92


Effects
Central Japan Railway Type 1 Company established standard None identified
Company & Type 3 which covers tracks, civil
Massachusetts Institute engineering structures,
of Technology electrical and signalling
(SHINKANSEN Japan) facilities and rolling stock. This
standard is approved by the
Transportation Ministry of
Japan.
Lyon Turin high Planning Based on the experience on the None identified
speed link France / to use Channel Tunnel.
Italy Type 3
Italy Increased The selection of these measures There is no common practice for DCP. [RD 14] There are no
separation of was the result of technical adverse affects on
track on considerations and empirical low speed lines
High Speed evidence on derailments on
Lines traditional lines.
Lyon Turin high Planning Based on the experience on the None identified
speed link France / to use Channel Tunnel.
Italy Type 3
High Speed Lines Type 1 No Information None identified
France
Page 67 of 92


Effects
SNCF Type 1 The approach on all railway lines is focused on the Robust kerbs are
[RD4] and provision of DCP where the risk is greatest. The not preferred since
Const- current practice is to use DCP Type 1. they interfere with
ruction mechanical track
design For structures under, adjacent to, or over the railway operations.
LGV used the approach of UIC leaflet 777-2, plus the
addition of a safety rail on existing LGV where this is
an adjacent structure nearby.
DCP Type 1 is used for railway bridges and viaducts

where the length is greater than 50m or the speed is
higher than 130Km/h, or where the risk is high i.e.
housing under the viaduct, or where a derailment off
the main line could infringe on the main line.
SNCF[RD14] There are exceptions. If a bridge is wide enough and Because of the
the track distance is wide enough (>4,5m), no torsion (change of
derailment provisions are installed, and at cant) of the track
transition curves where alignment changes into a some stiff vehicles
narrow curve (less than 2000m). may derail at very
low speeds due to
On high speed lines derailment provision is used. wheel climb. Some
Here a standard design is used (similar to Belgium). times this is solved
by lubrication of
the track.
Page 68 of 92


Effects
SNCF [RD15] The following high speed derailments have No Information
occurred:
• A TGV travelling at 270 km/h at Macon Loche,
France caused derailment of one bogie due to wheel
defect. No fatalities, 27 injured.
• A TGV travelling at 294 km/h at Haute
Picardie, France caused collapse of track structure
due to WW-I trench. No fatalities, 2 injuries.
• A Eurostar travelling at 300 km/h at Croisilles,
France caused by mechanical failure of bogie
component. Emergency braking caused additional
cars to derail. No fatalities, 14 injuries.
Derailments involving High Speed Trains Less than 200

km/h;
• A TGV train travelling at 120 km/h at Laval,
France caused by Derailment due to emergency
brake application in response to mud slide. No
fatalities, no injuries;
• A TGV train travelling at 130 km/h at Dax,
France caused Obstruction on track caused a
derailment and the overturn of a power car. No
fatalities, 5 injuries.
Page 69 of 92


Effects
Netherlands [RD14] Common practice is to use DCP only at bridges No Information
(steel) and flyovers (note a concrete plinth is
constructed on the Hemboog line).
Netherlands [RD3] A passenger train derailed in Baarn in 1999 at 40 km No Information
per hour with minor consequences. The cause of the
derailment was the failure of a wheel due to fatigue
cracks. As a result of the accident the investigation
recommendations were:
• Mitigating the technical risk of failure the
wheel sets by developing a maintenance strategy
• Reconsider the policy instrument of
compulsory certification of maintenance
companies (assessment of business processes).
Page 70 of 92


Effects
UK CTRL Type 1 Based on a Derailment CTRL recognised that no DCP is 100% effective and that None identified
Type 3 Containment Strategy the attention should be focused on preventing the causes
(ALARP Risk of derailments and rather than managing the consequence.
assessment)
The approach to DCP was based on their Train Accident
Model which is an averaging model in the sense that it
takes and average embankment etc.
There was no justification for the provision of DCP if

CTRL was to have been a passenger only line. The
additional risk introduced by mixed traffic required DCP
to by installed at strategic and high risk locations.
Due to the absence derailment data and evidence of DCP

effectiveness, the risk analysis used in assumptions and
estimated ranges and considered common derailments.
The report identified that check rails (DCP Type 1) and
guard rails (DCP Type 2) were not effective at high speed.
CTRL undertook an approach which was a variation on

UIC Leaflet 777-2 ‘Structures built over Railway lines-
Construction requirements in the track zone’. In addition
the positioning of the DCP Type 3 is designed so that
during a derailment the concrete kerb acts and the running
rail act as a barrier to the laterally displaced wheelset.
A key lesson from CTRL is that the theoretical approach to

DCP as a result of analysis has to be technically feasible
and practicable and compatible with the track bed design.
Page 71 of 92


Effects
UK There is no common practice for DCP in the UK [RD14] No Information
The risk of derailment is assessed on a case by case
basis in accordance with the ALARP principle used in
the UK. Therefore there is no single standard on
derailment prevention, however the most relevant is
GC/RC 5510 – Recommendations for the design of
bridges.
UK Heathrow Type 3 The HMRI required that Derailment containment consists of a central concrete No Information
Express Tunnels the project should up-stand in the four-foot area. The up-stand extends to
[RD6] incorporate derailment rail level, and permits 215mm excursion of a derailed
containment throughout wheel. At the locations of equipment to be installed in
the lengths of the tunnels. the four-foot; the derailment containment is stopped.
UK Rail Safety & The RSSB research paper concludes that when there is No Information
Standards Board an impact with Overhead lineside structures there may
Report be a significant amount of damage to the train. The
“Engineering severity is dependent on which part of the train absorbs
Overhead line the impact. A more significant finding for this report is
structure design to that the lineside structures can have a beneficial
cater for collision” containment effect.
[RD12]
Page 72 of 92

Organisation / Project Type of Why Comments Identified Adverse

DCP Effects
UK Rail Safety & This report produced for RSSB was a result of a review of No Information
Standards Board Report derailments in the United Kingdom over the period 1992-
“Engineering 2001 to identify the casual events and their associated
Derailment mitigating standards.
mitigation –
catergorisation of The derailments were split up into 3 categories:
past derailments” • Category 1: Derailment due to track or vehicle
[RD13] factors where either the track OR the vehicle were
non-compliant with the pertaining standards (32% of
derailments)
• Category 2: Derailment due to track or vehicle factors
where both the track AND the vehicle were compliant
with the pertaining standards (10 of derailments)
• Category 3: Derailment not due to track, vehicle or
track/vehicle interaction. (43 of derailments)
• Uncategorised (15% of derailments)
Key findings of the report showed that of the derailments

identified as category 1 (32% of all derailments) 92%
solely affected freight services.
The report identified that the speed of derailment was

known in 60% of Category 1 & 2 derailments. Nearly 50%
of the derailments occurred when the train was travelling
at less than 16km/h.
Page 73 of 92


DCP Effects
Economic Commission Extract: Mitigation of the consequence of accidents No Information
for Europe Infrastructure measures
Recommendations of Standard C.2 01 Derailment containment measures

the Multidisciplinary Derailment containment measures should be provided
group of experts on in all tunnels. The tunnel profile should be kept as free
safety in tunnels as possible from obstacles, which might snag a derailed
(rail) [RD9] train.
Korean High Speed Type 3 Concrete containment kerbs are used, which are high No Information
Railway enough to align with the bearing axle box of the train to
Source: [RD4] prevent undue force on the wheelsets or the train body.
These kerbs are situated sufficient distance from the
rails to allow ballast-cleaning machines.
Taiwan High Speed Type 3 Based on expert judgment Cost
Rail Corporation
New South Wales Statistics on: No Information
Australia [RD2] • Number of Derailments
• Number of Derailments by Line Type, by Rolling
Stock Type and by Top 5 Reasons
Page 74 of 92


DCP Effects
US AMTRAK None Amtrack does not have DCP on The focus is on derailment prevention through good DCP installed on
Source [RD 12] the track. The track is designed to track and train maintenance practices infrastructure
keep the train upright or within a interferes with
certain envelope. (The trains are maintenance and
equipped with locking couplers at times, can
which provide some constraint potentially become
between the vehicles to prevent a higher risk than
one vehicle from twisting relative no containment
to another). provision.
Federal Railroad The Federal Railroad Administration has undertaken Where barriers are
Administration, Safety extensive research including modelling of intrusion applied on both
of High Speed Guided barrier design. (DCP Type 3). They have found that sides of the track
Ground Transportation structural rather than earthwork barriers are more then high speed
Systems, Intrusion effective. In the instance of using structural barriers on and low speed
barrier Design Study. one side there were distinct differences in the output trains have a
[RD4] from the modelling. At high speeds the train tendency to get
experiences a glancing blow with the barrier and travel wedged between
down the track and come to rest in a shallow zig-zag. the two barriers,
Whereas at low speed the coached remain in contact and get pushed
with the barriers longer and rest in a sharper zig-zag. into the barriers
further by the
coaches behind.
Page 75 of 92

Appendix
D Causes of Derailment
D1. Introduction
The following causal factors are common reasons for derailments that have occurred in
the past on all types of trains - not just high-speed trains.
D2. Track Related
The discussion on track related derailments includes the track roadbed, track structures,
and any special track work included on the system.
The roadbed consists of the rails, the sleepers, the fasteners between the rails and
sleepers, and the ballast, concrete slab, or subgrade used to support the track. The entire
track structure is designed to support the weight of passing trains while maintaining the
required geometry of the track. The track structure is not perfectly rigid but flexes with
the passing of trains. The track structure, whether based on ballast, concrete slab, or
other subgrade material, must support the train and maintain the tolerances on the
geometry of the track. A subgrade that does not provide the required support will move
under the train and quickly deteriorate.
A common cause of subgrade problems is inadequate drainage. Inadequate drainage

can result in soft spots forming under the rail and sleepers. The soft areas allow the
track to sag under passing trains and allow variations in the alignment and cross-level of
the track. The more the track moves, the more the geometry of the track can deteriorate.
Excessive cross-level can allow a car to rock back and forth and derail by walking off of
the rail. More serious problems can cause the roadbed to collapse under the weight of a
train.
Another type of track condition that can cause a derailment is track buckling. Track
buckling usually occurs during hot weather on continuous welded rail. As the
temperature rises, the rail tends to expand (grow longer). The internal stresses from rail
heating are normally counteracted by the resisting forces of the track structure and
subgrade. If the resisting forces are inadequate, the rail will buckle and the track will
assume an “S” shape. A train travelling at a high enough speed will derail when it
encounters the sharp curve of the track buckle. Even if the first carriage to encounter the
rail buckle does not derail, the forces transmitted from the wheels to the track buckle
tend to accentuate the buckle causing following cars to derail.
Broken rail is another common cause of derailment. Broken rail is the physical parting
of the rail through a lateral break. The moving train causes the track to flex and pushes
down the leading edge of a rail break. The wheel then strikes the opposite (unloaded)
end of the broken rail until it fractures and can no longer support the train. Rail breaks
Page 76 of 92

are often initiated with an inclusion of an impurity in the rail from the time of
manufacture. Cracks propagate from the inclusion until the fatigue crack passes
completely through the rail.
Other types of rail breaks can also result in a derailment. Occasionally, a piece of rail
will break out and cause a derailment.
Finally, track caused derailments can also result from improper rail geometry. Rail gage
and cross-level and other critical dimensions on the rail must be tightly controlled to
prevent failure. Wide gauge, tight gage, excessive cross level variation, and other
geometry variations that cause the track to be out of tolerance can cause a derailment.
Inspection and maintenance of the right of way is key for detecting and correcting these
anomalies.
D3. Special Track Work
Special arrangements of track in the form of switches, turnouts, crossovers, or other

configurations can initiate or escalate a derailment. Defects in the adjustment and
maintenance of switches can cause an accident. They represent an incongruity in the
track structure and if not maintained, can quickly cause a derailment. Special track
work often occupies the space between the two running rails where it can cause an
obstruction for a derailed wheel. For example, a facing point turnout uses a curved rail
that guides a railed wheel from one running rail across the opposite running rail. The
configuration of a facing point switch can also capture a derailed wheel and direct it off
the side of the right of way. Through this process, the rail and roadbed can be heavily
damaged and a general derailment can occur.
Other issues associated with special track work can also contribute to derailments.
Components used for points, frogs, or the switch mechanism itself can become worn or
broken and cause a wheel to lose contact with the rail, climb on top of the rail, or split a
switch. When a train splits a switch, part of the train takes one route and the rest of the
train takes the diverging route.
D4. Vehicle Defects
Defects in the rail vehicle can also cause a derailment. Broken wheels, broken axles and
problems with the bogie or suspension are common causes of accidents. Wheels are
manufactured from cast steel and machined to provide the tread and flange on the
wheel profile. Wheels can crack or overheat and fail catastrophically. Wheel profiles
can be worn to the point where they cause the train to derail. Worn or defective wheel
treads can also allow a wheel to derail
Vehicle bogies can develop mechanical problems that make it difficult for the bogie to
rotate in curves. Excessive forces build up at the wheel flange forcing the flange to climb
over the rail and derail.
Page 77 of 92

D5. Operational Issues
Derailments may also be associated with operational issues such as collisions, excessive
speed, train handling, or failure to adhere to operating rules. Train handling
derailments are sometimes caused by the application of emergency brakes on curving or
undulating track. The forces that build up between carriages can become excessive
during a brake application and can actually push a carriage sideways out of the train.
Excessive speed is another common reason for a derailment. Failure to comply with
speed restrictions - especially on curves - is a significant cause of accidents.
D6. Collisions
Other derailments can be caused by collisions. Collisions can occur train-to-train, train
to highway vehicle, or train to obstruction. Although train-to-train collisions often result
in a derailment, the more important issue in that type of accident is the cause of the
initial train-to-train collision. Train to highway vehicle collisions are common in areas
where grade crossings are used to allow highway vehicles to cross the railway tracks.
Collisions with obstructions on the track are also common. The obstruction may be
natural (rock slides, mud slides) manmade debris from passing trains, or debris that has
been intentional placed on the track by vandals. In any case, hitting another train, a
highway vehicle, or an obstruction can cause serous derailments.
Page 78 of 92

Appendix
Concrete plinth for Rheda track
STARTER BARS 12
OPTIONAL: DEMU BAR CONNECTORS
5 12
INTERMEDIATE LAYER
FREE DRILLING ZONE
500
TYPICAL CROSS-SECTION AT CANT = 0
Slide 1 July 2004
Page 79 of 92

Derailment Provision
Concrete plinth – end sections
stirrup ∅16
4 ∅16 stirrup ∅12
per sleeper
500 mm
200 mm
1:75
4∅12
7500
per sleeper
1125
0
normal width
derailment provision
500
sleepers
inner width axle = 1437 – 37 (flange) = 1400
minimum width
derailment
2x (1400-1300)=200
half width Rheda = 1300
Slide 2 July 2004
Page 80 of 92

Appendix Accidents For Expert Panel

F Assessment
Introduction
This appendix contains accidents initiated by train, operations, infrastructure and

human factors causal events. The accidents will be reviewed by the expert panel in
accordance with the methodology described in section 1.3 of the main report.
F1. Accident 1 - Eschede, Germany
On Wednesday 3 June 1998, at about 10:59 am

local time, northbound ICE (Inter City
Express) Train 884 en route from Munich to
Hamburg derailed near Eschede, Germany.
The ICE Train consisted of two power units
(one on each end of the train) and 12
passenger cars. The accident occurred on the
Hanover to Hamburg line that was part of the
original ICE system. The Hanover to
Hamburg section of the ICE system was not newly built but was a renovated line that
was upgraded in 1991 from an existing rail line; consequently the line does not meet
all Deutsche Bahn (DB) engineering standards for a new high-speed rail line. Mixed
traffic, freight and passenger operates on this line. On this section of the rail line, the
maximum allowable speed is 200 km/h.
The area of the accident is fully equipped with an LZB signal system. The LZB on
this section of rail line was installed in 1965 and upgraded in 1990. LZB is an
Automatic Train Control (ATC) system overlaid on the current signal system. It is a
moving block system that continuously calculates the safe stopping distance for the
train and monitors the traffic ahead. The LZB continuously transmits and receives
train and signal information through an induction cable that runs down the middle
of the track structure. This induction cable is configured into circuit loops that span
the wayside signal blocks. When a train enters a loop, the wayside system takes and
records data from the train and time stamps it. The time stamp is only accurate to ±1
minute. This data is also available to the train dispatcher. In the event that
communication between the induction cable and the lead power unit is lost due to
malfunction or damage, the system reverts to the wayside signal system. Train
speed is also automatically governed to provide adequate braking distance between
the wayside signals. The LZB system is required on lines where the train speed is
greater than 160 km/h.
The accident sequence began about 5.6 km before the scene of the general derailment.
In this area, the track is tangent and there are no tight curves. The train was
operating at 180 km/h, 20 km/h below the authorized speed of 200 km/h for this
section of track. At this time, a wheel rim on a two-piece resilient wheel broke on the
3rd axle of the first car behind the lead locomotive. The tread of the wheel separated
Page 81 of 92

from the resilient core and wrapped itself around the track brake on the trailing
truck. Debris including broken bolts, pieces of under car bulkheads, and under car
baffles was found within a 100 m section of track. There was also evidence of
physical damage to concrete ties in this same area. Passengers from the first car
interviewed after the accident reported noise and vibration about 2 minutes before
the general derailment.
The damaged wheel cut the LZB inductive cable loop approximately 5.5 km south of
the bridge (loop 9). The wayside system recorded this event. The lead power unit
equipped with the receiving antennae, was past loop 9 and had entered loop 10 at the
time loop 9 was cut; thus the engineer received no indication of a problem with the
LZB system. The inductive cable loop varies in length but typically is 2 km long.
When the train entered loop 11, the induction cable was damaged again. The lead
power unit was still in loop 11 when communication between the train and the LZB
was lost. There was a delay in time in the recording of this event due to the
computer receiving an overload of information from a variety of sensors detecting
malfunctions in the train’s systems as a result of the derailment. Although there was
a delay in recording the data, the brakes were automatically applied when the
damage was detected. At the time of the general derailment, the train was still
operating at a speed of 180 km/h.
There are four tracks in the area of the accident, two main tracks and two siding
tracks. The main tracks were located between the two sidetracks and were
numbered track 1 and 2 from west to east. The accident train was travelling North
on track 2 at the time of the general derailment.
There were two turnouts in the area of the general derailment. The first turnout the
accident train encountered was a trailing point turnout that went from track 1 to
track 2. The second turnout was a facing point turnout from track 2 to the east siding
track. When the accident train approached the first turnout, the broken wheel tread,
still hung up on the track brake, fouled the guide rail of the first turnout. The force
of this collision caused the entire length of the 9-metre guide rail to be torn from the
roadbed and penetrate the floor of car 1. The guide rail then broke in two pieces.
The first piece of guide rail pierced the ceiling and water tank in the roof of car
number 1. The second piece of the guide rail went up through the floor of car 1 and
passed through the diaphragm of cars 1 and 2. The guide rail wedged itself against
car 2 causing the car to lean sideways. The driver of the train received a track brake
applied light when this occurred but did not experience any other sensation and did
not realise anything was wrong with the train.
After striking the guide rail, the train continued towards the second switch. At this
point the lead power unit and the lead bogie of car 1 traversed the switch still on the
rail. The trailing bogie of car 1, as a result of impacting the guide rail, veered
sideways causing the left wheel to impact and break the open switch point. The
switch point was damaged and the switch locks were broken. The trailing bogie of
car 1 continued past the switch on the wrong side causing the switch to be pushed
over, lining the tracks for the siding. Car 1 struck the wayside signal located just
prior to the bridge. Car 2 followed car 1 and proceeded straight through the switch.
The front truck of car 3 followed car 2 but the rear truck of car 3 was diverted to the
Page 82 of 92

siding track and derailed, this caused car 3 to be forced out towards the bridge
supports. The trailing end of car 3 knocked out the bridge supports. One of the
supports was found several hundred meters from the bridge and driven directly into
the ground.
Striking the bridge support caused the bridge to collapse. Each car is 23 m in length
and the train was travelling about 50 meters per second. One and a half additional
cars passed under the bridge before the bridge collapsed. The fourth car veered off
into the woods and the fifth car was partially crushed by the falling bridge. The
following seven passenger cars and the trailing power unit piled up on the south side
of the bridge.
Cars six and seven were directly under the concrete bridge and were crushed by the
falling bridge.
Some time during the accident sequence, the head power unit separated from the rest
of the train and the locomotive coasted to a stop at the Eschede station. Only then
did the engineer realise he had lost his train.
Ninety individuals died at the scene of this accident. Eleven more died of their
injuries in the weeks after the accident. It was estimated that 200 to 250 passengers
were injured in the accident. Neither the DB nor the local authorities were able to
provide a number of total passengers on board the train.
The Eschede accident was initiated when a resilient wheel mounted on the trailing
truck of the first passenger car broke. The wheel had succumbed to a fatigue crack.
The thickness of the broken wheel that initiated the derailment was found to be
under the condemning limit. There were no signs of impurities or inclusions at the
point where the crack initiated. The resilient wheel was specially designed and
tested for high-speed service on the ICE. Resilient wheels, however, are not normally
used in high-speed applications. Their use is normally confined to trams and light
rail vehicles where the consequences of wheel problems are tempered by the lower
speeds.
The original design of the ICE-1 series trains used a solid wheel. Resilient wheels
were applied after problems with the solid wheels developed. The solid wheels were
wearing eccentrically and developing an oval shape. The cause of the unusual wear
pattern was never determined. The resulting noise and vibration was very apparent
to the passengers. Although the resilient wheels also wear in an oval pattern, the
noise and vibrations were no longer transmitted into the car body. Later versions of
the ICE train-sets returned to solid wheels but used an air spring suspension to
isolate the car body from the noise and vibration.
Cracks in wheels can propagate very fast. To reduce the chance of a broken wheel,
ICE train-sets are fully inspected at the end of each run using ultrasonic testing
equipment. Resilient wheels are very effective at damping out vibration.
Unfortunately, the same qualities of resilient wheels that damp out vibrations also
impair the ability of ultrasonic wheel testing equipment to find internal defects.
Page 83 of 92

Wheel tread condition can also invalidate ultrasonic wheel tests. Flat spots, surface
cracking, marring and other common conditions found in a worn wheel mask the
defects in the same way that surface conditions on rail can mask internal track
defects. Ultrasonic testing - although somewhat effective on solid wheels - is not
adequate for detecting internal flaws on resilient wheels.
F2. Accident 2 – Waterfall, Australia
At about 7:15 am on Friday 31 January 2003,

State Rail Train C311 derailed and overturned
on a curve about 2 kilometres south of the
Waterfall Station. The train was carrying 47
passengers and 2 crewmembers. Six
passengers and the train driver were fatally
injured in the accident. Nineteen others were
seriously injured and required hospitalisation.
On the day of the accident, the 6:24 am

passenger service on the Illawarra Line from
Sydney Central Station to Wollongong was provided by a four car Tangara train set
designated G7. The train proceeded normally from Sydney to Waterfall, arriving at
the Waterfall Station about 2 minutes late. At 7.13 am, the train left Waterfall Station
travelling towards Helensburgh station. The Illawarra line between Waterfall and
Helensburgh is a two track main containing areas of twisting curves. The train was
travelling south on the “down” or easternmost track. At about 7.15 am, the train
entered a 240 metre radius left hand curve restricted to 60 km/h when the derailment
took place.
All cars in the train rolled onto their right side and landed on the “up” or
westernmost track. As the train slid down the adjacent track, it struck two overhead
catenary stanchions and then collided with the rock face of a cutting. The collision
with the rock face righted the first two cars. The train came to rest with the front two
cars in the upright position and the two rear cars on their right side.
The Special Commission of Inquiry who investigated the accident found that the
driver had been incapacitated by a heart attack shortly after leaving the Waterfall
Station. The train continued to accelerate and reached a speed in excess of 117 km/h
before overturning on the 60-km/h curve.
Page 84 of 92

F3. Accident 3 - Hatfield, United Kingdom
Details of the Incident

On 17 October 2000 four people were
killed and 70 injured as a result of a
derailment of an Intercity 225 near
Hatfield. The Great North Eastern
Railways Intercity 225 trainset was
travelling at or close to the line speed of
185 km/h.
The following hyperlink provides an
anmimation of the crash
http://www.guardian.co.uk/hatfieldtraincra
sh/flash/0,7365,383951,00.html
Causes of the derailment

The Hatfield derailment happened because a rail, in which there were multiple
cracks and fractures due to rolling contact fatigue (RCF), fragmented when a high-
speed train passed over it. The Investigation Board identified, under the heading
‘direct causes’, a number of interim recommendations intended to prevent this type
of catastrophic rail fracture occurring again.
In addition, there were a number of contributory factors which made it more likely
that this derailment would occur, and a range of aggravating factors may have made
the consequences of the derailment worse than they might otherwise have been.
The incident happened because a train travelled at the permitted speed over a rail
that had been identified as in poor condition, and which should have either been
replaced or a temporary speed restriction applied.
The following recommendations relate to the management systems which could have
prevented these circumstances.
• Health and safety management-
o Increased training and competence.
o Quicker and more responsive mechanisms established by which
employees can bring safety critical matters to the attention of
managers.
o Performance of infrastructure maintenance contractors and other
track-related contractors.
• Management of maintenance
o Implement an effective maintenance programme to ensure that the
probability of a safety critical rail fracture is as low as is reasonably
practicable.
• Inspection of track
o Current best practice in detecting RCF should be implemented i.e.
automated.
o Procedures for rail inspection, both visual and using NDT techniques.
Page 85 of 92

Aggravating factors
These interim recommendations address matters which, in the opinion of the

investigation board, may have made the outcome of the derailment worse than it
might have been.
• Train sets should be designed, built and maintained to maximise the chance
of their remaining upright and intact during high-speed derailment.
Particular aspects of rolling stock design which should be reviewed are:
o bogie and suspension component retention.
o Strengthening of the attachment systems should be considered;
o ‘Tightlock’ couplers; their propensity to open when rotated should be
assessed and the design loads reviewed.
o strength of vehicle roofs and walls;
o passengers seats; the risks to passengers as a consequence of seat
damage or failure should be reassessed
o design of catering facilities should be reviewed to minimise the risk to
staff in the event of an accident
• The design of overhead line equipment stanchions should be reviewed with a
view to making them less likely to penetrate passenger space in the event of a
collision. In addition, the risks associated with trains striking any trackside
equipment in a derailment should be assessed.
Key Findings
Key findings included:
• obvious and significant evidence of a rail failure.
• evidence of significant metal fatigue damage to the rails in the vicinity of the
derailment.
• the only evidence to date of wheel damage is consistent with the wheels
hitting defective track.
• there is no evidence, so far, of a prior failure of rolling stock.
• The most extensive damage appears to have been caused by derailed
carriages impacting line side structures.
Page 86 of 92

F4. Accident 4 - Mobile, Alabama, United States America
In the early morning hours of 22 September 22 1993,

Amtrak's "Sunset Limited" plunged into Big Bayou
Canot near Mobile, Alabama. Forty-seven people were
killed. It is the worst U.S. railroad accident in over 46
years.
At 2:53 am, Amtrak's "Sunset Limited" struck the girder

on a displaced span of the CSXT railroad bridge and
derailed into Big Bayou Canot. Access to the accident
site was difficult. The accident site could only be
reached by rail or by boat.
The first locomotive was buried 50 feet in the mud on

the upstream side of the bridge. The 3-man crew was
killed instantly and were trapped in the locomotive. The
second locomotive unit came to rest on the upstream
side of the bridge in about 15 feet of water.
The third locomotive and a baggage car landed on the downstream side of the bridge.
The next car in the train was an On Board Service Crew car. On Board Service Crew cars
are included to provide a place for the OBS crew to rest. Two OBS crewmen were killed
in the fire that resulted from diesel fuel spilled from the locomotives.
The first passenger coach was partially submerged but one end was supported by the
bridge structure. There were many fatalities in the first car, especially on the lower level.
The second passenger coach was completely submerged in the bayou. Most of the 42
passenger fatalities occurred in this car.
The third coach was left dangling off of the end of the bridge. The rest of the train
consisted of a lounge car, diner, and sleeper. These cars all derailed but remained on the
bridge or the right of way.
The steel girder span across the bayou was destroyed in the accident. A towboat had
struck the bridge and knocked it out of alignment.
The pilot took a wrong turn in the fog and struck the fixed bridge at Big Bayou Canot
displacing the bridge by 38 inches. The lead locomotive of the Sunset Limited struck the
displaced bridge at about 116 km/h - flew through the air - and embedded itself 50 feet
into the mud.
The National Transportation Safety Board conducted an accident investigation and

addressed several safety issues including towboat operator training and evaluation,
bridge risk assessment, bridge identification, emergency response and evacuation
procedures, and event recorder crashworthiness.
Page 87 of 92

Appendix

G1. Accident 5 - Potters Bar, United Kingdom
Details of the Incident

On Friday 10 May 2002, a train from London
Kings Cross to Kings Lynn derailed at high
speed on the points south of Potters Bar
station. Seven people lost their lives; over 70
people were injured, some seriously.
The train was a four-coach electric multiple

unit, powered through the first and last
coaches. At the time of the incident it was
travelling at about 156 km/h.
The rear part of the train derailed at points

numbered 2182A located about 160 metres
south of Potters Bar. The rear bogie of the third
coach and both bogies of the fourth coach
derailed. The fourth coach slewed sideways so
that it was moving broadside to the direction of
travel, skidded along the track and detached
from the rest of the train. As the fourth coach
struck Darkes Lane bridge, the rear bogie was
torn off and the coach became airborne. The
coach came to rest on its side almost
perpendicular to the running lines, straddled
across both island platforms and wedged under
the station canopy. The coach rolled nearly 360
degrees at least once before coming to a stop. A
small waiting shelter on the up platform was
partially demolished as the coach skidded
towards the canopies. Six passengers travelling
in the derailed coach were killed. A number of
passengers sustained varying degrees of
injuries.
The leading three coaches, with the trailing bogie of the third coach derailed, continued
travelling along the Down Fast line passing between the platforms at Potters Bar station.
The emergency braking system had been initiated and the coaches came to a halt with
the front coach about 400 metres from the northern end of the station platform.
Page 88 of 92

The rear bogie of the fourth coach caused considerable damage to the bridge that passes
over the main street (Darkes Lane) of Potters Bar. Debris from the bridge structure and
the underside of the rear coach fell on pedestrians and cars below, killing one person.
The following hyperlink provides an anmimation of the crash

http://www.guardian.co.uk/flash/0,5860,713560,00.html
Derailment mechanism
The lock stretcher bar of the points was subject to fatigue stresses and eventually failed
at one of its right-hand bolt holes, causing it to withdraw from its insulating jacket as the
train passed over the points, and allowing the switch rail to which it was attached at this
side to spring out against the right-hand stock rail. This resulted in the right-hand
switch rail being set for the turnout route, with the left-hand switch rail already set for
forward running.
This happened when the rear wheels of the third coach were travelling over the points.
The wheels on each axle were then forced in two opposing directions, derailing the rear
of the third coach and the fourth coach entirely. The rear of the fourth coach re-railed
and took the turnout route towards the Down Slow line.
The fourth coach hit the Darkes Lane bridge parapets, detached from the rest of the train
and became airborne. The rear bogie of this coach was ripped off along with underbody
equipment, causing damage to the bridge and causing debris to fall through the gap
between the bridge parapets onto pedestrians and vehicles below. The fourth coach then
slid across the station platforms, struck a waiting room and rolled through 360 degrees,
eventually coming to rest wedged under the station canopy roofs.
The incident took place over 5 to 6 seconds.
G2. Non-High Speed Derailment 1 (track related) Southall East, United

Kingdom
On Sunday 24 November 2002, a high speed train (HST) made up of eight coaches and
two power cars, was passing over a set of points east of Southall Station at
approximately 193 km/h, the leading bogie of coach D (the fifth coach) derailed towards
the down main line. The train remained upright and in line, and finally came to a halt
just before West Ealing station, some 3 km further on. There were no serious injuries.
The bogie was derailed when the flange of a wheel struck one half of a broken fishplate
which had lodged in the nose of a cast manganese steel crossover forming part of the
points.
There is some evidence that the derailed bogie was constrained by a length of rail in the
six foot between the up and the down lines. Had the coach been able to move further
across towards the down fast line, or had more vehicles derailed, then the consequences
of a collision with a train on the adjacent line could have been catastrophic.
Page 89 of 92

G3. Non-High Speed Derailment 1 (track related)
A track related derailment occurred in New York State in 1994. Amtrak Train No. 49,
derailed on 3 August 1994, near Batavia, New York. A total of 14 cars including all of
the passenger occupied cars were derailed in the accident.
The accident sequence began about 5km before the general derailment. The train was
operating at about 127 km/h when one car in the train derailed one axle. The derailed
car was dragged the 5km to the site of the general derailment. The general derailment
occurred in the vicinity of a crossover where 13 additional cars in the train derailed. The
9 rear cars of the train separated from the front of the train, struck a signal bridge, and
went down an embankment on the south side of the track. Five of the nine cars turned
on their sides.
There were no fatalities in the accident but 45 passengers and crew were injured. The
cause of the initial derailment was a condition on the track known as a crushed head.
The material handling car derailed one axle at this point and bounced along the ties for
over 5 km before encountering the turnout that caused the general derailment.
The general derailment took place within sight of a dragging equipment detector that
would have detected the derailed axle and require the locomotive engineer to stop the
train.
G4. Non High Speed Derailment 2 (Special Trackwork)
An accident that took place in 1992 in Lugoff, South Carolina illustrates how important
maintenance of special track work can be. On 31 July 31 1991, Amtrak train 82, the
Silver Star, was en route from Tampa, Florida to New York City. The train consisted of
two diesel electric locomotives, 3 baggage cars, and 15 passenger cars. At 5:01 a.m., the
last six passenger cars derailed on the CSX Transportation main track at the Orlon
crossover in Lugoff, South Carolina. The train was travelling north on a straight track
with a clear signal at an authorized speed of 129km/h.
The accident occurred on a single main track parallel to an auxiliary track. The
derailment occurred at the crossover switch that connects the main track and the
auxiliary track, also known as the Dupont Siding. The last six passenger cars (13 to 18)
derailed moving left (westward) towards the siding. The cars collided with the first of
nine freight cars parked in the siding. The collision caused a hopper car to turn over and
a wheel set (an axle and a pair of wheels) to penetrate the west side of the last passenger
car. The derailed passenger cars came to rest 0.5 km north of crossover. They remained
upright and parallel to the track.
After the accident, the main track crossover was found to have the connecting rod
disconnected from the switch stand crank. The switch point was not secured to the
stock rail, the cross pin that attached the switch stand crank to its spindle was not in
Page 90 of 92

place, and the crank had dropped into the safety plate. The cross pin was found near the
switch stand.
A total of 22 crew and 407 passengers were on board the train at the time of the
derailment. There were 8 fatalities and 77 injuries resulting from the accident.
The cause of the derailment was that the poorly maintained switch opened underneath
the passing train because of inadequate track inspections and switch maintenance.
G5. Non-High Speed Derailment 3 (Vehicle defect)
A defective wheel was responsible for a derailment that occurred on a train near
Lakeland, Florida. On 13 January 13 1994, a witness observed the train go by and saw
two pieces of a wheel fly off a passenger car and land in nearby woods. A company
employee onboard the derailed car knew there was a problem and headed back through
the train to the trainmaster's car to have him stop the train. The train continued 4.3 km,
across five grade crossings, with the broken wheel. When the train reached the Park
Spur turnout, 15 additional passenger cars and three freight cars derailed. Of the 16
derailed passenger cars, five turned on their sides; the rest remained upright. There
were two fatalities and five injures.
The Lakeland, Florida accident train was remarkable in that it travelled for over 4.3 km
with a derailed wheel. The wheel bumped along the track and left marks on the sleepers
and on the grade crossings. The general derailment only occurred when the derailed
wheel encountered the turnout.
G6. Non-High Speed Derailment 4 (Operational)
On 17 December 1991, Amtrak Train 87, operating south on CSX Transportation Inc.
track, derailed on a curve in Palatka, Florida. Train 87 consisted of a locomotive and
eight passenger cars. The locomotive and first six cars derailed. The derailment
occurred while train 87 was negotiating a 6 degree 6 minute curve to the right (west).
The derailed equipment struck two homes and blocked the street north of the Palatka
station. Eleven passengers sustained serious injuries and 41 received minor injuries.
Five operating crewmembers and four on-board service personnel had minor injuries.
The cause of the accident was the failure of the operating crew to slow the train to
negotiate the 48 km/h curve. At the time of derailment in the curve, the train was
travelling at approximately 113 km/h.
G7. Non-High Speed Derailment 6 (Collisions)
One such serious derailment occurred at a grade crossing in Bourbonais, Illinois. On 15

March 1999, Amtrak Train 59, with 207 passengers and 21 employees on board and
operating on Illinois Central Railroad (IC) main line tracks, struck and destroyed the
loaded trailer of a tractor-semi trailer combination that was traversing the McKnight
Page 91 of 92

Road grade crossing in Bourbonnais, Illinois. Both locomotives and 11 of the 14 cars in
the Amtrak consist derailed. The derailed Amtrak cars struck two of ten freight cars
standing on an adjacent siding. The accident resulted in 11 deaths and 122 people being
transported to local hospitals.
The cause of the accident was the failure of the truck driver to yield the right of way to
the train at the grade crossing.
G8. Non-High Speed Derailment 6 (Environment)
A train carrying 105 passengers derailed when it ploughed into a landslide caused by
torrential rain. The locomotive was travelling at 8 km/h due to the danger of landslides
in the area, and the first carriage slewed off the track but stayed upright. Most of the
passengers managed to get off the train themselves. The driver saw the mound of mud
and chalk as the train emerged from a tunnel near Redhill, Surrey, England, but could
not prevent the locomotive hitting it. Nobody was seriously hurt.
The landslide was in a stretch of track, which engineers for the infrastructure operator,
Network Rail, had omitted from recent reinforcement work. A kilometre north of the
Merstham tunnel, the engineers had recently built a retaining wall alongside the track
after deeming the area to be a "weak point" vulnerable to possible landslides.
Page 92 of 92

Derailment Containment Provisions PDF

Uploaded by

Copyright:

Available Formats

Derailment Containment Provisions PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Derailment Containment Provisions PDF

Uploaded by

Copyright:

Available Formats

Ref: R00673

Report on the findings of:

Overview of Current Practice

• No single current practice to derailment containment provision,

Estimation of Effectiveness of DCP

There is insufficient information to make a quantified decision on the effectiveness of DCPs

2. Independently review the HSL-Zuid Transportation System safety management

REPORT REFERENCE: R00673

Table 1 Activities for Part A 8

REPORT REFERENCE: R00673

1.1 Background to the Assignment

The goal of Derailment Containment Provisions (DCP) is to reduce the probability of

• Follow the track;

REPORT REFERENCE: R00673

1.2 The Assignment

In managing the delivery of the HSL-Zuid Transportation System and the

Part A An Overview of current practice and existing knowledge. Requiring

Part B An estimation of effectiveness of whether DCP will work for trains

DCP Type 1 provisions mounted between the running rails,

Part C An Assessment of the adverse effect of DCP.

1.3 Methodology and Approach

REPORT REFERENCE: R00673

Part A Overview of current practice and existing knowledge

Table 1: Activities for Part A

REPORT REFERENCE: R00673

b) Reviewed the Questionnaires obtained from High Speed

B2. Reviewed the accident synopsis, which provided factual information on

Table 2: Activities for Part B prior to Expert Panel Session

c) What caused the derailment Escalation to Occur?

No – DCPs were not in place in the area of the derailment

Table 3: Activities for Part B during the Expert Panel Session

REPORT REFERENCE: R00673

Part C Assessment of potential adverse effects of DCP

The assessment considered direct impacts, such as dragging train equipment,

Table 4: Activities for Part C

REPORT REFERENCE: R00673

A1. Complied a list of

A6.& C5. Collated Information

B2. Expert panel reviewed accident scenarios

B3. Identified causal factors

B4. Analysed escalation

B7. Identified steps to prevent derailment

B8. Expert Panel identified derailment causal

B9. Identified risks introduced by DCP Reported on the

Figure 1 - Methodology Flow Chart

REPORT REFERENCE: R00673

1.4 Report Structure

This report is structured as follows:

REPORT REFERENCE: R00673

2.1 Part A: Overview of Current practices and existing

Railway / Organisation Contact

Table 5: Railways/Organisations for Face- to-Face Interviews

REPORT REFERENCE: R00673

Railway / Organisation Contact

Table 6: Railways/Organisations for Telephone Interviews

Whilst undertaking the interviews, some organisations were willing to provide an

The Eschede accident is a major incident in German rail history and is

REPORT REFERENCE: R00673

DCP Type 1 protection is used especially on high embankments and on