Rail Accident Report
Freight train derailment at London Gateway,
Essex
24 December 2021
Report 14/2023
December 2023
This investigation was carried out in accordance with:
• the Railway Safety Directive 2004/49/EC
• the Railways and Transport Safety Act 2003
• the Railways (Accident Investigation and Reporting) Regulations 2005.
© Crown copyright 2023
You may reuse this document/publication (not including departmental or agency logos) free of charge in
any format or medium. You must reuse it accurately and not in a misleading context. The material must
be acknowledged as Crown copyright and you must give the title of the source publication. Where we
have identified any third party copyright material you will need to obtain permission from the copyright
holders concerned. This document/publication is also available at www.gov.uk/raib.
Any enquiries about this publication should be sent to:
RAIB Email: enquiries@raib.gov.uk
The Wharf Telephone: 01332 253300
Stores Road Website: www.raib.gov.uk
Derby UK
DE21 4BA
This report is published by the Rail Accident Investigation Branch, Department for Transport.
Preface
Preface
The purpose of a Rail Accident Investigation Branch (RAIB) investigation is to
improve railway safety by preventing future railway accidents or by mitigating their
consequences. It is not the purpose of such an investigation to establish blame or
liability. Accordingly, it is inappropriate that RAIB reports should be used to assign
fault or blame, or determine liability, since neither the investigation nor the reporting
process has been undertaken for that purpose.
RAIB’s findings are based on its own evaluation of the evidence that was available at
the time of the investigation and are intended to explain what happened, and why, in a
fair and unbiased manner.
Where RAIB has described a factor as being linked to cause and the term is
unqualified, this means that RAIB has satisfied itself that the evidence supports both
the presence of the factor and its direct relevance to the causation of the accident or
incident that is being investigated. However, where RAIB is less confident about the
existence of a factor, or its role in the causation of the accident or incident, RAIB will
qualify its findings by use of words such as ‘probable’ or ‘possible’, as appropriate.
Where there is more than one potential explanation RAIB may describe one factor as
being ‘more’ or ‘less’ likely than the other.
In some cases factors are described as ‘underlying’. Such factors are also relevant
to the causation of the accident or incident but are associated with the underlying
management arrangements or organisational issues (such as working culture).
Where necessary, words such as ‘probable’ or ‘possible’ can also be used to qualify
‘underlying factor’.
Use of the word ‘probable’ means that, although it is considered highly likely that the
factor applied, some small element of uncertainty remains. Use of the word ‘possible’
means that, although there is some evidence that supports this factor, there remains a
more significant degree of uncertainty.
An ‘observation’ is a safety issue discovered as part of the investigation that is not
considered to be causal or underlying to the accident or incident being investigated,
but does deserve scrutiny because of a perceived potential for safety learning.
The above terms are intended to assist readers’ interpretation of the report, and to
provide suitable explanations where uncertainty remains. The report should therefore
be interpreted as the view of RAIB, expressed with the sole purpose of improving
railway safety.
Any information about casualties is based on figures provided to RAIB from various
sources. Considerations of personal privacy may mean that not all of the actual effects
of the event are recorded in the report. RAIB recognises that sudden unexpected
events can have both short- and long-term consequences for the physical and/ or
mental health of people who were involved, both directly and indirectly, in what
happened.
RAIB’s investigation (including its scope, methods, conclusions and recommendations)
is independent of any inquest or fatal accident inquiry, and all other investigations,
including those carried out by the safety authority, police or railway industry.
Report 14/2023 December 2023
London Gateway
This page is intentionally left blank
Report 14/2023 4 December 2023
London Gateway
Freight train derailment at London Gateway,
Essex, 24 December 2021
Contents
Preface3
Summary7
Introduction9
Definitions 9
The accident10
Summary of the accident 10
Context11
The sequence of events17
Background information19
Analysis34
Identification of the immediate cause 34
Identification of causal factors 37
Identification of underlying factors 47
Data uncertainties and assumptions 52
Previous occurrences of a similar character 53
Summary of conclusions 55
Immediate cause 55
Causal factors 55
Underlying factors 56
Actions reported as already taken or in progress relevant to this report 57
Recommendations and learning point58
Recommendations58
Learning point 60
Appendices61
Appendix A – Glossary of abbreviations and acronyms 61
Appendix B – Derivation of traction and brake torques 62
Appendix C – Simulation of longitudinal train dynamics 64
Appendix D – Wagon derailment simulations 69
Appendix E – Urgent operating notice concerning trains with Ecofret 2 wagons 71
Report 14/2023 5 December 2023
London Gateway
This page is intentionally left blank
Report 14/2023 6 December 2023
London Gateway
Summary
Summary
At around 05:45 hrs on 24 December 2021, 5 wagons of a 33-wagon intermodal
freight train derailed at low speed as the train was entering the rail terminal at London
Gateway port, Essex. The derailment started when wheels on a wagon in the middle
part of the train suddenly lifted off the track, just before reaching the port boundary,
with the other wagons becoming derailed as they passed over points within the rail
terminal. While no one was injured, infrastructure damage disrupted rail freight access
into and out of the port for 14 days.
The first derailed wagon was the unloaded centre wagon of a triple-wagon, which was
made up of three individual wagons permanently coupled together. The wagons on
the train either side of the unloaded centre wagon were all loaded, and a train brake
application had recently been made when the derailment occurred. A longitudinal train
dynamic simulation model, specially developed as part of the investigation, showed
that compressive forces generated in the train were larger towards the middle of the
train, where the triple-wagon was located, and rapidly increased to a maximum as
the centre wagon approached the place where it became derailed. Data from the
train’s on-train data recorder showed that this was shortly after brake forces acting
on the locomotive had peaked and that the train’s brakes would have been starting to
release.
Derailments in 2015 involving an earlier version of the type of triple-wagon involved
demonstrated that triple-wagons of this design configuration were susceptible to
derailment when large longitudinal compressive forces are suddenly applied. Early
derailment simulation studies had confirmed this and helped inform a decision to
lengthen the bar coupler on the earlier version of these wagons to improve derailment
resistance. The wagon designers identified the need to make other design changes
when the new version of the triple-wagon was developed for unrelated reasons.
However, although the risks due to longitudinal compressive force continued to be
recognised, the need to make further design improvements to address them was
not deemed to be necessary. Derailment simulations undertaken in support of the
investigation helped show that the new version of the triple-wagon remained similarly
susceptible to derailment and confirmed that the lack of payload on the centre wagon,
and possibly in-service degradation of the bogie, increased the derailment risk.
The investigation found that the risks associated with the longitudinal dynamic
behaviour of long freight trains are not well understood in the rail industry and that
there are limited processes, tools, and knowledge resources available to assess
and manage them. This was identified as an underlying factor. The investigation
has identified the design management arrangements that were adopted during the
development of the new version of the triple-wagon as another underlying factor. It
is possible that these arrangements limited the designers’ ability to understand the
dynamic behaviour of the triple-wagon and did not result in the identification of critical
performance requirements.
Report 14/2023 7 December 2023
London Gateway
RAIB has made three recommendations. One is directed to VTG Rail, the owner
Summary
and developer of the triple-wagon, covering the need to identify and implement any
necessary design changes and operating restrictions. The second is directed to GB
Railfreight, the operator of the train, covering the need to develop instructions and
best practice guidance for its operations staff. The third is directed to RSSB covering
the need to raise the rail industry’s understanding of the derailment risks associated
with longitudinal compressive forces and the management strategies that need to be
introduced.
Report 14/2023 8 December 2023
London Gateway
Introduction
Introduction
Definitions
1 Metric units are used in this report, except when it is normal railway practice to
give speeds and locations in imperial units. Where appropriate the equivalent
metric value is also given.
2 All mileages are taken from a datum at London Fenchurch Street station. Left and
right orientations relate to the direction of travel of the train when the accident
occurred.
3 The term ‘rail industry’ refers to the various organisations and parties, each with
individual responsibilities and duties, that co-operate to deliver rail services on the
mainline railway in Great Britain.
4 The term ‘wagon’ refers to an item of freight rolling stock that is braked but
unpowered. A freight train comprises a number of wagons that are coupled
together and hauled by a locomotive. The term ‘triple-wagon’ refers to three
wagons that are permanently coupled together, while ‘twin-wagon’ refers to
two permanently coupled wagons. The term ‘single wagon’ is used to refer to a
wagon that is not permanently coupled to another wagon. The term ‘vehicle’ is
used to refer to an item of rolling stock in general, in this case either a wagon or a
locomotive.
5 The term ‘coupler’ refers to any arrangement of equipment designed to connect
two vehicles, thereby controlling the degree to which they can separate or draw
closer together.
6 The report refers to ‘Ecofret 1’ and ‘Ecofret 2’ wagons. These are variants of the
FWA type of intermodal wagon (see paragraphs 47 to 58).
7 The report contains abbreviations. These are explained in appendix A.
Report 14/2023 9 December 2023
London Gateway
The accident
The accident
Summary of the accident
8 At around 05:45 hrs on 24 December 2021, five wagons within freight train
reporting number 4L471 derailed on the Thames Haven line on the approach to
London Gateway port, near to Stanford-le-Hope, Essex (figure 1). Train 4L47
was the scheduled GB Railfreight (GBRf) 22:35 hrs (on 23 December) intermodal
service from the freight terminal at Hams Hall, near Birmingham, to London
Gateway port. The service was carrying freight containers.
9 Train 4L47 had been routed off the Down Tilbury line at Thames Haven Junction.
The train had just started to enter the port infrastructure when the derailment
occurred. The 12th wagon derailed to the left on a section of line without points or
crossings (termed as ‘plain line’) approaching LG11 signal, just before reaching
the boundary between Network Rail’s infrastructure and that managed by the port
operator (figure 2). The locomotive’s on-train data recorder (OTDR) registered
that the train brakes were being applied at the time of the derailment and that the
train was rapidly slowing to a speed of 9 mph (14.5 km/h). The 11th, 13th, 14th and
15th wagons subsequently derailed as a consequence.
10 Realising something was wrong, the driver brought the train to a stand on the
Port Arrivals line, 740 metres beyond LG11 signal. The train remained coupled
together during the derailment, with all the wagons staying upright. Figure 3
shows the position of the train after the derailment.
Location of accident
© Crown Copyright. All rights reserved. Department for Transport 100039241. RAIB 2023
Figure 1: Extract from Ordnance Survey map showing the location of the accident at London Gateway.
1
An alphanumeric code, known as the ‘train reporting number’, is allocated to every train operating on Network
Rail infrastructure.
Report 14/2023 10 December 2023
London Gateway
The accident
To Pitsea and Not to scale
Shoeburyness
Some lines omitted for clarity
Stanford-le-Hope Infrastructure boundary
station Network Rail DP World
The Bridleway
Path of train 4L47 Camera and
level crossing
scanning
Extent of derailment To rail
equipment
Point of derailment terminal
Down
LG11 sidings
T hames
Haven
line
Up Th
ames H LG1 LG2 Thames Haven Single
line
aven lin
e Port Arrivals line
Port Departures line
LG3
Thames Haven
Junction London
Gateway Port
ne
Summit
li
Initial
line
bury
y
n Til
ilbur
peak
Dow
Up T
To
1 in 328
1 in 374
1 in 141
1 in 181
1 in 108
London
LG11
Figure 2: Track layout on Thames Haven line showing the gradient profile and the route of the train.
11 No one was injured in the accident, although there was major damage caused
to the railway infrastructure and six wagons required repair. The derailment also
resulted in major disruption to rail freight access into the port.
Context
Location
12 London Gateway port is a deepwater container terminal on the Thames, located
around 40 km east of central London. Constructed on the site of a former oil
refinery, it opened for operation in 2013. The Thames Haven line connects the
mainline railway to the rail terminal at the port. The line and associated railway
infrastructure were redesigned and renewed as part of the construction work for
the new port. The new rail terminal is capable of handling trains up to 775 metres
long.
Report 14/2023 11 December 2023
London Gateway
The accident
Down Thames Haven line
Wagon 13
Wagon 12 (hidden)
Wagon 11
Direction
of travel
Port
Arrivals
line
Port
Thames Haven
Departures
Single line
line
Figure 3: Position of the train after the derailment (image courtesy of British Transport Police).
13 Trains join the Thames Haven line at Thames Haven Junction (26 miles
41 chains) on the London, Tilbury, and Southend line. It initially comprises two
lines: the Down Thames Haven, used by trains entering the port, and the Up
Thames Haven, used by departing trains. Trains entering the port pass LG11
signal, then a user worked level crossing (The Bridleway) (28 miles 7 chains) and,
shortly afterwards, the infrastructure boundary (28 miles 9 chains) between the
mainline railway and the port. The two lines then join to form the Thames Haven
Single line at LG1 points. The single line continues under the overbridge carrying
the port access road, after which LG2 and LG3 points provide access to the Port
Arrivals and Port Departures lines and the rail terminal sidings (figure 2).
Report 14/2023 12 December 2023
London Gateway
14 Trains routed towards London Gateway port enter the Down Thames Haven line
The accident
on a right-hand curve and climbing gradient. After reaching an initial peak, the
curvature on the line eases to the left and the railway continues to climb, reaching
a final summit at 27 miles 45 chains, just over one mile (1.7 km) beyond the
junction. The line then descends on a gradient of 1 in 108 and becomes level
shortly before LG11 signal. The gradient profile is shown at figure 2.
15 Various maximum permitted line speeds apply for trains routed into the port rail
terminal. These are:
• a 25 mph (40 km/h) limit from Thames Haven Junction on the Down Thames
Haven line to 28 miles 33 chains on the Thames Haven Single line, shortly after
the overbridge
• a 15 mph (24 km/h) limit from 28 miles 33 chains to 29 miles 74 chains on the
Thames Haven Single line, and also on the Port Arrivals and Port Departures
lines
• a 5 mph (8 km/h) limit beyond 29 mile 74 chains on the Thames Haven Single
line.
16 Train movements outside the port rail terminal are controlled by lineside signals
operated from Network Rail’s Upminster signalling control centre. LG11 signal
controls access into the rail terminal. LG11 signal, and the signals used to control
movements within the rail terminal, are operated from the London Gateway rail
control centre.
Organisations involved
17 Network Rail owns, manages, and maintains the railway infrastructure on the
approach to the infrastructure boundary where the 12th wagon (wagon 122)
derailed. This section of line lies within Network Rail’s Anglia route, which is part
of its Eastern region.
18 DP World owns, manages, and maintains the railway infrastructure within London
Gateway port (beyond the infrastructure boundary) where the four other wagons
(wagons 11, 13, 14 and 15) consequently derailed and train 4L47 came to a
stand.
19 GBRf operated train 4L47 and employed its driver. It leased wagon 12, an FWA
type wagon, and the 11 other FWA type wagons on the train. It also either leased
or owned the other vehicles on the train, including the locomotive, and was
responsible for the maintenance for most of them. DP World contracted GBRf to
operate the port rail terminal and to provide the ground staff required.
20 VTG Rail owned and maintained the FWA wagons on the train, including
wagon 12, and leased them to GBRf. All the FWA wagons on the train were of
the Ecofret 2 variant, a wagon of recent design. VTG Rail specified the Ecofret 2
design requirements and led the development, procurement and approval work
needed to obtain authorisation for placing the wagons into service. VTG Rail
appointed consultants to assist it with the work.
2
Wagons are referred to in the report by their position along the train.
Report 14/2023 13 December 2023
London Gateway
21 Wabtec UK Ltd3 (referred to as ‘Wabtec’ in the remainder of this report) designed,
The accident
manufactured, and supplied the bogies that VTG Rail required be fitted to the
Ecofret 2 wagons. It supported VTG Rail with the approval work needed to obtain
authorisation for placing Ecofret 2 wagons into service.
22 WH Davis designed and manufactured the Ecofret 2 wagon underframe. It was
based on the design of the Ecofret 1 wagon, an earlier FWA wagon variant, the
design rights for which VTG Rail already owned. WH Davis also assembled
the wagons and supported VTG Rail with the approval work needed to obtain
authorisation for placing Ecofret 2 wagons into service.
23 VTG Rail appointed AEGIS Certification Services Limited (ACS) to undertake
conformity assessments for the Ecofret 2 wagon against applicable technical
standards and safety requirements. This was part of the approval work needed to
obtain authorisation for placing Ecofret 2 wagons into service.
24 The Rail Safety and Standards Board (RSSB) is the organisation that supports
the rail industry in improving its safety performance. Its principal activities include:
managing research and development programmes; developing, agreeing and
maintaining standards; and providing insight, analysis and guidance on railway
systems, safety and health.
25 Network Rail, DP World, GBRf, VTG Rail, Wabtec, WH Davis, ACS and RSSB
freely co-operated with the investigation.
Train involved
26 Train 4L47 was formed of a class 66 diesel-electric locomotive and a mixture of
33 intermodal flat wagons. These were:
• four Ecofret 2 triple-wagons, each comprising three wagons permanently
coupled with a bar coupler between each wagon (12 wagons altogether)
• six FEA type twin-wagons (a mixture of FEA-A, FEA-B and FEA-C types), each
comprising a pair of wagons permanently coupled with a bar coupler between
them (12 wagons altogether)
• nine FEA-S type single wagons.
Each triple-wagon, twin-wagon, single wagon and the locomotive were coupled to
each other using a conventional screw coupling and side buffer arrangement.
27 Wagon 12, the first wagon to derail, was the centre wagon of an Ecofret 2
triple- wagon. As such, it was permanently coupled (by a bar coupler) to two outer
wagons. These were wagon 11 in front, and wagon 13 behind. Except for wagon
12, all the wagons on the train carried at least one freight container. Table 1 is
a summary of the consist of train 4L47 and the payload recorded on the train
document. The train had an approximate length of 621 metres.
3
Previously known as Wabtec Rail Ltd.
Report 14/2023 14 December 2023
London Gateway
Connection to
The accident
Vehicle Type Number Payload4 Outcome
next vehicle
Locomotive Class 66 66779 - SC+B
Wagon 1 FEA-C 650026 Two containers (29.3t + 30.0t) BC
Wagon 2 FEA-C 650025 Two containers (29.7t + 19.4t) SC+B
Wagon 3 FEA-S 640632 Two containers (31.9t + 29.8t) SC+B
Wagon 4 FWA-Ecofret 2 7045202301 One container (20.7t) BC
Wagon 5 FWA-Ecofret 2 7045210510 Two containers (30.1t + 26.6t) BC
Wagon 6 FWA-Ecofret 2 7045202319 One container (12.0t) SC+B
Wagon 7 FEA-A 630037 Two containers (16.0t + 29.8t) BC
Wagon 8 FEA-A 630038 Two containers (13.8t + 27.2t) SC+B
Wagon 9 FEA-S 640666 Two containers (4.4t + 29.3t) SC+B
Wagon 10 FEA-S 640909 Two containers (27.4t + 29.7t) SC+B
Wagon 11 FWA-Ecofret 2 7045202574 One container (18.9t) BC Derailed later
Wagon 12 FWA-Ecofret 2 7045210643 No container BC First to derail
Wagon 13 FWA-Ecofret 2 7045202566 One container (20.0t) SC+B Derailed later
Wagon 14 FEA-B 643020 Two containers (29.3t + 10.0t) BC Derailed later
Wagon 15 FEA-B 643019 One container (29.9t) SC+B Derailed later
Wagon 16 FWA-Ecofret 2 7045202335 One container (29.4t) BC
Wagon 17 FWA-Ecofret 2 7045210528 One container (28.4t) BC
Wagon 18 FWA-Ecofret 2 7045202327 One container (29.4t) SC+B
Wagon 19 FEA-A 630019 One container (14.2t) BC
Wagon 20 FEA-A 630020 Two containers (43.4t + 28.4t) SC+B
Wagon 21 FEA-S 640915 Two containers (18.6t + 25.6t) SC+B
Wagon 22 FEA-C 650032 Two containers (29.4t + 30.4t) BC
Wagon 23 FEA-C 650031 One container (30.4t) SC+B
Wagon 24 FEA-S 640686 Two containers (28.4t + 29.3t) SC+B
Wagon 25 FEA-S 640660 Two containers (17.9t + 29.8t) SC+B
Wagon 26 FEA-B 640621 Two containers (27.8t + 13.8t) BC
Wagon 27 FEA-B 640622 Two containers (29.2t + 29.3t) SC+B
Wagon 28 FEA-S 640642 Two containers (29.3t + 29.2t) SC+B
Wagon 29 FWA-Ecofret 2 7045202111 One container (4.4t) BC
Wagon 30 FWA-Ecofret 2 7045210411 Two containers (29.0t + 29.8t) BC
Wagon 31 FWA-Ecofret 2 7045202103 One container (11.0t) SC+B
Wagon 32 FEA-S 640662 Two containers (26.4t + 28.0t SC+B
Wagon 33 FEA-S 640691 Two containers (29.3t + 30.2t) -
SC+B Screw coupling plus side buffers
BC Bar coupler
Wagon type:
Single wagon
Twin-wagon
Triple-wagon
Table 1: Train 4L47 consist and recorded payload. 4
4
Payload in tonnes (t). Containers on the derailed wagons were examined and weighed after the derailment. The
measurements suggested that the actual train payload may have differed from that recorded on the train document.
RAIB has taken account of this in the supporting wagon derailment simulation work (appendix D).
Report 14/2023 15 December 2023
London Gateway
Railway and port infrastructure
The accident
28 The track on the Thames Haven line was laid for the port’s opening in 2013. It
comprises CEN60 rail5 supported on concrete sleepers and stone ballast. The
rails are seated on polymer pads and secured to the sleepers with steel rail
fastenings.
29 Network Rail’s track geometry measurement train ran over the Down Thames
Haven line on 13 July 2021. This was the last recording run before the derailment.
No track geometry faults were recorded in the vicinity of the derailment that
required intervention6 before train 4L47 ran over the line on 24 December 2021.
RAIB compared data from the recording run with track gauge and cross level
measurements that Network Rail made immediately after the derailment. This
indicated that the track had not significantly deteriorated in the meantime.
30 On entry to the port rail terminal, trains pass a set of lineside cameras and optical
character recognition (OCR) scanners. These collect information about the train
and the container payload (figure 4). The equipment is located just after the port
access road overbridge (figure 2). It includes lights to illuminate the train as it
passes.
Figure 4: Camera and scanning equipment on
entry to the port (diagram courtesy of DP World).
Staff involved
31 The driver of train 4L47 joined GBRf in January 2019 with no prior experience as
a train driver. They were assessed as competent to operate class 66 locomotives
in October 2019 and assessed as competent to operate trains over the route into
London Gateway port in January 2021. They had regularly driven trains into and
out of the port since then. The driver had no previous safety incidents recorded by
GBRf.
External circumstances
32 The nearby weather station in Corringham (2.5 km to the north-east) recorded an
air temperature of 7.7°C at the time of the derailment and a wind speed of 1 km/h.
There was no report of precipitation. It was dark when the derailment occurred.
5
A type of flat-bottom rail having a weight of 60 kilograms per metre.
6
According to Network Rail company standard NR/L2/TRK/001 ‘Inspection and maintenance of permanent way’.
Report 14/2023 16 December 2023
London Gateway
The sequence of events
The sequence of events
Events preceding the accident
33 Train 4L47 departed from Hams Hall on time at 22:35 hrs on 23 December 2021.
The train was routed via Whitacre Junction, Rugby and Watford Junction and
arrived at Wembley European Freight Operating Centre (EFOC) at 00:52 hrs on
24 December where there was a change of drivers.
34 After preparing the locomotive and checking the handbrakes on the train were
released, the driver who operated train 4L47 from Wembley EFOC to London
Gateway port was ready to depart. GBRf drivers needed to know if the train
included Ecofret 2 wagons because this required them to apply the brake in a
specific manner (see paragraph 138a). GBRf stated that drivers would know if
there were Ecofret 2 wagons as this was detailed on the train document. The
driver who operated train 4L47 to London Gateway, however, stated that they
used the running brake test7 (required to take place at the start of the journey) to
determine if there were Ecofret 2 wagons, since it was their experience that trains
containing this type of wagon responded differently during these tests.
35 The train departed from Wembley EFOC at around 03:45 hrs, three minutes early.
It was routed along the North London line to Stratford and then onto the London
Tilbury and Southend line at Barking. The driver recalled that the train felt “heavy”,
and that the journey had been slow because of an unusually large number of
restrictive signals, but that nothing untoward occurred during this part of it. Train
4L47 passed over Thames Haven Junction and onto the Down Thames Haven
line at 05:22 hrs, six minutes later than scheduled.
Events during the accident
36 Figure 5 is an analysis of the locomotive OTDR recording of the journey along the
Down Thames Haven line, starting at Thames Haven Junction. It shows:
• the throttle number (between 1 and 8) that was selected by the driver to control
traction effort
• the train brake pipe pressure (which is measured at the locomotive) and the
corresponding locomotive brake cylinder pressure generated in response to
braking changes made by the driver
• the gradient profile (track elevation)
• the train’s speed and the maximum permitted line speed at the location
• the location of important infrastructure features and of train 4L47 at the time of
the derailment, including the position of wagon 12 (leading bogie).
7
Running brake tests are required to test the effectiveness of the train’s brakes in accordance with the Rule Book
GERT8000 Module TW1 ‘Preparation and movement of trains’ (issue 17, September 2021).
Report 14/2023 17 December 2023
London Gateway
LG11 signal
The sequence of events
Summit Scanners
Point of and camera
30
derailment equipment
25
20
15
10
5
0
0 500 1000 1500 2000 2500 3000 3500
Distance from Thames
Haven Junction (metres) Single brake Double brake
applications application
Speed (mph) Permanent speed restriction (mph) Brake pipe pressure (bar)
Locomotive brake cylinder pressure (bar) Traction (throtle number) Track elevation (metres)
Train 4L47 Wagon 12 (lead bogie)
Figure 5: Analysis of the locomotive OTDR for the journey along the Down Thames Haven line.
37 Train 4L47 passed over Thames Haven Junction at around 22 mph (36 km/h).
The locomotive OTDR recorded a mixture of traction effort demands and single
brake applications as it continued to the summit and then descended the 1 in
108 gradient to LG11 signal. The driver recalled the signal clearing as the train
approached. The OTDR recorded two consecutive brake applications (a ‘double
brake application’) shortly after the front of the train passed the signal. At this
time, wagon 12 was approaching the point of derailment. Together with the
derailment marks and other evidence (see paragraphs 42-44), the OTDR analysis
shows that the leading bogie on wagon 12 derailed as the brakes were releasing
after the second of the two brake applications.
38 The driver recalled a sudden jolt and, looking back, noticed dust coming from the
train. On realising something was wrong, they brought the train to a stand.
Events following the accident
39 The driver secured the train and walked back to confirm what had happened.
On returning to the cab, they made a GSM-R call to the signaller to report the
derailment. The signaller notified the London Gateway rail control centre. A
member of ground staff arrived, and both they and the driver acted to protect the
train.
40 After the accident, the driver was tested for the presence of both non-medical
drugs and alcohol at London Gateway port. The driver tested clear for both.
41 It was 14 days before temporary repairs could be made to the railway and the
rail connection into the London Gateway port reopened. An assessment of final
infrastructure repairs identified the need for extensive renewal work including the
replacement of nearly 700 metres of track and the installation of a complete new
set of points.
Report 14/2023 18 December 2023
London Gateway
Background information
Background information
Examination of the track and derailment site
42 Marks and damage on the track were found on the approach to LG11 signal. They
were consistent with:
• both left-hand wheels on a single bogie lifting up onto the rail
• the left and right-hand wheels of both wheelsets then becoming airborne for
around two metres
• the right-hand wheels then landing on the sleepers between the rails (the
‘four- foot’) and the left-hand wheels landing in the track bed alongside the
railway (the ‘cess’).
43 The first identified derailment mark was on the left-hand rail, approximately
11 metres on the approach to LG11 signal. This was designated the point of
derailment and numbered as sleeper 0.8 The second derailment mark was
identified at sleeper -3, around 2 metres beyond the point of derailment (figure 6).
The separation of the marks was consistent with the 1.8 metre wheelbase of the
TF20 bogies fitted to wagon 12 (see paragraphs 49 and 58). This, and the form
of the marks (short, and at an angle to the rail), shows that both wheelsets on the
same bogie had lifted suddenly, and simultaneously left the track on a trajectory
towards the left.
First landing marks
Sleeper -3
~2 metres
Direction
of travel
Point of derailment
(sleeper 0)
Figure 6: Derailment marks and track damage in the vicinity of the designated point of derailment
(photographs courtesy of Network Rail).
8
The sleepers were sequentially numbered as part of the track examination, with sleeper 0 being the nominal point
of derailment and negative numbers indicating sleepers beyond this point.
Report 14/2023 19 December 2023
London Gateway
44 Other marks were found further along the track. These took the form of damage
Background information
to sleepers in the four-foot and cess. There was an absence of marks between
sleeper 0 and sleeper -3, evidence of one wheelset running derailed between
sleeper -3 and sleeper -6 and, after this, evidence of two wheelsets running
derailed. This indicates that the two wheelsets on the bogie landed more or
less at the same time, the leading wheelset close to sleeper -6 and the trailing
wheelset close to sleeper -3. They had remained airborne in the meantime
(figure 7). There were no marks to suggest that more than one bogie ran derailed
on the Down Thames Haven line. RAIB has concluded, from other evidence, that
the wheelsets which derailed at this point were on the leading bogie of wagon 12
(see paragraph 66).
Landing mark -
LG11 signal leading wheelset
Sleeper -6
Landing mark -
trailing wheelset
Sleeper -3
Direction
of travel
Figure 7: Derailment marks and track damage on the approach to LG11 signal and The Bridleway level
crossing beyond (photograph courtesy of Network Rail).
45 Closed-circuit television (CCTV) images from the lineside cameras (paragraph 30)
showed that three bogies in total had derailed by the time wagon 12 was running
on the Thames Haven Single line (figure 8). These were the trailing bogie on
wagon 11 and both bogies on wagon 12. All three of these bogies were shown
running derailed to the left. RAIB concluded that the trailing bogie on wagon 12
and the trailing bogie on wagon 11 had been drawn into derailment due to the
path that the already derailed leading bogie of wagon 12 was forced to take as it
ran through LG1 points, where the Up Thames Haven and Down Thames Haven
lines join to form the single line (paragraph 13). Marks and damage on the track
at LG1 points support this.
Report 14/2023 20 December 2023
London Gateway
Background information
Wagon 12 -
leading bogie
Wagon 11 - Wagon 12 -
trailing bogie trailing bogie
Figure 8: CCTV images of train 4L47 passing the London Gateway port cameras and OCR scanners
(images courtesy of DP World).
46 When the derailed train reached LG2 points, the leading bogie on wagon 11 and
both bogies of wagon 13 also derailed (figure 9).
Wagon 13 Wagon 12 Wagon 11
Direction of travel
Figure 9: Final position of Wagons 11, 12 and 13 (photograph courtesy of VTG Rail).
Ecofret 2 wagons
47 Intermodal wagons operating on the mainline railway in Great Britain (GB)
have been historically designed to accommodate container payloads of up
to 60 feet (18.3 metres) in length. This means that they can typically carry a
40- foot container plus a 20-foot container, three 20-foot containers or two 30-foot
containers. Examples of such wagons include the FEA type wagons within train
4L47.
Report 14/2023 21 December 2023
London Gateway
48 With 40-foot containers becoming more popular (and 20 and 30-foot containers
Background information
becoming scarcer), freight train operators were finding many of their wagons
were carrying solitary 40-foot containers, resulting in wasted capacity. VTG Rail
identified a market for a shorter intermodal wagon that carried only a single
40- foot container.9 This led to it contracting an engineering consultant to design
the Ecofret 1 wagon. A prototype Ecofret 1 triple-wagon was in built in 2012, with
Ecofret 1 production versions being built both in triple-wagon and twin-wagon
configurations.
49 A continuing demand for intermodal wagons led to the development of the Ecofret
2 wagon, based on the earlier Ecofret 1 design. VTG Rail identified that the
following changes to the Ecofret 1 design were needed:
• A bogie that would enable compliance with new ride requirements.10 VTG Rail
opted to specify a bogie of the TF20 type that Wabtec was in the process of
developing.
• Brake equipment from a different supplier. The aim was to avoid problems that
had been experienced with the brake equipment fitted to Ecofret 1 wagons.
• Flexibility to carry one 40-foot or two 20-foot containers on any of the three
wagons (rather than only the centre wagon which Ecofret 1 allowed) of a
triple- wagon.
50 In other areas it was decided to keep the wagon designs as similar as possible.
This was to aid maintenance and spares management. The objective of
minimising change led to VTG Rail deciding to specify the same longer bar
coupler that had been retrofitted to Ecofret 1 triple-wagons during early service
(see paragraphs 76 and 77).
51 VTG Rail led a team to project manage the various development tasks and
engaged consultants to assist. The team included representatives from its
suppliers, Wabtec and WH Davis. The division of the engineering responsibility
can be summarised as below:
• VTG Rail – wagon technical specification and overall design, and the
management of the approval work needed
• WH Davis – modifications of the Ecofret 1 wagon underframe design including
incorporation of the new brake equipment, the revised container securing
arrangements and the interfaces for the TF20 bogie
• Wabtec – design and development of the TF20 bogie and work needed to
demonstrate compliance with wagon ride and running safety requirements that
were identified as relevant.
52 Figure 10 shows the three wagons making up a typical Ecofret 2 triple-wagon.
Each has a welded steel underframe comprising a central longitudinal rectangular
box (the ‘underframe spine’) from which the bolster, headstock and other
fabrications provide the interfaces for the bogies, couplers and buffers, brake
equipment and the spigots needed for securing the containers.
9
The ability to carry two 20-foot containers was also required, but this flexibility was only provided on the centre
wagon of an Ecofret 1 triple-wagon.
10
VTG Rail identified Railway Group standard GM/RT2141 issue 4 as an applicable requirement. This introduced
the need to demonstrate that vehicles are not at risk of derailment due to repeated vertical track geometry
irregularities (also known as ‘cyclic top’).
Report 14/2023 22 December 2023
London Gateway
Background information
Bolster
Underframe spine
Spigot
Brake equipment
Outer headstock Inner headstocks Outer headstock
Outer wagon Centre wagon Outer wagon
Figure 10: An Ecofret 2 triple-wagon.
53 The bar couplers that connect Ecofret 2 triple-wagons together (paragraph 26)
are anchored within a central pocket on the inner headstocks (the ‘headstock
pocket’). With reference to figure 11, the bar coupler arrangement comprises
a rigid drawbar11 (1) that connects to spring units (2), one on each wagon, via
spherical bearings (3). Each spring unit is made up of three elastomeric ring
elements. These are preloaded by being clamped between two steel plates
(4), which locate and secure each spring unit within the headstock pocket (5).
The spring units enable the drawbar couplings to compress (or extend) when
the externally applied longitudinal force exceeds the preload amount in either
compression (causing the coupled wagons to draw closer together) or tension
(causing the coupled wagons to separate). The material used in the spring
components enables the bar coupler arrangement to absorb energy when
alternating forces apply.
54 The spherical bearings allow the rigid drawbars to pivot in both the vertical
longitudinal (‘pitch’ rotation) and horizontal (‘yaw’ rotation) planes. Relatively large
angles of movement are possible. No mechanism or design feature is provided
to restrict these rotations before the rigid drawbar contacts the sides, or top and
bottom, of the headstock pocket.
55 The outer wagons are nearly a metre longer than the centre wagons. This is
to accommodate the conventional screw coupling and side buffer arrangement
needed to couple to other vehicles (paragraph 26). With reference to figure 12,
the arrangement comprises a conventional draw hook and screw coupling (6).
This connects to a drawgear spring unit (7) on the outer headstock via a clevis
and pin (8). The drawgear spring unit comprises two side-by-side stacks of
elastomeric ring elements. These are also preloaded. Conventional buffer units
(9) with internal spring elements (10) are bolted to the outer headstock, one on
either side.
11
1.6 metres long, as adopted for the lengthened Ecofret 1 bar coupler arrangement (see paragraphs 76 and 77).
Report 14/2023 23 December 2023
London Gateway
1.60 m
Background information
5.Headstock
2. Spring unit pocket
3. Spherical
bearings
1. Rigid
drawbar
4. Steel plates
Figure 11: Bar coupler arrangement at the inner headstock locations. Photograph shows the general
arrangement and the space available for pitch and yaw rotation within the headstock pocket.
9. Side buffer unit
7. Drawgear
spring unit
6. Drawhook
and screw
coupling
8. Clevis
and pin
10. Spring elements
Figure 12: Screw coupling and side buffer arrangement at outer headstock locations. Photograph shows
the general arrangement when coupled to another vehicle – a class 66 locomotive.
56 When coupled, the link on the screw coupling is placed on the draw hook of the
other vehicle (or vice versa) and the screw coupling tightened. If this results in
the buffer faces being in contact (without buffer unit compression),12 then any
longitudinal compressive force generated along the train will act only on the buffer
unit pairs, with the coupled vehicles coming closer together as the buffer units
deflect. Insufficient tightening of the screw coupling results in a gap between the
buffer faces. This is referred to as ‘slack’, which needs to be taken up (closed)
before the buffer units can touch and start to deflect.
12
The discussion on the behaviour of a screw coupling and side buffer arrangement relates to operation on
nominally straight track. It does not take into account effects such as vehicle yaw rotation that occurs on curved
track.
Report 14/2023 24 December 2023
London Gateway
57 With the buffer faces in contact (but without buffer unit compression), any
Background information
longitudinal tensile force generated acts only on the connected drawgear spring
unit pairs. In this case, the coupled vehicles will separate, but only once the
drawgear spring unit preload is overcome and the drawbar spring units can
start to deflect. If any slack was present (buffer faces not in contact), then this
would need to be taken up before this force can act. The material in the spring
components enables the arrangement to absorb energy when alternating forces
apply.
58 The TF20 bogie (figure 13) has a fabricated steel bogie frame supported by four
coil spring primary suspension units, one at each axlebox, with the axleboxes
being also connected to the bogie frame via a primary traction rod arrangement.
The bolster on the wagon underframe is supported by a pair of rubber secondary
suspension springs. Primary and secondary hydraulic dampers are fitted to help
control dynamic behaviour. The bogie is designed to run with the ‘P8’ type wheel
profile that was developed by British Rail.
Secondary Secondary
suspension spring suspension damper
Primary Primary Primary
suspension traction suspension
unit rods damper
Figure 13: TF20 bogie.
Ecofret 2 wagons - design approval and maintenance
59 In September 2019, VTG Rail drafted a document setting out its strategy for
obtaining authorisation from the Office of Rail and Road (ORR)13 for placing
Ecofret 2 wagons into service on the GB mainline railway. This included
demonstrating that the wagons complied with technical specifications for
interoperability (TSI) and national technical rules (NTR), as was then required by
The Railways (Interoperability) Regulations 2011.14 VTG appointed ACS to act as
both the Notified Body15 (NoBo) and Designated Body16 (DeBo) for the verification
work required.
13
The safety authority for railways in Great Britain.
14
https://www.legislation.gov.uk/uksi/2011/3066/contents/made.
15
A body which assesses and verifies the conformity of projects and systems to TSIs. TSIs have now been
replaced in the UK by National Technical Specification Notices (NTSNs). The assessment and verification of
conformity to NTSNs is by an Approved Body (ApBo). ACS also carried out this independent assessor role. https://
www.orr.gov.uk/sites/default/files/2021-07/guidance-on-approach-to-authorisation-under-the-interoperability-
regulations.pdf.
16
A body which assesses and verifies the conformity of projects and systems to NTRs.
Report 14/2023 25 December 2023
London Gateway
60 VTG Rail also recognised that the new Ecofret 2 wagons constituted a significant
Background information
change under the Common Safety Method on Risk Evaluation and Assessment17
(CSM REA), primarily due to the change of bogie. Because of this, they needed
to carry out a programme of hazard identification (recording the identified hazards
in a hazard log), risk evaluation and hence demonstrate that risks were managed.
VTG Rail appointed ACS to act as the Assessment Body (AsBo) to undertake the
required independent assessment of compliance with the CSM REA.
61 Ecofret 2 wagons were to be the first wagons to use the new TF20 bogie. Wabtec
sought to use this opportunity to gain approval for the bogie as an ‘interoperability
constituent’.18 The intent of this was to avoid extensive additional testing and
analysis work when supplying the bogie to other wagon manufacturers. Wabtec
carried out the approval work for the bogie, arranged for its assessment and
obtained approval certificates. These certificates were referenced by ACS as
evidence of compliance within the overall wagon approval documentation.
62 Wabtec used an industry-standard railway vehicle dynamics computer simulation
package to help assess compliance with requirements for dynamic behaviour
and safety against derailment on twisted track.19 This only required development
and validation of computer simulation models that represented Ecofret 2 wagons
as single vehicles. No need was identified to develop a simulation model
representing a coupled triple-wagon. Supporting analysis work identified a running
behaviour issue that led Wabtec to conclude that special devices were needed to
control primary lateral suspension movement. It was this that resulted in Wabtec
making a late modification to the TF20 bogie (see paragraph 67).20
63 On 30 March 2021, ORR issued an ‘authorisation to place into service’ for
the batch of Ecofret 2 wagons that included wagons 11, 12 and 13 of train
4L47. ORR can only issue this where an applicant has satisfied ORR that the
‘project sub-system’ (in this case the Ecofret 2 wagon) is technically compatible
with the railway on which it is to operate, that the various required tests have
been satisfactorily completed and that the relevant supporting information
and documents have been drawn up or prepared.21 The batch comprised 84
triple- wagons, or 252 wagons in total.
64 VTG Rail’s inspection and maintenance regime for Ecofret 2 wagons comprised
a weekly in-service inspection, which is essentially a visual examination of the
overall wagon condition, and a more detailed annual vehicle inspection and
brake test (VIBT). The triple-wagon comprising wagons 11, 12 and 13 entered
service on 19 April 2021. The first VIBT was not due until 19 April 2022, after the
derailment.
17
A regulation which describes how the safety impacts of significant changes are assessed. https://www.orr.gov.uk/
sites/default/files/2023-07/common-safety-method-for-risk-evaluation-and-assessment.pdf.
18
A component, group of components or assembly of equipment that is incorporated or intended to be
incorporated into a subsystem upon which the interoperability of the rail system depends directly or indirectly. The
concept of a constituent includes both tangible objects and intangible objects such as software.
19
Chapters 4, 5, 6.1 and 7 of BS EN14363:2016 ‘Railway applications – Testing and Simulation for the
acceptance of running characteristics or railway vehicles – Running Behaviour and stationary tests’.
20
Wabtec explained that when no payload was carried (‘tare condition’), these devices were not required at speeds
below 50mph (80 km/h).
21
‘Approach to authorisation under the Interoperability Regulations’, Guidance to Applicants, ORR, 2021.
Report 14/2023 26 December 2023
London Gateway
Examination of the derailed wagons
Background information
65 RAIB examined wagons 11, 12 and 13 (which together formed a Ecofret 2
triple-wagon). It recorded the condition of the suspension, wheels, bogie brake
equipment and couplers and found that the damage to the wheel treads and
flanges and the brake equipment was consistent with all the bogies of these
wagons having run derailed. RAIB also measured the wheel profiles and relevant
wheelset dimensions for use in subsequent wagon derailment simulation studies
(see paragraph 82c and appendix D).
66 RAIB identified witness marks on the headstock pockets where wagon 11 and
wagon 12 were coupled together (figure 14). The marks were consistent with
the leading end of wagon 12 running derailed to the left and the rigid drawbar
coming in to contact with the headstock pockets as a result. The marks were only
found on one side of the headstock pockets. From this, and the available CCTV
evidence (paragraph 45), RAIB has concluded that the leading bogie of wagon 12
was the first to derail.
Contact No Contact No
damage damage damage damage
on left on right
Wagon 11 inner headstock (trailing end) Wagon 12 inner headstock (leading end)
Figure 14: Witness marks on the headstock pockets on wagon 11 and wagon 12 after recovery (rigid
drawbar is shown disconnected from wagon 11, and after a weld repair had been done).
67 Special suspension movement control devices were added to the TF20 bogie as
a late modification (paragraph 62). Referred to as ‘primary lateral buffers’ (PLBs),
they comprise a sliding frame designed to capture and restrain a vertical metal
lug on top of each axlebox. The lug is informally known as the ‘shark fin’. With
reference to figure 15, the sliding frame (11) is secured to the underside of the
bogie frame, the shark fin passing through a rectangular void (12). Two friction
strips (13) attached to the sliding frame are designed to bear against the sides
of the shark fin. These allow the shark fin (and axlebox) to slide up and down.
The side-to-side (lateral) movement of the axlebox is restrained by two laminated
elastomeric springs (14).
68 RAIB identified signs of significant damage and degradation on several PLBs. On
wagon 12, nearly all the friction strips had detached and there was evidence that
some of the laminated elastomeric springs had split. Degradation of the PLBs on
Ecofret 2 wagons was a problem that existed before the derailment, and work
was being carried out to investigate it when the derailment occurred.
Report 14/2023 27 December 2023
London Gateway
Background information
13. Friction strips 11. Sliding frame Friction strips missing
14. Laminated 12. Rectangular void (to
elastomeric springs accommodate the shark fin) Shark fin on axlebox
Figure 15: Primary lateral buffer device (diagram courtesy of VTG Rail). Photograph shows the
condition of a PLB device as found on wagon 12.
Derailment risk due to longitudinal train dynamics
69 Design and operational issues associated with longitudinal train dynamics have
long been recognised as factors affecting derailment risk. British Rail’s internal
derailment investigation guidance from the 1980s22 provides useful historical
context. It mentions traction and braking effects, buffer interaction, and coupling
tightness as important influences.
70 RAIB has investigated one previous derailment event where longitudinal
train dynamic behaviour was identified as a factor, at Neville Hill depot on
13 November 2019 (RAIB report 13/2020). The accident involved a relatively
new Intercity Express Train which collided at low speed with the rear of another
train moving in the same direction. The longitudinal compressive forces that were
generated resulted in wheelsets on the Intercity Express Train derailing to the
side by up to 1.25 metres, on three separate vehicles.
71 Following discussions with RSSB, RAIB has concluded that the rail industry
maintains an awareness that such issues can affect the safe running of freight
trains. However, there has been little recent rail industry research aimed at
understanding the risks relevant to current freight operations, or developing
guidance, processes, or standards to manage them.
22
‘A guide to the investigation of derailments’, 5th edition, British Railways Board, 1985.
Report 14/2023 28 December 2023
London Gateway
72 The wagons involved in this derailment were of a type that is currently outside the
Background information
scope of European Standard BS EN15839:2012, ‘Railway Applications – Testing
for the acceptance of running characteristics of railway vehicles - Freight wagons -
Testing of running safety under longitudinal compressive forces’.23 This European
standard is identified in the TSI for freight wagons (WAG TSI) as an ‘additional
optional condition’ which defines an acceptance criteria for vehicles with standard
side buffers and screw couplings which are operated in trains where ‘high
longitudinal compressive forces may occur’. The standard cites S-shaped curves
(where side buffer contact could occur) to be the special concern and specifies
a test track comprising curves of 150 metre radius separated by a nominal
6-metre-long straight section. There are plans to revise this European standard to
incorporate wagons of the type that were first to derail at London Gateway (see
paragraph 152).
Previous derailments involving Ecofret 1 and Ecofret 2 wagons
73 Two derailments of Ecofret 1 triple-wagons occurred after they entered service, at
Felixstowe on 28 August 2015 and at Peterborough on 17 September 2015 (see
paragraphs 174 and 175). The two trains were of similar length, and both were
braking at the time. In each instance, the triple-wagons that derailed were in the
middle part of the train, with the centre wagons carrying no load. Nearly all the
other wagons on the trains were loaded. Neither derailment occurred on Network
Rail managed infrastructure.
74 GBRf and VTG Rail investigated both derailments and identified that high
longitudinal compressive forces were the likely cause. VTG Rail subsequently
commissioned a vehicle dynamics specialist to undertake detailed computer
simulations to understand the effect of longitudinal compressive forces on Ecofret
1 triple-wagons. It used the same vehicle dynamics computer simulation package
that was used in this investigation to evaluate the susceptibility of wagon 12 to
derailment by this means (see paragraph 82c). The simulations included the effect
of track geometry irregularities.
75 The centre wagon was found to be most prone to derailment by this means. VTG’s
specialist reported that an increasing longitudinal compressive force resulted in:
• the wagon underframes initially rotating when viewed from above (yaw rotation),
all in the same direction
• the flanges of the left-hand wheels on the leading bogies then contacting the rail,
with the flanges of the right-hand wheels on the trailing bogies doing the same
• the wheel flanges being further pressed into rail contact as the developed lateral
forces increased, and the wheelsets on the most prone bogie (in some cases
the leading bogie on the centre wagon, in others the trailing bogie) starting
to roll about the wheel flange-rail contact point with the wheel opposite to the
contacting wheel then lifting off the rail
• the whole bogie then finally lifting and derailing.
The longitudinal compressive forces were simulated by applying equal and
opposite end loads on the outer wagons.
23
BS EN 15839:2012 lists ‘permanently coupled units with a bar coupler between the vehicles’ as one of the types
of vehicles that are out of scope.
Report 14/2023 29 December 2023
London Gateway
76 VTG Rail’s engineering consultant (paragraph 48) later developed a simulation
Background information
model that was simpler, and which became referred to as the ‘quasi-static model’.
The consultant concluded that, for an Ecofret 1 triple-wagon, the quasi-static
model predicted similar behaviour to the earlier detailed computer simulations.
Other simulations using the quasi-static model suggested an Ecofret 1
triple- wagon was more susceptible to derailment than an Ecotfret 1 twin-wagon.
The quasi-static model was also used to evaluate options for improving the
derailment resistance. It included work supporting a decision to lengthen the bar
coupler.
77 The decision to change the bar couplers on Ecofret 1 triple-wagons was recorded
in engineering documentation issued in March 2017. VTG Rail explained that an
operating restriction was imposed until the modification work was complete. This
restriction was intended to prevent Ecofret 1 triple-wagons from running in certain
payload conditions identified as making derailment more likely. VTG Rail reported
that Ecofret 1 wagons have operated without incident since the modification was
implemented.
78 A collision involving a GBRf intermodal train and agricultural equipment at Kisby
user worked crossing on 19 August 2021 resulted in the derailment of an Ecofret
2 triple-wagon (RAIB Report 12/2022). The collision occurred on a section of
nominally straight track, with the Ecofret 2 triple-wagon being part way along the
train and none of the wagons carrying a container. The Ecofret 2 triple-wagon
derailed in the same manner as an Ecofret 1 triple-wagon was predicted to under
longitudinal compressive force (paragraph 75), with the wheels on the leading
bogie of the centre wagon derailing to the left and the wheels on the trailing bogie
derailing to the right (figure 16).
Outer wagon Centre wagon Outer wagon
Direction of travel
Figure 16: Ecofret 2 triple-wagon that derailed at Kisby on 19 August 2021.
Report 14/2023 30 December 2023
London Gateway
79 The damage to the front of the train at Kisby indicates that significant longitudinal
Background information
compressive forces were generated in the accident, just as they were during
the collision at Neville Hill depot (paragraph 70). Therefore, this damage, the
observed position of the wheels after derailment and the design similarities
between Ecofret 1 and Ecofret 2 triple-wagons have led RAIB to conclude
that the derailment of the wagon at Kisby was probably caused by longitudinal
compressive force.
Testing and analysis
80 The accident at Kisby demonstrated the probability of an Ecofret 2 triple- wagon
derailing in a similar manner to that seen at London Gateway because of
longitudinal compressive force. However, unlike Kisby, no collision occurred
at London Gateway, and therefore, for longitudinal compressive force to have
initiated this derailment, it needed to have been generated in another way.
Longitudinal train dynamic effects due to brake applications were identified
as the reason for the unsafe longitudinal compressive forces in the Ecofret
1 triple- wagon derailments at Felixstowe and Peterborough. Given the
brake application changes at the time of the derailment at London Gateway
(paragraph 37), this was identified by RAIB as a potential cause of the derailment
of wagon 12.
81 The simulation models developed by VTG Rail’s vehicle dynamics specialist
and its engineering consultant both predicted that longitudinal compressive
forces of around 400 kN24 would be sufficient to cause an Ecofret 1 triple- wagon
to derail with bar couplers of the original length (paragraphs 74 and 76). In
seeking to compare this with the longitudinal compressive forces likely to be
experienced in service, VTG Rail’s engineering consultant hypothesised a
worst-case force of around 500 kN during braking. However, this hypothesis
largely relied on assumptions and simplifications relating to retardation rates and
speed- dependent braking effects published in RSSB research.25 This meant that
the supporting analysis could not be extended and used to establish how the
actual brake applications, made around the time that train 4L47 derailed, related
to the nature and significance of the longitudinal compressive forces generated.
82 Understanding the nature of the actual braking forces acting along train 4L47 (and
the traction forces provided by the locomotive), their effect on the longitudinal train
dynamics, and ultimately the risk of derailment, became critical to determining the
likelihood of the derailment having been caused by this mechanism. Developing
this understanding involved:
a. Estimating the brake and traction torques acting on the wheelsets of the
locomotive and the individual wagons of train 4L47 and how these varied as
the train travelled along the Down Thames Haven line. These are referred to
in this investigation report as the ‘locomotive brake torque time history’, the
‘locomotive traction torque time history’ and the ‘wagon brake torque time
history’. Their derivation involved use of measurements made on the brake
system of representative test trains and data from the OTDR on train 4L47
recording how it was operated. Appendix B contains further explanation on
how the estimated torque time histories were derived, and the testing involved.
24
A force of 10 kN is approximately equal to the weight of a mass of one tonne.
25
RSSB report, T999-01 ‘Review of braking tables in RGS GK/RT0075 Lineside signal spacing and speed
signage’.
Report 14/2023 31 December 2023
London Gateway
b. Determining how the derived traction and brake torque time histories affected
Background information
the longitudinal compressive forces generated within train 4L47 as it travelled
along the Down Thames Haven line. This involved the development of a
computer model to simulate the longitudinal train dynamic behaviour of train
4L47. This was then used to undertake baseline simulations and sensitivity
studies to understand how longitudinal compressive force (determined from
the forces calculated in the couplers connecting the locomotive and wagons
together) was affected by factors including coupler characteristic and brake
application. Appendix C includes further explanation on how the computer
model was developed and the findings from the baseline simulations and
sensitivity studies.
c. Evaluating the susceptibility of wagon 12 to derailment under longitudinal
compressive force. This involved the development of a vehicle dynamics
simulation model representing the complete Ecofret 2 triple-wagon comprising
wagons 11,12 and 13, and use of this to understand the behaviour of wagon
12 when a longitudinal compressive force is suddenly applied, the derailment
sequence and the forces associated. Appendix D includes further explanation
on how the wagon derailment simulation model was developed and the
findings from the baseline simulations and sensitivity studies.
Train braking
83 All the wheels on train 4L47 were braked using a conventional single pipe
automatic air brake system. With reference to figure 17, the system comprises
a distributor valve (15) and an auxiliary air reservoir (16) on the locomotive and
each wagon. These are connected by a single brake pipe (17) that runs the length
of the train, with flexible hoses (18) provided between the individual vehicles.
Brakes blocks on each wheel (19) apply when the distributor valves feed air to the
brake cylinders (20) from the auxiliary air reservoirs.
84 The automatic air brake is controlled from the driver’s brake valve (21) in the
locomotive cab. The driver’s brake valve is used to vary the pressure in the
brake pipe, either by venting the brake pipe or by feeding it with air from the main
air reservoir (22). The brake distributor valves vary the pressure in the brake
cylinders (and therefore the force applied to the brake blocks) depending on the
brake pipe pressure. The brakes are released (no pressure in the brake cylinders)
when the brake pipe is fully charged (at nominally 5 bar). A full service brake
application is made when the brake pipe is vented to a pressure of approximately
3.35 bar. An emergency brake application is made when it is fully vented (zero
pressure). The automatic air brake system is charged from an air compressor on
the locomotive (23).
85 The brake pipe pressure signal needs to propagate along the length of the train
and command the individual distributor valves to vary the local brake cylinder
pressure. This means that on a long train it can take several seconds more for the
brakes to apply (or release) on a wagon at the rear of the train compared to the
front. This can lead to a significant variation in braking force along the train.
Report 14/2023 32 December 2023
London Gateway
86 Brake distributor valves have a set of chokes that can be set to either ‘passenger’
Background information
or ‘goods’ settings. In the passenger setting, the train can stop in a shorter
distance because the wagon brakes apply more quickly. In the goods setting a
slightly delayed brake application is made. This helps ensure that the brakes on
the wagons throughout a long freight train apply in a more consistent manner.
The brake distributor valves on train 4L47 were all in the passenger setting.
This was so the train was able to achieve the braking performance needed to be
compatible with its maximum operating speed, which was 75 mph (121 km/h),
and the signal spacing on the route.
Wagon Locomotive
21. Driver’s brake valve
22. Main air reservoir
23. Air compressor
17. Brake pipe
To next
wagon
15. Distributor 20. Brake
valve cylinder
18.
Flexible
Hose
19. Brake block 16. Auxilliary
and wheel air reservoir
Figure 17: Single pipe automatic air brake system (simplified).
Report 14/2023 33 December 2023
London Gateway
Analysis
Analysis
Identification of the immediate cause
87 Longitudinal compressive forces generated within train 4L47 during braking
were sufficient to cause the wheels on the leading bogie of wagon 12 to
suddenly lift and run derailed.
88 The witness marks on the track (paragraphs 42 to 44), CCTV images from the
port lineside cameras (paragraph 45), and the contact damage to headstock
pockets from the bar coupler between wagon 11 and wagon 12 (paragraph 66),
were all consistent with:
• the left-hand wheels on the leading bogie of wagon 12 rapidly and
simultaneously contacting the left-hand rail and then lifting over the railhead
• both bogie wheelsets then becoming airborne, landing to the left and running
derailed parallel to the track.
89 The derailments at Felixstowe, Peterborough and Kisby (paragraphs 73 and 78)
had significant similarities with the circumstances and outcome of the derailment
at London Gateway. All three involved FWA type triple-wagons (either Ecofret 1 or
Ecofret 2) that were in the middle part of the train, and the individual wagons were
connected using the same (or near identical) bar coupler arrangement. The first
wagons to derail were unloaded centre wagons and there is evidence suggesting
that the wheels derailed rapidly, probably by lifting,26 when significant longitudinal
compressive forces would have been suddenly generated (either during braking
or as the result of a collision). This supports the conclusion that wagon 12 was
susceptible to derailment by this means.
90 The wagon derailment simulations undertaken in support of the investigation of
this derailment (paragraph 82c and appendix D) predicted behaviour that was
consistent with the characteristics of these earlier derailments. In summary, and
with reference to figure 18, as a sustained longitudinal compressive force of
increasing magnitude was suddenly applied:
• The underframe of wagon 12 started to lift (stage A).
• The underframe of wagon 12 then started to rotate when viewed from above
(yaw rotation) (stage B).
• As the compressive force was increased further, the wheel flanges on the
leading bogie started to contact one of the rails, while on the trailing bogie the
wheel flanges started to contact the opposite rail (stage C).
• Wheels opposite to the wheel flanges in contact then started to unload,
eventually lifting off the rail (stage D).
• Finally, one of the bogies rose clear of the rails, derailing fully as a result (stage
E, not illustrated).
26
No wheel flange marks were identified on the railhead at Felixstowe. This suggested to investigators that the
wheels did not climb over the rail, but rather that they lifted.
Report 14/2023 34 December 2023
London Gateway
Stage A: Underframe lift
Analysis
Direction of travel
Lift Lift
Stage B: Underframe yaw rotation
Yaw rotation
Stage C: Wheel flanges in contact with opposite rails
Trailing bogie right-hand
wheels forced into contact
Leadling bogie left-hand
wheels forced into contact
Stage D: Non-contacting wheel starts to lift
Wheel flange Wheel
in contact starts to lift
Figure 18: Predicted wagon derailment sequence.
91 VTG Rail’s specialist had predicted similar behaviour for an Ecofret 1 triple-wagon
in simulation studies undertaken following the derailments at Felixstowe and
Peterborough (paragraph 75).
Report 14/2023 35 December 2023
London Gateway
92 The longitudinal train dynamics simulation studies showed that a high longitudinal
Analysis
compressive force would have been rapidly generated within the train at the time
of the derailment (paragraph 82b and appendix C). Baseline simulations predicted
that the maximum longitudinal compressive force occurred as the leading bogie
of wagon 12 was approaching the point of derailment. Sensitivity studies showed
that refining the traction and brake torque time histories, so that the calculated
speed profile better matched that recorded by the locomotive OTDR, significantly
increased the magnitude of the maximum longitudinal compressive force but did
not alter its location. Altering the characteristics of the couplers, for instance, by
introducing the effect of slack (at side buffer and screw coupling arrangements)
and reducing their capacity to absorb energy was also found to increase the
magnitude of the maximum longitudinal compressive force, again without altering
its location. Figure 19 shows the calculated forces in the couplers on wagon 12
and speed profile for sensitivity study C4. Of the sensitivity studies undertaken,
this study calculated the greatest maximum longitudinal compressive force
(451 kN).
Speed profile (metres per second)
Predicted speed
OTDR speed
Track elevation
Locomotive traction
Wagon 12 (lead bogie)
at point of derailment
and brake torque
Thames Haven Junction
Wagon 12 brake torque
Coupler force (kN)
Coupler force
wagon 11 to 12
Coupler force Maximum longitudinal
wagon 12 to 13 compressive force =
451 kN
Figure 19: Predicted speed profile and coupler force – sensitivity study C4.
Report 14/2023 36 December 2023
London Gateway
93 The upper graph plot shows the predicted speed profile and that recorded by
Analysis
the locomotive OTDR. The track gradient profile and applied traction and brake
torques (at the locomotive and wagon 12) are shown for pictorial reference. The
lower graph plot shows the predicted variation of the force in the couplers at
either end of wagon 12. OTDR analysis shows how the double brake application
made by the driver of train 4L47 coincided with wagon 12 approaching the
point of derailment location (paragraph 37). Figure 19 shows that the maximum
longitudinal compressive force was predicted to occur at the same time. Other
sensitivity studies show how replacing the double brake application with a brake
application of different form and size could significantly affect both the magnitude
and location of the maximum longitudinal compressive force (appendix C).
Together these show that the high longitudinal compressive force that occurred
was because of the double brake application and that it is evident that the force
generated was sufficient to result in the derailment of the leading bogie of wagon
12.
94 RAIB has investigated several freight derailments on plain line track. It has
found that these more usually involve conventional flange climb mechanisms
(such as the derailment at Reading West Junction on 28 January 2012, RAIB
report 02/2013), or behaviours associated with cyclic top (such as the derailment
near Gloucester on 15 October 2013, RAIB report 20/2014). RAIB found no
evidence that the causes frequently associated with these types of derailment
mechanism were relevant to the accident at London Gateway. Routine track
geometry measurements had not recorded any significant faults in the vicinity
(paragraph 29) and the train had not been travelling at a particularly high speed
(paragraph 9).
95 The longitudinal train dynamics studies predicted a maximum longitudinal
compressive force location that was consistent with the derailment outcome.
However, RAIB was mindful that the greatest magnitude calculated (451 kN, for
sensitivity study C4) was less than the baseline wagon derailment simulations
predicted as being needed for full derailment (650 kN (see paragraph 119)).
96 RAIB has concluded that the gap between the predictions probably arises from
data uncertainties and assumptions that were then needed to allow the simulation
work to proceed. Therefore, the inconsistency is explainable and, given the
significance of the other supporting evidence (and the lack of evidence supporting
an alternative derailment means), does not negate the conclusions drawn. RAIB
did not explore further refinement and validation of the simulation models as
it considered this would not to be proportionate to the identification of relevant
safety learning.
97 The significance of the data uncertainties and assumptions is discussed in
paragraphs 170 to 172.
Identification of causal factors
98 The accident occurred due to a combination of the following causal factors:
a. Wagon 12 was susceptible to derailment under longitudinal compressive force;
its condition, and the loading of the Ecofret 2 triple-wagon of which it was a
part, increased this risk (paragraph 99).
Report 14/2023 37 December 2023
London Gateway
b. The longitudinal compressive forces generated within train 4L47 during
Analysis
braking and which acted on wagon 12 were large, significant, and sudden.
(paragraph 125).
Each of these factors is now considered in turn.
Susceptibility to derailment under longitudinal compressive load
99 Wagon 12 was susceptible to derailment under longitudinal compressive
force; its condition, and the loading of the Ecofret 2 triple-wagon of which it
was a part, increased this risk.
100 The Ecofret 2 triple-wagon comprising wagons 11, 12 and 13 was a development
of the Ecofret 1 triple-wagon, a type of triple-wagon that had previously derailed
on plain line track because of longitudinal compressive forces generated during
train braking (paragraphs 73 and 74). Although wagon 12 was fitted with the
longer bar coupler (paragraph 76), this improvement did not fundamentally alter
the behaviour of the bar coupler arrangement under longitudinal compressive
force, since the bar coupler remained free to rotate within the headstock pocket
(paragraph 54). Ecofret 2 triple-wagons remained susceptible to derailing in
the same way, with a derailment outcome depending on the magnitude (and
nature) of the longitudinal compressive force, the payload on the three wagons
(paragraph 77) and the effects of changes introduced as part of the Ecofret 2
wagon development (paragraph 49).
101 This causal factor arose due to a combination of the following:
a. The design process for Ecofret 2 wagons did not fully recognise the need for
there to be a change in design from the Ecofret 1 wagon to improve behaviour
under longitudinal compressive forces (paragraph 102).
b. It is possible that degradation and wear had resulted in the bogies on
wagon 12 being in a condition that made the wheel flanges more prone to
being forced into contact with the rail under longitudinal compressive force
(paragraph 117).
c. The way that wagons 11, 12 and 13 were loaded made the wheels of wagon
12 more prone to lifting under longitudinal compressive force (paragraph 121).
Each of these sub-factors is now considered in turn.
Wagon design
102 The design process for Ecofret 2 wagons did not fully recognise the need
for there to be a change in design from the Ecofret 1 wagon to improve
behaviour under longitudinal compressive forces.
103 Earlier derailments and follow-up simulations had led to an understanding of
the mechanism by which the centre wagon of an Ecofret 1 triple-wagon was
susceptible to derailment under longitudinal compressive force (paragraphs
74 to 76). Witness marks and damage to the track and derailed wagons were
all consistent with a mechanism of this nature resulting in the derailment at
London Gateway (paragraphs 42 to 44 and 66), while the wagon derailment
simulations (paragraph 82c and appendix D) further supported this as the
derailment cause. The derailment at Kisby showed how the centre wagon of
an Ecofret 2 triple- wagon was similarly susceptible to derailment by this means
(paragraph 79).
Report 14/2023 38 December 2023
London Gateway
104 A further conclusion of the early derailment simulation studies was that
Analysis
triple- wagons of this type are likely to be more susceptible to derailment than
twin-wagons (paragraph 76).
105 Other options for improvement had been considered before it was decided to
lengthen the bar coupler on Ecofret 1 triple-wagons. A number of these may have
benefited the wagon design by changing the fundamental characteristic behaviour
of the Ecofret 1 triple-wagon under longitudinal compressive force. They included:
• Fitting side buffers between the centre and outer wagons, either in a
conventional (two buffer units per headstock) or diagonal (one buffer unit per
headstock) arrangement. The objective was to provide a means of energy
absorption that would reduce the effect of the longitudinal compressive force.
The buffers may also have helped to provide a moment to resist underframe
yaw rotation (paragraph 90).
• Fitting cup-and-cone devices between the centre and outer wagons to protect
against the wagons overriding each other as they get closer together, so in
effect locking the wagon ends together.
• Restricting bar coupler rotation by some means, thereby helping prevent the
coupled wagons coming out of line.
106 All these options were discounted for Ecofret 1 triple-wagons because of the
complexity of the design and modification work involved. The development of the
new Ecofret 2, however, presented an opportunity to reconsider improvement
options. Despite this, design activities and objectives either led to the conclusion
that the Ecofret 2 triple-wagon would behave satisfactorily under longitudinal
compressive force or did not result in work that might have indicated otherwise.
This meant that there was no reconsideration of potential options to improve the
design’s performance in this respect. The design activities and objectives involved
are discussed below in more detail (see paragraphs 107 to 116).
Design commonality
107 Other than the bogie and brake equipment changes, VTG Rail sought to keep the
Ecofret 1 and Ecofret 2 wagon designs as similar as possible (paragraph 50). This
particularly concerned the wagon underframe and the bar coupler arrangement.
The technical specification prepared for the Ecofret 2 wagon required the wagon
underframe to be ‘essentially the same as on Ecofret 1’, with some ‘minor
change’ needed to accommodate specified equipment differences. It specifically
required use of the same, longer bar coupler that had been later fitted to Ecofret 1
triple- wagons.
108 VTG Rail explained that no Ecofret 1 triple-wagons had derailed because of
longitudinal compressive force after the bar coupler was lengthened. On this
basis, use of the same, lengthened, bar coupler arrangement was felt to be
justified for Ecofret 2 triple-wagons.
Design process
109 VTG Rail and its suppliers separately assessed aspects of vehicle behaviour.
Each used different methods to do so.
Report 14/2023 39 December 2023
London Gateway
110 Wabtec focused on the wagon ride and derailment resistance performance
Analysis
(traditionally concerning the risk of derailment by flange climb and cyclic top
mechanisms) and developed single-vehicle dynamic simulation models to assist
with this (paragraph 62). Related performance requirements are generally
specified in TSI and NTR documents, and simulation models of this type are often
used for assessing and demonstrating compliance.
111 The assessment of risks due to longitudinal compressive forces was excluded
from Wabtec’s work. VTG Rail sought to focus on this issue because of its
overall responsibility for vehicle design (paragraph 51). It used the quasi-static
simulation model to support this, with VTG Rail asking Wabtec to supply the
parameter data needed to represent the bogie in the model.27 The need to
enhance the single-vehicle dynamic model to carry out this assessment was not
considered necessary. Such a model could have provided greater insight into the
behaviour of the Ecofret 2 triple-wagon under longitudinal compressive force (see
paragraph 169).
Design approval – interoperability requirements and technical standards
112 VTG Rail identified the standards applicable to its strategy for obtaining
authorisation for the Ecofret 2 wagon to enter service (paragraph 59). In doing
this, VTG Rail found no requirement which mandated a demonstration of safe
running under longitudinal compressive force or any criterion that could have
been used to assess this and to highlight concern. Its conclusion was consistent
with a list of standards provided to RAIB that RSSB considered would have been
applicable.
113 Criteria for assessing safe running under longitudinal compressive force are
defined in EN15839 and included as an ‘additional optional condition’ in the
WAG TSI (paragraph 72). ACS listed the standard in the technical file it prepared
as part of its NoBo role, but subsequently concluded that compliance was not
required since the Ecofret 2 triple-wagon was out of scope. Even if the wagon
had been within the scope of EN15839, RAIB has not been able to determine
any fundamental changes to the design which are likely to have resulted. This
is because EN15839 cites operation over S-shaped curves to be the special
concern, whereas wagon 12 derailed on a section of plain line track comprising
mainly straight track and plain curves of large radius.
Design approval – residual risk
114 ACS reviewed VTG Rail’s hazard log as part of its AsBo role (paragraph 60). It
identified the risks due to longitudinal compressive force as one of two issues that
remained open. ORR required the resolution of these issues as a condition of its
letter authorising the placing into service of the wagons (paragraph 63). VTG Rail
used the explicit risk estimation approach permitted by CSM REA to develop an
argument justifying closure.
27
The outer wagons were treated as being fully loaded and the centre wagon unloaded.
Report 14/2023 40 December 2023
London Gateway
115 The safety justification that VTG Rail developed was based on a combination of
Analysis
calculation and evidence of service performance for the Ecofret 1 twin- wagon.
The quasi-static simulation model had predicted that the longer bar coupler
increased the capacity of the Ecofret 1 triple-wagon to resist longitudinal
compressive force from around 400 kN to 600 kN (paragraph 76). However, when
the same simulation model was used to assess the derailment resistance for
the Ecofret 2 triple-wagon (paragraph 111), the TF20 bogie parameters meant
that the capacity to resist derailment due to longitudinal compressive force was
predicted to be 510 kN, again with the longer bar coupler.
116 VTG Rail’s engineering consultant had hypothesised a worst-case longitudinal
compressive force during braking of around 500 kN for the Ecofret 1 wagons
(paragraph 81). In the file note28 that it prepared, VTG Rail stated this to be the
‘maximum expected’ longitudinal compressive force ‘to be experienced in service’;
in an earlier file note29 it argued that the fact that no Ecofret 1 twin-wagon had
derailed since their introduction in 2012 helped to justify this figure. As this value
was less than the capacity of 510 kN calculated for the Ecofret 2 design, the
likelihood of derailment due to longitudinal compressive force was considered
‘improbable30’. On this basis, while the severity level was deemed ‘catastrophic’,
the derailment risk was assessed as being ‘tolerable’.
Bogie condition
117 It is possible that degradation and wear had resulted in the bogies on
wagon 12 being in a condition that made the wheel flanges more prone to
being forced into contact with the rail under longitudinal compressive force.
118 Wagon derailment simulations (paragraph 82c and appendix D) established that
lateral clearances within the bogie primary and secondary suspension needed to
be taken up before the longitudinal compressive force acting on the bar coupler
arrangement could result in a wheel flange being brought and forced into rail
contact. Sensitivity studies showed that the size of the lateral clearances could
detrimentally affect the capacity of the Ecofret 2 triple-wagon to resist derailment.
119 RAIB identified signs of significant damage and degradation on several PLB
devices on wagon 12 (paragraph 68). The PLBs were fitted to remedy a running
behaviour issue that was unrelated to the concerns about the risk of derailment
due to longitudinal compressive forces (paragraph 62). In the baseline derailment
simulation case, they were modelled with additional clearance to account for the
degradation resulting from the detachment of the friction strips, with a longitudinal
compressive force of 460 kN predicted as being needed for the wheel flanges to
contact opposite rails (stage C in the derailment sequence (paragraph 90)) and
650 kN for full derailment (stage E in the derailment sequence). Sensitivity studies
predicted that when the PLBs were modelled as being intact, the forces predicted
as being needed increased to 500 kN and 680 kN respectively. This supports the
degradation of the PLBs reducing the amount of longitudinal compressive force
required for the wheel flanges to contact the opposite rails, and for a bogie to
ultimately derail.
28
VTG Rail file note, FWA-FN10, issue 3.
29
VTG Rail file note, FWA-FN10, issue 2.
30
‘Improbable’ in this context was defined as an event which occurs at intervals between 100 and 1000 years.
Report 14/2023 41 December 2023
London Gateway
120 It is possible that some of the PLB damage and degradation observed occurred
Analysis
after the derailment and as a consequence of it occurring. In addition, the
sensitivity studies suggest that PLB degradation has a much less significant effect
than other factors such as payload condition (see paragraph 124). However, the
studies highlight the important role that the bogie suspension is likely to play in
the resistance to derailment under longitudinal compressive force, particularly
with respect to the control of lateral suspension movement.
Payload condition
121 The way that wagons 11, 12 and 13 were loaded made the wheels of wagon
12 more prone to lifting under longitudinal compressive force.
122 Wagon 12 carried no container payload. This was the same loading condition
as the centre wagons in previous reported Ecofret 1 and Ecofret 2 triple-wagon
derailments (paragraph 89). This lack of payload resulted in two vulnerabilities
which made wagon 12 more prone to lifting under longitudinal compressive force
and its attendant consequences:
• The lack of payload caused the bar couplers to become inclined, because the
outer wagons (wagons 11 and 13) were loaded and therefore lower on their
suspension. This provided a line of action resulting in forces that tended to lift
the centre wagon (figure 20).
• The lack of payload reduced the wheel load that the lifting forces needed to
overcome.
Centre wagon (unloaded) Outer wagon (loaded)
Height difference due to
suspension deflection
Lifting force
Suspension deflection Line of action of Bar coupler
exaggerated bar coupler force inclination
Figure 20: Bar coupler inclination and resulting centre wagon lifting force.
123 The modification to lengthen the bar coupler on the Ecofret 1 triple-wagon
(paragraph 76) had helped to address the first vulnerability because it reduced
the angle of inclination and hence the tendency for the centre wagon to lift when
longitudinal compressive forces were applied. A loading restriction had been
introduced to manage this vulnerability following the derailments at Felixstowe
and Peterborough. The bar coupler modification had allowed this restriction to be
removed (paragraph 77). A loading restriction of this nature was never applied
to Ecofret 2 wagons, and RAIB found no evidence of any other instruction that
meant wagons 11, 12 and 13 had been loaded contrary to the rules and guidance
in force at the time of the accident.
Report 14/2023 42 December 2023
London Gateway
124 The lightest container on the driver’s train document had a mass of 4.4 tonnes
Analysis
(table 1). Sensitivity studies predicted that, for this relatively modest payload,
the longitudinal compressive force needed for the wheel flanges to contact the
opposite rails (stage C in the derailment sequence) increased to 540 kN (from
460 kN in the baseline derailment simulation case) and to 930 kN (from 650 kN
in the baseline derailment simulation case) for full derailment (stage E in the
derailment sequence) (appendix D). With none of the three wagons loaded, the
longitudinal compressive forces predicted as being needed increased to 510 kN
and 850 kN respectively.
Generation of longitudinal compressive force
125 The longitudinal compressive forces generated within train 4L47 during
braking and which acted on wagon 12 were large, significant, and sudden.
126 Under normal operation, longitudinal train dynamic forces are generated within
the couplers on a train (such as the bar couplers and the conventional screw
coupling and side buffers on train 4L47) and act on the individual vehicles at
the coupler attachment points. The magnitude, sense (compressive, acting to
squeeze the vehicles together, or tensile, acting to separate the vehicles) and
variation of these forces depends on the complex interaction of the train’s physical
properties and the applied forces and torques acting. For long locomotive-hauled
freight trains, the following elements are important:
• the traction torques provided by the locomotive that are used to propel
(accelerate) the train
• the brake torques that are applied to the wheelsets on the locomotive and each
wagon to decelerate the train
• the gravitational forces acting on the locomotive and each wagon as the train
progresses over the gradient profile
• the resistance forces acting to slow the motion of the train (for instance, due to
the effects of track curvature and friction in the axle bearings)
• the masses of the locomotive and wagons and the rotational inertia of the
wheelsets (and other rotating parts) that the applied forces and torques need to
accelerate and decelerate
• the force-displacement (stiffness), slack and friction properties of the couplers
that, together with the train mass and inertia properties, help to characterise the
natural longitudinal dynamic behaviour of the train.
127 Traction torques generally result in longitudinal forces that are tensile. Brake
torques can result in negligible longitudinal forces if they apply simultaneously on
each vehicle and are in proportion to the mass of the vehicle being braked. With
an automatic air brake system, this will rarely be the case because propagation
and timing effects mean that brake torques do not apply simultaneously along the
train (paragraph 85). As a result, significant longitudinal compressive (or tensile)
forces can be generated.
Report 14/2023 43 December 2023
London Gateway
128 RAIB sought to characterise and understand the significance of the above
Analysis
elements on the longitudinal compressive forces generated within train 4L47 at
the time of the derailment by:
• test and analysis work to estimate the brake and traction torques that were
acting along train 4L47 (appendix B)
• the development and use of the longitudinal train dynamic simulation model
(appendix C).
129 This work provided evidence that:
a. Wagon 12 was within a part of the train where the longitudinal compressive
force generated during braking was likely to be higher than at other locations
along the length (paragraph 130).
b. A double brake application, made shortly after passing LG11 signal, resulted
in a sudden and large longitudinal compressive force being generated that
acted on wagon 12. This coincided with wagon 12 approaching the point of
derailment (paragraph 133).
Each of these sub-factors is now considered in turn.
Wagon location
130 Wagon 12 was within a part of the train where the longitudinal compressive
force generated during braking was likely to be higher than at other
locations along the length.
131 Wagon 12 was in the middle part of the train. The longitudinal train dynamics
simulation model predicted that, around the time of the derailment, the
longitudinal compressive forces were highest in the coupler connecting wagons
16 and 17, which was the coupler at the centre of train 4L47. The longitudinal
compressive forces acting in the couplers either side of wagon 12 were predicted
to be only slightly less (figure 21).
Coupler forces:
Wagon 12 (leading bogie)
at point of derailment
Wagons 11 to 12
Wagons 12 to 13
Wagons 16 to 17
Locomotive to wagon 1
Wagons 32 to 33
Figure 21: Predicted variation of the force in the couplers at different positions along train 4L47 around
the time of the derailment – sensitivity study C4. Plot shows the forces predicted in the couplers: either
side of wagon 12; between wagons 16 and 17; between the locomotive and wagon 1; and between
wagons 32 and 33.
Report 14/2023 44 December 2023
London Gateway
132 While wagon 17 was the centre wagon of an Ecofret 2 triple-wagon, it was
Analysis
carrying a relatively heavy container (table 1). RAIB concluded that the high
longitudinal compressive force generated in this part of the train, in combination
with the lack of container payload (paragraph 121), made wagon 12 particularly
vulnerable to the effects of longitudinal compressive force. The compressive
forces in the couplers at the outer ends of the train (between the locomotive and
wagon 1, and between wagons 32 and 33) were predicted to be negligible by
comparison.
Train brake application
133 A double brake application, made shortly after passing LG11 signal,
resulted in a sudden and large longitudinal compressive force being
generated that acted on wagon 12. This coincided with wagon 12
approaching the point of derailment.
134 OTDR data from train 4L47 shows three single brake applications being made
before the double brake application (figure 5). The last two of these were made
while the train was descending the gradient on the approach to LG11 signal. Each
was followed by a full traction effort demand. It is possible that this indicates that
the driver was having difficulty in accurately controlling the train’s speed. RAIB
found evidence that other drivers of trains that included Ecofret 2 wagons had
experienced difficulty like this, especially when attempting to control their train at
low speed.
135 The double brake application was made shortly after the last of these full traction
effort demands, and OTDR analysis shows that wagon 12 was approaching the
point of derailment as the brakes were then releasing (paragraph 37). Baseline
simulation and sensitivity studies predicted that the longitudinal compressive
forces acting on wagon 12 would have suddenly reached a maximum at this point
(appendix C). Sensitivity study C4 predicted the greatest maximum longitudinal
compressive force, 451 kN (paragraph 92 and figure 19).
136 Further sensitivity studies investigated the significance of the double brake
application (appendix C) by replacing the associated brake torques in sensitivity
study C4 with those from:
• one of the single brake applications made on the approach to LG11 signal
(sensitivity study C5), where the maximum longitudinal compressive force
reduced to 125 kN
• a larger single brake application of approximate equivalence to the double brake
application (sensitivity study C6), where the maximum longitudinal compressive
force reduced to 335 kN.
137 The significant force reduction shown in these studies led RAIB to conclude that
the sudden longitudinal compressive force that caused the derailment was the
result of the double brake application. Other effects, such as gradient profile, were
less significant.
Report 14/2023 45 December 2023
London Gateway
138 Witness and other evidence indicated a number of possible reasons why the
Analysis
double brake application was made when it was. These were:
a. Enhanced initial brake applications
In May 2021, GBRf issued urgent operating notice UON/21/006 because of
technical issues that were believed to concern how the brake distributors fitted
to Ecofret 2 wagons interacted with the brake cylinders (appendix E). GBRf
had worked with VTG Rail to understand the issues and decide the action
needed before issuing this notice.
When making an initial brake application, testing had shown that drivers
needed to reduce the brake pipe pressure to the lower value of 4.3 bar
(instead of a usual 4.6 bar). This was to avoid a reported problem with the
brake system venting air on the Ecofret 2 wagons.
This became referred to as making an ‘enhanced initial brake application’.
Drivers were required to brake in this way when operating any trains that
included Ecofret 2 wagons, which meant that it became more difficult to make
a long sustained single brake application without the train decelerating too
rapidly. Many drivers had found it necessary to adapt their normal driving style,
and the need to make a traction effort demand after an enhanced initial brake
application was cited as a means of avoiding an unplanned stop. It is possible
that the double brake application was an attempt to limit an increase in train
speed following a traction effort demand such as this.
b. Line side lighting
The driver regularly operated into London Gateway port (paragraph 31) and
reported finding the brightness of the lights located by the port camera and
OCR scanner equipment to be a distraction (paragraph 30). Knowing there
was a 15 mph (24 km/h) permanent speed restriction shortly after the lights,
they might have decided to make the double brake application to reduce
the train’s speed in advance, so avoiding the need to make a further brake
application as the train passed the lights.
c. Train mass
The driver recalled that train 4L47 felt “heavy” on the journey from Wembley
EFOC (paragraph 35). It is possible this influenced their decision regarding
the amount of braking force that was required while most of the wagons on the
train were still descending the gradient (figure 5).
Driver guidance and training
139 Freight train operators are responsible for defining the driving techniques
they expect their drivers to use and for managing how drivers comply. GBRf’s
instructions for train operation are defined in a company document called the
‘General Operating Appendix’. The document is divided into modules, including:
• Module A6 – Automatic air brake regulations
• Module B2 – Professional driving mindset
• Module B3 – Professional driving skills
• Module B7 – Economic driving
• Module B8 – Seasonal risk.
Report 14/2023 46 December 2023
London Gateway
140 These modules included considerations such as the need to check that the
Analysis
automatic air brake is functioning correctly (the running brake test, paragraph 34),
precautions when braking on the approach to signals and other places where it is
necessary to stop, and the need to anticipate low adhesion conditions. However,
RAIB found no reference in these modules to a need to drive in a particular
way with the aim of controlling or limiting train longitudinal forces and shocks.
Derailment risk mitigation is identified in the modules, but only in the context of
avoiding excessive speed and operations involving points. GBRf’s manual for the
class 66 locomotive includes no other driving instructions or guidance relevant to
the control of longitudinal train dynamic effects.
141 Although not documented in standards or guidance, GBRf explained that brake
and traction management is covered by instructor feedback during the 225 hours
of initial training that its drivers need to complete. This includes teaching drivers
that ‘large shocks’ can result in buffer and coupling damage and failure. However,
there is nothing similar concerning how train handling effects can also cause
derailment.
142 GBRf’s employee development system (EDS) includes various routine
assessments and reviews for managing driving competency after the initial
training period. GBRf provided a list of the driving-related criteria that are covered.
Included were topics such as the economic use of the traction controller and
braking techniques for mitigating the risk of wheel slide. RAIB did not identify any
criteria relating to derailment risk resulting from the generation of longitudinal
compressive forces.
143 GBRf operations standards managers (OSMs) routinely review OTDR records as
part of the EDS process. RAIB asked an OSM from GBRf about the OTDR record
from train 4L47. They concluded that an OTDR review would not have raised
undue concern or triggered the need for a support plan or further reviews. It would
probably have only resulted in constructive feedback about the benefits of braking
earlier and longer.
Identification of underlying factors
Understanding of derailment risk due to longitudinal compressive force
144 Derailment risks associated with the longitudinal dynamic behaviour of long
freight trains are not widely understood within the rail industry and there
are limited supporting processes, tools and knowledge available to assess
and manage them.
145 Although the rail industry was historically aware of issues such as traction and
braking effects and buffer interaction, RAIB found that it had undertaken little
research to further develop its understanding of longitudinal train dynamic
behaviour and the risk it poses to modern freight train operation (paragraphs
69 and 71). Longitudinal train dynamics have not been a recurrent factor in
RAIB freight derailment investigations, and it is probable that derailments of
this nature have not been occurring as regularly as they once did. This may be
due to the type of rolling stock now in service and recent operating trends. With
fewer derailments of this type occurring, it is possible that the need to develop
understanding in the field has not been considered a priority by the rail industry.
Report 14/2023 47 December 2023
London Gateway
146 The derailments involving Ecofret 1 and Ecofret 2 wagons are a reminder that the
Analysis
risks arising from longitudinal compressive forces remain. In addition, the demand
for shorter intermodal wagons, the increased use of permanently coupled wagons
with bar couplers in between, and proposals to operate longer and heavier trains
suggest that such risks may become more significant. The passenger train
derailment at Neville Hill (paragraph 70) highlights that these risks are not only
relevant to freight train operation.
147 The investigation identified four areas where there was a particular lack of
understanding and supporting resources.
Derailment resistance assessment
148 An assessment of rail vehicle derailment resistance requires:
• a criterion for quantifying derailment resistance; in this case, for instance,
the longitudinal compressive force needed to cause the first wheel to derail
(‘full derailment’) or to bring it to position where it is about to derail (‘incipient
derailment condition’)
• running conditions against which derailment safety is to be assessed; in this
case, for instance, the magnitude of the maximum longitudinal compressive
force that is considered likely to act.
149 The parties involved in the development of the Ecofret 2 wagon did not know of a
standard criterion for assessing the capacity of a triple-wagon to resist derailment
under longitudinal compressive force (paragraph 112). It was decided to adopt
an approach that involved the use of the simplified quasi-static simulation model
to determine the longitudinal compressive force at which the ‘Y/Q derailment
quotient’31 exceeded a standard critical value of 1.232 (paragraph 111). VTG
Rail’s engineering consultant had previously looked to validate this approach
via comparison with the detailed Ecofret 1 triple-wagon computer simulations
(paragraph 74).
150 The Y/Q derailment quotient is conventionally used to assess resistance to
derailment due to flange climbing. In this mechanism, sustained forces need
to act at the wheel-rail contact point to promote the wheel flange to gradually
climb onto and over the railhead. The short derailment marks (paragraph 43)
and supporting wagon derailment simulations (appendix D) showed that the
mechanism which caused wagon 12 to derail was more complex. It involved
forces at the wheel-rail interface that acted suddenly, and a sequence of
intermediate steps, with full derailment probably occurring in a chaotic fashion.
RAIB is not aware of work to fully validate the ability of standard simulation
methods to predict full derailment by a mechanism of this complexity, or to
determine the degree to which earlier steps in the sequence are indicative of an
incipient derailment condition.
151 The parties involved were also not aware of a standard that defined the maximum
longitudinal compressive force likely to act. RSSB was similarly not aware of
any such standard. VTG Rail’s engineering consultant had instead needed to
hypothesise a worst-case force (paragraph 81).
31
The ratio of the lateral (Y) and vertical (Q) forces acting at the wheel-rail contact point.
32
As defined in several technical standards, including BS EN14363.
Report 14/2023 48 December 2023
London Gateway
152 RAIB found evidence of longitudinal compressive force values (termed ‘endurable
Analysis
longitudinal compressive force’) in the assessment criteria defined in EN15839
(paragraph 72). However, it was not able to determine the degree to which
this assessment method was considered relevant to the derailment location
(paragraph 113). While the Ecofret 2 triple-wagon is a type of wagon that is
currently out of scope of EN15839, a footnote in the standard explains that
‘acceptance conditions’ are being developed for inclusion. At the time of writing
this report, RAIB has not been able to establish the progress of this work.
153 A suitably developed and robust derailment resistance assessment method would
have simplified the work undertaken to develop the Ecofret 2 wagon design
and given greater confidence regarding its ability to resist derailment due to
longitudinal compressive forces. Such a method may also be applicable to the
assessment of other wagon types.
Longitudinal train dynamic analysis
154 RAIB found evidence that the parties involved in the development of the
Ecofret 1 and Ecofret 2 wagons had discussed the idea of a longitudinal train
dynamics simulation model. This occurred both when they were developing an
understanding of the early Ecofret 1 triple-wagon derailments and as part of the
work to obtain authorisation for the Ecofret 2 triple-wagon to enter service. RAIB
found no evidence of work to develop such a model. This was probably due to a
lack of knowledge of a suitable existing analysis tool, or due to the complexities
inherent in developing one.
155 These complexities were illustrated by the work needed to develop the
longitudinal train dynamic simulation model that was critical to determining and
understanding the cause of the derailment at London Gateway (paragraph 82b
and appendix C). This involved the need for fundamental checks to validate
model behaviour against both basic physical principles and data recorded by the
locomotive OTDR (paragraph 92). It also involved the need for external expert
review and advice, including help from those having experience of the longitudinal
dynamic behaviour of freight trains operating overseas.
156 Certain simulation model parameters required information that could not readily
be provided. Further validation of the simulation model, for instance, using
dynamic vehicle testing, was outside of the scope of RAIB’s investigation
(paragraph 96).
157 A suitably developed and validated simulation model and readily accessible
parameter information would have helped those developing the Ecofret 2 wagon
to evaluate the maximum longitudinal compressive force likely to act. Such a
model would also help operators of long freight trains to assess train configuration
and operation options designed to limit such forces, and thereby define
restrictions that should be applied.
158 RAIB is aware of alternative simulation tools that have been developed for
longitudinal train dynamics analysis.33 It has not assessed the use of these, or
their limitations and benefits.
33
For instance, Wu Q et al, ‘International benchmarking of longitudinal train dynamics simulators: results’, Vehicle
System Dynamics, 2018, Vol 56, No 3, 343-365.
Report 14/2023 49 December 2023
London Gateway
Train brake force analysis
Analysis
159 RAIB concluded that neither GBRf nor VTG Rail knew of brake testing or other
data that could be used to help determine the forces acting on the individual
wagons within the train. The static brake testing undertaken on the two
representative intermodal trains was the pragmatic means of acquiring the data
needed to allow RAIB’s investigation to proceed (paragraph 82a and appendix B).
RSSB had commissioned some potentially relevant laboratory testing in the late
1990s. However, this was limited and RSSB knew of no other testing that had
sought to measure brake system pressures on a long freight train or a laboratory
equivalent.
160 A literature search undertaken as part of this investigation found reference to the
use of pneumatic simulation tools for predicting brake pipe and cylinder pressure
values on long freight trains.34 However, RAIB was not able to find evidence of
wide-scale rail industry adoption of such tools.
161 The availability of relevant and reliable brake test data, or a suitably developed
and validated pneumatic simulation tool of the train air brake system, would have
helped those developing the Ecofret 2 wagon to assess the brake forces acting
along the train. This could have been used to help determine the maximum
longitudinal compressive force likely to act. A pneumatic simulation model
could be of use to the rail industry in general as it could help operators of long
freight trains assess options for brake system configuration (such as choice of
passenger or goods distributor timings) and manage the implications of brake
system performance issues. It could also be used to inform the development of
operating standards and restrictions.
Freight train operation and handling
162 GBRf had provided no specific guidance to its operating staff on how to manage
the effects of longitudinal train dynamic behaviour to avoid the risk of derailment
(paragraphs 140 and 141). RSSB does not publish standards that it would expect
to be used for this purpose.
163 The specification of rules and guidance in the relevant train operating standard
could have reinforced practice that was designed to help control and limit train
longitudinal forces. As well as driving techniques, such rules and guidance could
also cover relevant aspects of train preparation, such as the degree to which
screw couplings need to be tightened (and slack allowed) and any necessary
restrictions on train consist configuration and loading.
34
For instance, Catone L, ‘TrainDy: the new Union Internationale de Chemins de Fer software for freight
interoperability’, Proceedings of the Institution of Mechanical Engineers, Vol 225 Part F: Journal of Rail and Rapid
Transit.
Report 14/2023 50 December 2023
London Gateway
Approach to wagon design
Analysis
164 VTG Rail adopted design management arrangements that possibly limited
its ability to understand the dynamic behaviour of the triple-wagon as
a complete vehicle system and did not result in it identifying critical
subsystem performance requirements that were associated with the
behaviour of the train as a whole.
165 VTG Rail established a steering group to co-ordinate the work of its main
suppliers (Wabtec and WH Davis) in developing the design of the Ecofret 2
triple- wagon (the ‘vehicle system’). They worked together using a consortium-type
project management arrangement, with engineering responsibilities being largely
divided between these suppliers according to the subsystems within their scope,
principally the underframe, couplers and bogie (paragraph 51).
166 The responsibilities for evaluating vehicle ride and running safety (largely
to demonstrate compliance with identified standards (paragraph 62)) were
separated from those for evaluating derailment resistance under longitudinal
compressive force (paragraphs 111 and 114 to 116). RAIB found evidence that
this did not result in use of what would possibly have been the best approach for
understanding the dynamic behaviour of the vehicle system or the forces relating
to it.
167 A technical specification was prepared for the wagon, and this effectively defined
the vehicle and the subsystem requirements. However, the technical specification
largely cascaded vehicle system requirements to the subsystems according to
the desired design solution (or feature) rather than on the basis of performance.
For instance, the bar coupler (part of the coupler subsystem) was specified with
reference to the supplier’s component drawing, its size (1600 mm long) and its
design features (incorporation of a ‘rubber damper’).
168 While the ‘draw force’ (the longitudinal tensile force acting on the bar coupler) was
mentioned as a performance-related requirement, the technical specification was
silent on the maximum longitudinal compressive force to be catered for. In the
absence of published values (paragraph 112), efforts to define and validate the
maximum longitudinal compressive force by consideration of the overall behaviour
of the train (for instance, by calculation, simulation or test) could have provided
greater confidence in the derailment resistance of an Ecofret 2 triple-wagon.
169 The division of responsibilities also meant that a computer simulation model
representing the dynamic behaviour of the complete Ecofret 2 triple-wagon
was not available. Such a model (like that developed for the wagon derailment
simulations carried out during the investigation (appendix D)) could have
provided greater insight into the behaviour of the triple-wagon under longitudinal
compressive force and the complex interactions of the coupler and bogie
subsystems. Obtaining approval for the bogie as a separate interoperability
constituent (paragraph 61) may have reinforced the responsibility division.
Report 14/2023 51 December 2023
London Gateway
Data uncertainties and assumptions
Analysis
170 RAIB was mindful that the magnitude of the maximum longitudinal compressive
forces predicted by the longitudinal train dynamics model were less than that
calculated as being needed for full derailment by the wagon derailment simulation
model. This probably arose from data uncertainties and the assumptions that
were then needed (paragraphs 95 and 96).
171 Assumptions necessary in the development of the longitudinal train dynamics
model are probably particularly important in this respect. Several of these related
to uncertainties associated with effects that could result in greatly increasing
the magnitude of the predicted maximum longitudinal compressive force. They
included:
a. Coupling tightening and slack
The rail industry has long recognised the importance of coupling tightness and
slack as a contributor to the generation of longitudinal forces and derailment
risk. However, there was a lack of information on the amount of slack within
train 4L47. In sensitivity study C1 it was assumed to be zero (appendix C). In
sensitivity study C2 a nominal slack of ±25 mm was introduced. This increased
the predicted maximum longitudinal compressive force by around 30 kN. The
actual amount of slack could have been much greater than this nominal figure
and the amount of longitudinal compressive force thereby also greater.
b. Coupler friction losses
Coupler friction is also recognised as a factor that is important to longitudinal
train dynamic behaviour. The only reliable information on this was for the
Ecofret 2 wagon coupler components.35 In the baseline simulation, it was
assumed that the friction model developed for the Ecofret 2 bar coupler was
appropriate to all other couplers on the train (appendix C). Sensitivity study
C4 predicted that a 50% reduction in friction, when combined with slack of
±25 mm, increased the longitudinal compressive force by around 130 kN.
c. Air brake system equipment
It was not possible to use the actual brake pipe and brake cylinder pressures
from train 4L47 when determining the wagon brake torque time histories.
Measurements from representative trains were instead processed to estimate
representative values (appendix B). There was insufficient information
to establish the degree to which the air brake system equipment on the
representative trains differed in detail from that on train 4L47, for instance,
in terms of specification, setting, condition, behaviour, and performance.
Sensitivity study C1 showed how the magnitude of the vehicle brake torques
could affect the maximum longitudinal compressive force (appendix C). As
well as affecting the brake torque magnitude, air brake system equipment
differences may have affected the time at which the brake torques applied.
Other brake equipment differences, such as the configuration and condition of
the mechanical brake rigging, may also have affected the brake torque.
35
There was some limited information for some components on the FEA wagons.
Report 14/2023 52 December 2023
London Gateway
d. Speed dependent braking effects
Analysis
The double brake application resulted in train 4L47 rapidly slowing. The
OTDR recorded the locomotive’s speed reducing to less than 9 mph
(14.4 km/h) before starting to increase as traction was demanded (figure 5).
The trajectory of the reducing speed suggests that it is possible that wagons
on the train (whose speed is not recorded by the locomotive’s OTDR), may
have momentarily been travelling significantly slower. Sensitivity study C4,
which used the refined traction and brake torque time histories, predicted
something similar (figure 19). Data published by RSSB suggests that train
brake forces can rapidly increase at slow speed. This is due to variation in the
brake block and pad friction characteristic. VTG Rail’s engineering consultant
hypothesised that this speed dependent effect could as much as double the
longitudinal compressive force.
e. Train resistance
In the absence of experimental or other data, the longitudinal train dynamics
simulation studies assumed that the train resistance forces would act to slow
the locomotive and each wagon at the same rate. Deviation from this would
result in the generation of additional longitudinal compressive (or tensile)
forces.
Other uncertainties, such as the lack of information on the characteristics of some
of the coupler-related components, and the actual mass of the containers, might
also have affected the accuracy of the calculations.
172 Other assumptions led RAIB to conclude that the baseline wagon derailment
simulation model probably overpredicted the capacity of wagon 12 to resist
derailment due to longitudinal compressive force. These principally concerned
the friction elements used to represent the friction losses in the bar coupler
(appendices C and D). The formulation of the friction elements meant that their
line of action remains fixed and does not follow that of the ‘pinlink’ elements they
are parallel to, which are able to rotate. As a result, the friction-related component
of the bar coupler force was not accurately aligned. Separate simulation studies
suggested that this alignment inaccuracy could result in an overprediction of
resistance of around 100 kN or more. If correct, then this would mean that
the wagon was more vulnerable in practice to derailment due to longitudinal
compressive force.
Previous occurrences of a similar character
173 Research published in 200636 reviewed UK railway derailment data for the
ten- year period between 1992 and 2002. It identified 33 derailment events that
were due to longitudinal train dynamic effects. Of these, 27 events related to the
presence of sudden severe longitudinal braking forces. The research identified a
further four derailments that were due to buffer interaction (referred to as ‘buffer
locking/forces’).
36
RSSB report ‘Cost effective reduction of derailment risk: initial analysis’, January 2006. Research project T357.
Report 14/2023 53 December 2023
London Gateway
174 The first known derailment of an FWA type wagon due to longitudinal compressive
Analysis
force occurred on 28 August 2015 on the entry line to the north and central rail
terminals within the Port of Felixstowe. The train comprised a class 66 locomotive
hauling 31 container flat wagons, which were a mixture of FEA and Ecofret 1
type wagons. The centre and leading outer wagons of an Ecofret 1 triple-wagon
derailed, which were the 14th and 15th wagons on the train. Another derailment
occurred at Peterborough on 17 September 2015. On this occasion, the train
comprised two class 66 locomotives (one of which was not providing any traction
effort) and 32 container flat wagons. These were a mixture of FEA and Ecofret 1
type wagons. The centre and trailing outer wagons of an Ecofret 1 triple-wagon
derailed. These were also the 14th and 15th wagons on the train. Both trains were
operated by GBRf.
175 Both derailments occurred at low speed on straight plain line track. Only the
leading outer wagon of the Ecofret 1 triple-wagon at Felixstowe was carrying a
container, while the centre wagon was not loaded in either derailment. The rail
industry investigation work undertaken in response to these derailments led VTG
Rail to decide to increase the length of the bar coupler on Ecofret 1 triple-wagons
(paragraph 76).
176 The derailment of an empty Ecofret 2 triple-wagon occurred as a result of the
collision of a GBRf intermodal train at Kisby level crossing on 19 August 2021.
RAIB has considered the mechanism that caused the wagon to derail as part of
this investigation (paragraph 79).
177 The only similar derailment that RAIB has investigated occurred at Neville Hill
depot on 13 November 2019 (paragraph 70). This involved the collision of a
relatively new Intercity Express Train. The investigation found that the train
was susceptible to derailment due to high longitudinal compressive forces.
This susceptibility was due to the design of the bar coupler arrangement which
connected the vehicles together. This arrangement allowed the bar coupler to
rotate relatively freely, and in this regard was like the Ecofret 2 triple-wagon. The
impact of these features on derailment resistance had not been appreciated when
the train was designed.
Report 14/2023 54 December 2023
London Gateway
Summary of conclusions
Summary of conclusions
Immediate cause
178 Longitudinal compressive forces generated within train 4L47 during braking were
sufficient to cause the wheels on the leading bogie of wagon 12 to suddenly lift
and run derailed (paragraph 87).
Causal factors
179 The causal factors were:
a. Wagon 12 was susceptible to derailment under longitudinal compressive force;
its condition, and the loading of the Ecofret 2 triple-wagon of which it was
a part, increased this risk (paragraph 99). This causal factor arose due to a
combination of the following:
i. The design process for Ecofret 2 wagons did not fully recognise the
need for there to be a change in design from the Ecofret 1 wagon to
improve behaviour under longitudinal compressive forces (paragraph 102,
Recommendation 1).
ii. It is possible that degradation and wear had resulted in the bogies on
wagon 12 being in a condition that made the wheel flanges more prone
to being forced into contact with the rail under longitudinal compressive
force (paragraph 117, Recommendation 1).
iii. The way that wagons 11, 12 and 13 were loaded made the wheels of
wagon 12 more prone to lifting under longitudinal compressive force
(paragraph 121, Recommendation 1).
b. The longitudinal compressive forces generated within train 4L47 during
braking and which acted on wagon 12 were large, significant, and sudden
(paragraph 125). This causal factor arose due to a combination of the
following:
i. Wagon 12 was within a part of the train where the longitudinal
compressive force generated during braking was likely to be higher than
at other locations along the length (paragraph 130, Recommendation 2).
ii. A double brake application, made shortly after passing LG11
signal, resulted in a sudden and large longitudinal compressive
force being generated that acted on wagon 12. This coincided with
wagon 12 approaching the point of derailment (paragraph 133,
Recommendation 2).
Report 14/2023 55 December 2023
London Gateway
Underlying factors
Summary of conclusions
180 The underlying factors were:
a. Derailment risks associated with the longitudinal dynamic behaviour of long
freight trains are not widely understood within the rail industry and there are
limited supporting processes, tools and knowledge available to assess and
manage them (paragraph 144, Recommendation 3).
b. VTG Rail adopted design management arrangements that possibly limited
its ability to understand the dynamic behaviour of the triple-wagon as a
complete vehicle system and did not result in it identifying critical subsystem
performance requirements that were associated with the behaviour of the train
as a whole (paragraph 164, Learning point 1).
Report 14/2023 56 December 2023
London Gateway
Actions reported as already taken or in progress relevant to
Actions reported as already taken or in progress relevant to this report
this report
181 Following the Ecofret 2 wagon derailments at Kisby and London Gateway, VTG
Rail asked its engineering consultant to revisit the earlier quasi-static simulation
studies (paragraphs 111 and 115). The conclusion drawn from this revised
simulation was that the derailments were due to the effects of longitudinal
compressive forces and that a centre wagon payload of more than five tonnes
greatly increased derailment resistance. As a result, VTG Rail issued temporary
loading instructions which required the centre wagon of an Ecofret 2 triple-wagon
to always carry a container, unless both the outer wagons were unloaded. GBRf
issued urgent operating notice UON/21/007 to ensure loading is undertaken in
compliance with this instruction.
182 GBRf is assisting VTG Rail to progress modification work to remedy the brake
distributor issues on Ecofret 2 wagons that had led to GBRf issuing an urgent
operating notice (paragraph 138a). It is anticipated that this will remove the
requirement for drivers to make enhanced initial brake applications when
operating trains that include Ecofret 2 wagons.
183 Wabtec is progressing work to resolve the underlying reasons for the in-service
degradation of the PLBs on TF20 bogies. It has reported that the bogies on the
majority of wagons have now been retrofitted with PLBs of a new stronger design.
Report 14/2023 57 December 2023
London Gateway
Recommendations and learning point
Recommendations and learning point
Recommendations
184 The following recommendations are made:37
1 The intent of this recommendation is to ensure that the vulnerability of
the Ecofret 2 triple-wagon to derailment due to longitudinal compressive
forces is fully understood and that the risk of this occurring is
appropriately controlled.
VTG Rail, supported by Wabtec UK Ltd, should investigate the dynamic
behaviour of Ecofret 2 triple-wagons to better understand the operating
conditions that result in longitudinal compressive forces acting on these
wagons, their behaviour under such forces, and their capacity to resist
derailment. This investigation should be used to identify appropriate
mitigation measures such as design changes and operating restrictions,
and should specifically include consideration of:
• improvements to the bar coupler arrangement
• control of lateral suspension movement
• wagon payload restrictions
(paragraphs 179a.i, 179a.ii and 179a.iii).
This recommendation may also be applicable to other owners
or operators of rolling stock made up of freight wagons that are
permanently coupled together using bar couplers at both ends.
37
Those identified in the recommendations have a general and ongoing obligation to comply with health and safety
legislation, and need to take these recommendations into account in ensuring the safety of their employees and
others.
Additionally, for the purposes of regulation 12(1) of the Railways (Accident Investigation and Reporting) Regulations
2005, these recommendations are addressed to the Office of Rail and Road to enable it to carry out its duties under
regulation 12(2) to:
(a) ensure that recommendations are duly considered and where appropriate acted upon; and
(b) report back to RAIB details of any implementation measures, or the reasons why no implementation measures
are being taken.
Copies of both the regulations and the accompanying guidance notes (paragraphs 200 to 203) can be found on
RAIB’s website www.raib.gov.uk.
Report 14/2023 58 December 2023
London Gateway
Recommendations and learning point
2 The intent of this recommendation is to ensure that freight trains are
configured and operated in such a way that the longitudinal compressive
forces generated in service are not sufficient to cause derailment.
GB Railfreight, working where appropriate with relevant industry
working groups, should review the risks associated with longitudinal
train dynamics when operating freight trains. It should use this to inform
the development of instructions, best practice guidance and training
for operations staff so that longitudinal compressive forces are not
generated that exceed the derailment resistance of the wagons forming
the train. This work should include consideration of:
• the management of brake application and traction demands
• the effects of the train air brake system configuration, setting,
degradation and operation
• the control of slack and pre-load in screw coupling and buffer
arrangements
• the need for train configuration and payload restrictions and limits
(paragraphs 179b.i and 179b.ii).
This recommendation may also be applicable to other freight operating
companies.
3 The intent of this recommendation is to establish a robust and pragmatic
industry-wide framework for managing the risks of freight train
derailments due to longitudinal train dynamic effects.
RSSB working with the freight operating companies should initiate
a programme of work to raise the rail industry’s understanding of
longitudinal compressive forces in freight trains and the associated
derailment risk. This work should incorporate:
• identifying, developing and validating analysis tools and techniques
• determining and validating relevant derailment criteria
• reviewing and developing a suitable derailment resistance assessment
process
• identifying limits for longitudinal compressive force that can be
generated in operational service, and the practical means by which it
can be assured that these are not exceeded.
RSSB should develop a timebound programme for the implementation
of any appropriate changes identified, such as new or revised standards,
guidance and operating rules (paragraph 180a).
Report 14/2023 59 December 2023
London Gateway
Learning point
Recommendations and learning point
185 RAIB has identified the following learning point:38
1 It is important that parties responsible for the introduction of rail vehicle
products and equipment into service, such as new and modified rolling
stock and interoperable components, adopt processes and procedures
which ensure the use of best available engineering design and analysis
practice.
38
‘Learning points’ are intended to disseminate safety learning that is not covered by a recommendation. They are
included in a report when RAIB wishes to reinforce the importance of compliance with existing safety arrangements
(where RAIB has not identified management issues that justify a recommendation) and the consequences of failing
to do so. They also record good practice and actions already taken by industry bodies that may have a wider
application.
Report 14/2023 60 December 2023
London Gateway
Appendices
Appendices
Appendix A – Glossary of abbreviations and acronyms
ACS AEGIS Certification Services Limited
ApBo Approved Body
AsBo Assessment Body
CCTV Closed-circuit television
CSM REA Common Safety Methods for Risk Evaluation and Assessment
DeBo Designated Body
EDS Employee development system
EFOC European Freight Operating Centre
GB Great Britain
GBRf GB Railfreight
GSM-R Global system for mobile communications - railways
NoBo Notified Body
NTR National Technical Rule
NTSN National Technical Specification Notice
OCR Optical character recognition
ORR Office of Rail and Road
OSM Operations standards manager
OTDR On-train data recorder
PLB Primary lateral buffer
RSSB Rail Safety and Standards Board
TSI Technical Specification for Interoperability
VIBT Vehicle inspection and brake test
VTG Vereinigte Tanklager und Transportmittel GmbH
WAG TSI Technical specification for interoperability relating to the
subsystem ‘rolling stock – freight wagons’
Report 14/2023 61 December 2023
London Gateway
Appendix B – Derivation of traction and brake torques
Appendices
B1 In line with normal practice, the locomotive OTDR only recorded the brake
pipe and brake cylinder pressures on the locomotive itself (paragraph 36).
Although some historical supplementary rail industry data was available, it was
not possible to use this to work out the corresponding brake pipe and brake
cylinder pressures at each individual wagon.39 RAIB concluded that brake system
pressure measurements on a representative train were needed.
B2 Two intermodal trains were made available on separate dates for brake testing.
Each was formed of a class 66 locomotive and a similar number of wagons.
Ecofret 2, FEA-A, FEA-B and FEA-C type wagons were included on both trains.
Pressure transducers were added to various brake pipe and brake cylinder test
points. The driver’s brake valve was operated with the train stationary and the
resulting pressure measurements were recorded using a data acquisition unit
(figure B1).
Pressure
transducer
Figure B1: Brake testing arrangements. Inset shows pressure transducer fitted to a test point in the train
brake pipe. (Main photograph courtesy Railway Braking Associates Limited).
39
In the late 1990s, RSSB commissioned research that investigated the effect of train length on braking
distance. Supporting laboratory testing measured the pressures within a single pipe automatic air brake system
at various train length positions. However, the work was focused on brake application and did not include
measurements relating to brake release which were also needed.
Report 14/2023 62 December 2023
London Gateway
B3 This testing had two main objectives:
Appendices
• To obtain pressure measurements to help characterise the behaviour of the
various wagon brake equipment types on train 4L47. The first train was mainly
used for these tests.
• To simulate the brake applications made by the driver of train 4L47 on the
Down Thames Haven line and record the corresponding brake pipe and brake
cylinder pressures at selected positions along the train: the locomotive; the
first wagon; the 33rd wagon (the location of the last wagon on train 4L47); and
nominated wagons in between. The second train was used for these tests.
B4 The wagon brake equipment included ‘load weigh valves’ which are used to vary
the brake cylinder pressure (and therefore the wagon brake force) according to
wagon suspension displacement (which is proportional to the wagon payload).
B5 Many of the wagons on the tested trains carried no container payload, whereas
nearly all the wagons on train 4L47 were laden (table 1). As a means of
accounting for this, the load weigh valves on the wagons of the second train were
manually configured to behave as though a full container load was being carried.
B6 RAIB used the test data, the locomotive’s OTDR record (figure 5), and a
combination of brake system design and parameter information to estimate:
• the magnitude of brake torque acting on the wheelsets40 of each wagon of train
4L47
• how this varied in time as the train travelled from Thames Haven Junction to
where it came to rest.
This is referred to in the report as the ‘wagon brake torque time history’.
B7 RAIB used a combination of locomotive design and test data to estimate the
brake torque time history for the locomotive (‘locomotive brake torque time
history’).41 For this it was able to use the brake pipe and brake cylinder pressures
recorded by the locomotive’s OTDR directly (paragraph B1). It estimated the
traction torques acting to accelerate the train (‘locomotive traction torque time
history’) in a similar way.
B8 When aggregated, the individual torque time histories define how the traction and
brake forces distribute along the train and vary with time.
40
The wagons in the longitudinal train dynamics model were represented as having only two wheelsets, whereas
each wagon in train 4L47 had four (see appendix C). Therefore, the brake torques were calculated for each pair of
wheelsets.
41
The locomotive in the longitudinal train dynamics model was represented as having only two wheelsets,
whereas a class 66 locomotive has six. Therefore, the brake torques were calculated for a set of three wheelsets.
Report 14/2023 63 December 2023
London Gateway
Appendix C – Simulation of longitudinal train dynamics
Appendices
C1 In work arising from the investigation of the early Ecofret 1 triple-wagon
derailments, it had been necessary to make some significant assumptions and
simplifications when hypothesising a worst case longitudinal compressive force
(paragraph 81). RAIB concluded that a longitudinal train dynamics simulation
model was needed to assess and understand these forces and how they varied.
With no pre-existing analysis available, the development of a simulation model
was required. For train 4L47, this needed to take into consideration:
• the inertia properties of the locomotive and each of the 33 wagons
• the characteristics of the couplers connecting the locomotive and wagons
• the gradient
• the traction and braking forces.
C2 A simulation model was developed using the longitudinal analysis capabilities of
an industry-standard railway vehicle dynamics analysis package. The train was
modelled as a string of 34 mass elements connected using pairs of special spring
elements (‘bumpstop’ elements) to represent the coupler non-linear stiffness,
with friction elements added in parallel to represent the energy absorbing
characteristics. The mass elements represented the inertia properties of the
non-rotating parts of each of the vehicles. Separate ‘wheelset’ elements, two per
vehicle for simplicity, were used to represent the rotating parts (figure C1).
8 7 6 5 4 3 2 1 Locomotive
18 17 16 15 14 13 12 11 10 9
26 25 24 23 22 21 20 19
33 32 31 30 29 28 27
Wagon 13 Wagon 12 Wagon 11
Mass element
Brake torque Non-linear stiffness element Wheelset element
Figure C1: Longitudinal train dynamics simulation – model of train 4L47.
Report 14/2023 64 December 2023
London Gateway
C3 The gradient profile of the Down Thames Haven line was defined in a separate
Appendices
track model.42 Simulations of train 4L47 running along the line involved defining
a starting speed and the traction and brake torques acting on the wheelset
elements to accelerate and decelerate the train as it progressed, step by step.
C4 Data of sufficient integrity was available for most of the model elements.
However, there was some information that not could not easily be sourced.
Table C1 summarises the main assumptions that were adopted for the baseline
simulation case.
Assumption adopted
Information deficiency
- baseline simulation case
Lack of design and test data for the bar Parameter data for components of the same
coupler, screw coupling and side buffer generic type used
arrangements on some of the FEA wagon
types
Characteristics of the screw coupling and Parameter data for components of the same
side buffer arrangements on locomotive generic type used
66779
Lack of data on the degree to which screw Parameters for all side buffer and screw
couplings were tightened and the resulting coupling arrangements based on the buffer
slack or additional preload (paragraphs 56 faces just touching (no slack or additional
and 57) preload)
General lack of data on coupler Generic friction model developed and
friction (and other energy absorption) validated for a coupler for which standard test
characteristics likely in service data was available (the Ecofret 2 bar coupler
arrangement). The same friction model was
then used for all other couplers.
In the generic friction model, the force in the
friction elements (friction force) was equal
to 50% of the force in the corresponding
bumpstop element
Table C1: Parameter assumptions adopted for the longitudinal train dynamics baseline simulation case.
C5 In the baseline simulation, the locomotive traction torque, locomotive
brake torque and wagon brake torque time histories were used as derived
(appendix B). It predicted a train speed profile that favourably matched
that recorded by the locomotive OTDR. It also predicted that the maximum
compressive force in the bar couplers on wagon 12 (that is to say, the
longitudinal compressive force acting on wagon 12) occurred as the double
brake application was releasing (paragraph 37). This coincided with the time
that wagon 12 would have been approaching the point of derailment (figure C2).
The maximum longitudinal compressive force was calculated to be 203 kN and it
had rapidly increased to this value. Other longitudinal compressive force peaks
coincided with brake application events.
42
Two kilometres of track, level for one kilometre then an uphill constant gradient for another kilometre, were
added ahead of Thames Haven Junction. This was to ensure that the locomotive and all wagons were subject to
the same gradient effects before the train reached the junction and the traction and braking torques were applied.
Report 14/2023 65 December 2023
London Gateway
Appendices
Speed profile (m/s)
Predicted speed
OTDR speed
Track elevation
Locomotive traction
and brake torque
Wagon 12 (leading bogie)
Wagon 12 brake torque
at point of derailment
Thames Haven Junction
Coupler force (kN)
Maximum longitudinal
Coupler force compressive force =
wagon 11 to 12 203 kN
Coupler force
wagon 12 to 13
Figure C2: Predicted speed profile and coupler force – baseline simulation. The upper graph plot shows
the predicted speed profile and that recorded by the locomotive OTDR. The track gradient profile and
applied traction and brake torques (at the locomotive and wagon 12) are shown for pictorial reference
only. The lower graph plot shows the predicted variation of the force in the couplers at either end of
wagon 12.
C6 Several steps and assumptions were involved in estimating the traction and
brake torque time histories and, because of this, a degree of uncertainly must
be recognised in the results. RAIB found that an improved speed match could
be obtained by factoring the traction and brake torque values (referred to as the
‘refined traction and brake torque time histories’). No change was made to the
time at which the torques applied. Simulations using the refined traction and
brake time histories are referred to as sensitivity study C1. They predicted the
maximum longitudinal compressive force occurred at the same location, but its
magnitude was greatly increased.
Report 14/2023 66 December 2023
London Gateway
C7 Other sensitivity studies investigated the significance of the following on the
Appendices
magnitude and location of the maximum longitudinal compressive force acting on
wagon 12:
• coupler slack (sensitivity study C2)
• coupler friction (sensitivity study C3)
• coupler slack and friction (sensitivity study C4)
• the double brake application that was made before the derailment (sensitivity
studies C5 and C6).
C8 Table C2 summarises each of the sensitivity studies and the magnitude and
location of the predicted maximum longitudinal compressive force event.
Sensitivity Description Predicted maximum longitudinal
study compressive force
Magnitude Location
Baseline simulation 203 kN Coincident with leading
bogie of wagon 12
passing over the point of
derailment
C1 Refined traction and brake torque 322 kN Coincident with leading
time histories (paragraph 92) bogie of wagon 12
passing over the point of
Other parameters as baseline
derailment
simulation
C2 ±25 mm slack at all couplers 353 kN Coincident with leading
comprising screw coupling and side bogie of wagon 12
buffer arrangements passing over the point of
derailment
Other parameters as sensitivity
study C1
C3 Friction force equal to 25% of the 324 kN Coincident with leading
force in the bumpstop element at all bogie of wagon 12
couplers passing over the point of
derailment
Other parameters as sensitivity
study C1
C4 ±25 mm slack at all couplers 451 kN Coincident with leading
comprising screw coupling and side bogie of wagon 12
buffer arrangements; and friction passing over the point of
force equal to 25% of the force derailment
in the bumpstop element at all
intervehicle connections
Other parameters as sensitivity
study C1
Report 14/2023 67 December 2023
London Gateway
Appendices
Sensitivity Description Predicted maximum longitudinal
study compressive force
Magnitude Location
C5 Brake torques associated with a 125 kN Coincident with
single brake application instead of replacement single brake
the double brake application. application
Brake torques for the replacement
single brake application equivalent
to that of a single brake application
made earlier on the Down Thames
Haven line.
Other parameters as sensitivity
study C4
C6 Brake torques associated with a 325 kN Coincident with
single brake application instead of replacement single brake
the double brake application. application
Brake torques for the replacement
single brake application increased
to match the approximate overall
magnitude of the double brake
application.
Other parameters as sensitivity
study C4
Table C2: Sensitivity studies – summary and results. Baseline simulation results are included for
reference.
C9 In summary:
• Sensitivity study C1 showed how refinement of the traction and brake torques
resulted in greatly increasing the magnitude of the maximum longitudinal
compressive force, without altering its location and, therefore, its significance
to the derailment. The calculated speed profile closely followed the speed
recorded by the locomotive’s OTDR up to where the double brake application
started to release. At this location the predicted speed continued to reduce,
initially following the trajectory recorded by the locomotive OTDR, but reaching
a lower minimum value before rising again.
• Sensitivity studies C2, C3 and C4 showed how changing the coupler
characteristics could also greatly increase the magnitude of the maximum
longitudinal compressive force, again without altering its location on the line.
The speed profile calculated in all three studies was nearly identical to that
calculated in sensitivity study C1.
• The greatest maximum longitudinal compressive force was calculated for
sensitivity study C4. Sensitivity studies C5 and C6 showed how making an
alternative brake application could have reduced the magnitude of this force.
C10 Figure 19 shows the predicted speed profile for sensitivity study C4 and the
predicted variation of the force in the couplers at either end of wagon 12.
Report 14/2023 68 December 2023
London Gateway
Appendix D – Wagon derailment simulations
Appendices
D1 RAIB concluded that a new wagon derailment simulation study was needed to
understand the longitudinal compressive forces likely to present a derailment
risk to wagon 12. The bar coupler and bogie changes (paragraphs 49 and
76) introduced to the Ecofret 2 design meant that it was not possible to
reuse analysis from the earlier Ecofret 1 triple-wagon derailment studies
(paragraph 74).
D2 Wabtec constructed a vehicle model of the Ecofret 2 triple-wagon that derailed
using the single-vehicle computer models that it had already developed and
validated (paragraph 62). RAIB provided the design information for the bar
coupler components, as supplied by VTG Rail, and the details of the container
payload, as recorded on the train document (table 1).
D3 ‘Pinlink’ elements were needed to represent the behaviour of the bar couplers
connecting the wagons to each other. These are special rod-like elements
that can rotate freely about their end pivot points, and transfer axial forces.
The location and movement of the pivot points (on the connected vehicle
underframes) determine the element length and the line of action of the axial
force. A pinlink element can suddenly rotate (buckle) if the applied compressive
force becomes too great. This is an important characteristic of the element since
it means the resulting rapid change in the direction of the axial force, and how it
acts on the connected underframes, is represented.
D4 The non-linear stiffness characteristic of the two spring units in the bar coupler
arrangement (figure 11) were defined by reference to supplied test information.
The friction losses in the bar coupler were represented using a friction model
like the one adopted in the longitudinal train dynamics simulation model
(appendix C). The containers on wagons 11 and 13 were represented as
separate mass elements, tied to the wagon underframe at the spigot locations.
D5 RAIB adjusted the model that Wabtec supplied. For the baseline simulation case,
this included:
• Introducing nominal lateral drawbar pivot offsets; VTG Rail’s engineering
consultant (paragraph 76) had found that such offsets can be significant to
initiating the behaviour found to occur during such derailments; the value
chosen (±10 mm) was within what could easily be explained due to the
general build and maintenance tolerances and possible relative positioning of
wheelsets on the track.
• Altering the mass properties of the containers on wagons 11 and 13 to reflect
measurements made after the derailment (table 1); the heights of each of the
drawbar pivots were corrected to account for the corresponding change in
suspension deflection.
• Altering the stiffness characteristics of the PLBs to account for degradation
(paragraph 68).
D6 Track geometry recordings from 13 July 2021 (paragraph 29) and wheel profile
measurements (paragraph 65) were used to create the required track irregularity
and wheel-rail contact definitions.
Report 14/2023 69 December 2023
London Gateway
D7 The baseline simulation study focused on the effect of rapidly applying a
Appendices
sustained longitudinal compressive force on wagon 12 as the Ecofret 2
triple- wagon approached the derailment location. This was achieved by defining
equal and opposite longitudinal forces that acted on wagon 11 and wagon 13
and increasing these in 10 kN steps. The same approach had been used by VTG
Rail’s vehicle dynamics specialist in the earlier Ecofret 1 triple-wagon derailment
studies. The baseline simulations predicted a derailment sequence that was
similar to that described for the earlier Ecofret 1 studies (paragraphs 90 and 91);
with reference to figure 18, the following longitudinal compressive forces were
required:
• 460 kN for the wheel flanges on both bogies of wagon 12 to contact opposite
rails (stage C)
• 600 kN: for a non-contacting wheel to start lifting off the rail (stage D)
• 650 kN: for full derailment (stage E).
D8 Sensitivity studies investigated the significance of the following on the
longitudinal compressive forces required:
• PLB condition (sensitivity study D1)
• wheel profile condition (sensitivity study D2)
• wagon payload (sensitivity studies D3 and D4).
D9 Table D1 summarises each of the sensitivity studies and the required longitudinal
compressive forces that were predicted.
Sensitivity Description Predicted longitudinal compressive
study force required
Stage C Stage D Stage E
Wheel Non- Full
flanges in contacting derailment
contact with wheel starts
opposite rails to lift
Baseline simulation 460 kN 600 kN 650 kN
D1 PLBs intact 500 kN 600 kN 680 kN
Other parameters as baseline simulation
D2 Wheel profiles in accordance with 440 kN 600 kN 640 kN
specified P8 wheel profile
Other parameters as baseline simulation
D3 Container payload of 4.4 tonnes on 540 kN 790 kN 930 kN
wagon 12
Other parameters as baseline simulation
D4 No container payload on wagon 11, 12 510 kN 740 kN 850 kN
or 13
Other parameters as baseline simulation
Table D1: Wagon derailment simulations. Sensitivity studies – summary and results. Baseline simulation
results are included for reference.
Report 14/2023 70 December 2023
London Gateway
Appendix E – Urgent operating notice concerning trains with Ecofret
Appendices
2 wagons
UON/21/006
URGENT OPERATING NOTICE
Ecofret2 Wagons –Technical Issues
Summary
Following reports of braking technical issues from Train Managers operating trains
with the new Ecofret2 wagons. GBRf have completed some static testing of these
vehicles at Peterborough and also dynamic testing with an independent Brake
Specialist riding GBRf trains with Ecofret2 wagons in the consist.
The static testing identified an issue during the initial brake application with the
vehicles in a ‘tare’ condition. However, tests found that if the brake pipe is reduced to
4.3bar, instead of 4.6bar, then an ‘Initial’ brake application in ’tare’ would occur.
GBRf are waiting for VTG to confirm the reason behind this issue and proposed
engineering solution, but believes this is down to how the control side of the brakes
(manufactured by Dako) interacts with the brake cylinders (manufactured by Wabtec
Faiveley).
It has also been noted that the Dako supplied regulator has been occasionally
venting air continuously following brake applications between 5.0bar & 4.5bar
(reference ‘Air Flow’ fluctuations). GBRf have failed to repeat this issue on static
testing, but the loss of air was quite obviously occurring on trains containing Ecofret2
wagons. The problem was reported to be alleviated by making a more substantial
brake application of around 4.3bar.
This affects vehicles starting with the following number range: FWA C
704520xxxx (end platforms) and 704521xxxx (centre platforms).
Action Required
Therefore whilst operating an intermodal services which have Ecofret2 wagons within
the consist, see number range above, the following applies:
in order to facilitate an initial brake application drivers MUST reduce the Brake
Pipe Pressure to 4.3 Bar. (this includes when required to carry out a Train
Manager only brake continuity test), and
If any unusual fluctuations of the Air Flow gauge are experienced, this should be
rectified by use of applying ‘Overcharge’ in the first instance. If this fails to
stabilise the fluctuation, the train should be inspected in line with existing
operating procedures.
NOTE: This issue does NOT affect the vehicles ability to develop
its full brake force potential.
Issued by:
GBRf Standards & Training Dept. 1 of 1 May 2021
Report 14/2023 71 December 2023
London Gateway
This page is intentionally left blank
Report 14/2023 72 December 2023
London Gateway
This report is published by the Rail Accident Investigation Branch,
Department for Transport.
© Crown copyright 2023
Any enquiries about this publication should be sent to:
RAIB Email: enquiries@raib.gov.uk
The Wharf Telephone: 01332 253300
Stores Road Website: www.raib.gov.uk
Derby UK
DE21 4BA