Signalling Maintenance Benchmarking Report

BENCHMARKING STUDY
MAINTENANCE OF MODERN SIGNALLING

SYSTEMS
SEPTEMBER 2013
Benchmarking Analysis & Reviews
Authors
Name Directorate Title

Aydin Sheibani S&SD Principal, Benchmarking Analysis & Reviews
Signalling maintenance Page 2 of 57

UNDERGROUND
Confidential and Commercially Sensitive

CONTENTS
1 Executive Summary ..................................................................................................... 4

2 Introduction .................................................................................................................. 7
3 Study participants ......................................................................................................... 7
4 Approach and methodology .......................................................................................... 9
5 Structural and explanatory factors .............................................................................. 11
6 Cost analysis and benchmark range........................................................................... 40
7 Conclusions................................................................................................................ 45
8 Recommendations ..................................................................................................... 47
Appendix 1 .......................................................................................................................... 48
Appendix 2 .......................................................................................................................... 49
Appendix 3 .......................................................................................................................... 56
Appendix 4 .......................................................................................................................... 57
Signalling maintenance Page 3 of 57

UNDERGROUND

1 Executive Summary
1 This report summarises the results from the study commissioned by London
Underground into the maintenance of modern signalling systems. The principal aim of
the study is to gain a fuller understanding of the approach taken by organisations in
the maintenance of these modern systems.
2 The study was conducted as Type 1 case study with all CoMET and Nova metros
invited to participate. The data was gathered through a detailed questionnaire,
followed up with a series of clarification questions and teleconferences with the
participants.
3 Data has been gathered for 12 lines from 6 CoMET and Nova member metros across
Europe and Asia that have installed modern ATO systems:
(a) London (2 lines) – Victoria line and Jubilee line
(b) Barcelona (1 line group) – line 9/10
(c) Paris (2 lines) – Line 1 and Line 3
(d) Guangzhou (5 line groups) – Line 1, Lines 2 & 8, Line 3, Lines 4 & 5 and
Guangfo line
(e) Taipei (1 line) – Wenhu line
(f) Hong Kong (1 line) – West Rail line
4 In order to better understand the cost drivers and allow a more meaningful comparison
to be made, the analysis was undertaken on the following basis:
(a) Firstly the asset base, system characteristics, performance requirements and
any other structural or explanatory factors were understood;
(b) Secondly the resultant maintenance regime to meet these requirements and
the subsequent organisation structure, including the different types of labour
resource were compared; and
(c) Finally the labour rates were looked at along with other costs and minor capex
to understand the overall cost profile for each metro and link these to the
structural and explanatory factors that help explain the differences
5 The base year for all data collected in the study is 2011/12 (or 2011 calendar year)
unless otherwise stated.
Maintenance of modern signalling systems Page 4 of 57

UNDERGROUND

6 The key findings from the study are as follows:

(a) London has a significantly higher city wage rate than the home cities of the
other metros in the study group. This was tested and shown to have a positive
correlation with maintenance cost;
(b) Metros that have built an incident response organisation to resolve faults more
swiftly have higher costs. The London lines perform better than the rest of the
study group, with only Hong Kong achieving a comparable Mean Time to
Repair (MTTR);
(c) The use of predict and prevent and in particular remote condition monitoring
technology has been proven to greatly reduce the number of incidents;
(d) A significant number of signalling failures are protected in systems that have
greater redundancy built-in. This can account for c. 10% of failures;
(e) Operational staff are involved in the first line response and can often resolve
signalling failures on Paris Metro L1 & L3, Barcelona Line 9/10 and Hong
Kong WRL. This helps metros reduce the time taken to resolve incidents;
(f) Lines with a higher density of lineside assets have higher maintenance costs;
(g) During failure it is better to replace a subset of assets rather than target the
specific component in order to resolve the issue as fast as possible, for
example removing a sub rack from train borne equipment. This reduces the
time taken to diagnose the issue.
(h) Shorter lines tend to have a higher unit cost per track km, suggesting an
element of fixed cost is involved;
(i) There is a relationship between intensity of service and MTTR, with metros
running a more trains per hour (TPH) having a lower MTTR. Victoria line has
the highest TPH across the peer group and the shortest MTTR;
(j) Some metros have third line support which is provided by a central workshop,
for example in Hong Kong and Taipei. These activities have been excluded
from the submitted costs;
(k) Hong Kong metro are performing the same maintenance activities less
frequently than on the Jubilee Line without compromising on the safety or
performance of the asset. Both lines have the same signalling system.

UNDERGROUND

(l) Metros can reduce failures by adapting the maintenance regime so that the
frequencies of inspections/maintenance are based on the criticality of asset
locations.
(m)Hong Kong have a similar system of approval for changes to the maintenance
regime as London but test the benefit of the change first as a trial before
rolling out to the whole line
7 The cost of maintenance varies significantly across metros and lines, but adjusting the
costs to align with London city wage rates, gives the following results:
8 Given the key findings and conclusions from the study, the following actions are
recommended:
(a) Discuss through teleconferences or visits, how other metro’s use of multi-
skilled operational staff to respond to and resolve failures is a key contributing
factor to reducing MTTR. Barcelona, Hong Kong and Paris all use operational
staff in resolving incidents;
(b) A discussion to take place with Taipei on how it has used predict and prevent
and in particular remote condition monitoring technology to cut the number
service affecting failures;

UNDERGROUND

(c) Undertake a more detailed comparison of maintenance activities between the

Jubilee Line and Hong Kong’s West Rail Line on Thales SelTrac S40 system
assets; and
(d) Speak to Guangzhou about the level of redundancy in their systems and how
this has helped prevent a significant proportion of failures;
2 Introduction
9 London Underground (LU) has installed CBTC signalling on the Victoria and Jubilee
Lines and is in the process of installation on the Northern Line and Sub-surface
network (Metropolitan, Circle, Hammersmith & City and District Lines).
10 The implementation of a new generation of technology creates an opportunity for a
substantial re-evaluation of the approach to signals maintenance. For example:
(a) There will be significant reduction on the amount of line-side equipment,
especially mechanical equipment.
(b) Both line-side and on-train equipment will become more standardised.
(c) The remote diagnostic capabilities of modern signalling systems should
enable better preventative and corrective intervention, with the need for fewer
locally based call staff, other than at critical locations.
(d) The characteristics of this technology mean that future skill and capability
requirements will be biased towards computer and control systems, providing
opportunity to introduce a more diverse workforce at lower cost
11 Therefore LU has undertaken this study with the main objective of getting a fuller
understanding of the approach taken by other organisations in the maintenance of
modern signalling systems.
3 Study participants
12 The following metros participated in the study:

UNDERGROUND

13 The study participants have submitted data for the following lines.
(a) London (2 lines) – Victoria line and Jubilee line
(b) Barcelona (1 line group) – line 9/10
(c) Paris (2 lines) – Line 1 and Line 3
(d) Guangzhou (5 line groups) – Line 1, Lines 2 & 8, Line 3, Lines 4 & 5 and
Guangfo line
(e) Taipei (1 line) – Wenhu line
(f) Hong Kong (1 line) – West Rail line

UNDERGROUND

4 Approach and methodology
14 The members of the study group were asked to submit data in the form of a
questionnaire (attached as Appendix 1) and more specifically around the following key
areas:
(a) System Characteristics – high level summary of quick facts about the metro
e.g. number of stations, track length etc. This information has already been
gathered.
(b) Asset Base – a breakdown of the modern signalling equipment as well as any
legacy assets
(c) Maintenance – maintenance organisation, frequency and costs including
incident response.
(d) Minor Capex – a breakdown of the minor capex programme over 10 years
(e) Performance – service affecting failures (SAFs), mean distance between
failures (MDBF) or mean time between failures (MTBF) and mean time to
repair (MTTR)
15 This was followed up with a series of clarification questions and teleconferences with
the participants to understand the level of information presented. What follows is a
summary of the information collected and the subsequent analysis carried out to
provide a meaningful comparison.
16 The base year for all data collected in the study is 2011/12 (or 2011 calendar year)
unless otherwise stated. A summary of the data completeness achieved for each
question is shown as Appendix 2 and the full set of submissions by the metros as
Appendix 3.
17 Before cost comparisons were undertaken, the cost data was first normalised for
Purchasing Power Parities (PPP). PPP normalisation takes account of the different
currencies and levels of purchasing power in each country. This is based upon a

UNDERGROUND

basket of comparable goods and services across countries. Therefore all costs shown
are in $US and PPP normalised. The factors 1 used for each metro are shown below:
18 The approach taken is to understand the maintenance requirements from the bottom
up, in order to better gauge the cost drivers. This is illustrated in the figure below.
Asset Base
• Modern
Assets (CEUs) Performance Structural and
• Legacy assets requirements Explanatory
& failure factors
rates
• SAFs, MDBF,
MTBF, MTTR Other
• Safety structural and
• Incident explanatory
response • Employer
• Reliability • System type contributions
growth • Access Maintenance • Materials
• Legacy issues Regime • Other
Overheads
• Inspections Org & Other
• Preventative
Structure Costs
• Reactive
• Corrective • Incident TO
• Maintenance Labour
TO
• Point Fitters Costs
• Supervisors
• Other
Cost
Minor Capex
profile
1
Source: World Bank PPP conversion International $2010

UNDERGROUND

19 With this in mind, the analysis has been undertaken in the following order:
(a) Firstly the asset base, system characteristics, performance requirements and
any other structural or explanatory factors were understood;
(b) Secondly the resultant maintenance regime to meet these requirements and
the subsequent organisation structure, including the different types of labour
resource were compared; and
(c) Finally the labour rates were looked at along with other costs and minor capex
to understand the overall cost profile for each metro and link these to the
structural and explanatory factors that help explain the differences.
5 Structural and explanatory factors
20 Structural factors are network characteristics that will have an impact on cost levels
but are considered unchangeable by the organisation, or perhaps only changeable in
the long-term.
21 Explanatory factors are defined as network characteristics that also have an impact on
cost levels but can possibly be influenced in the short to medium term. These are
potentially opportunities for improving efficiencies.
22 These structural and explanatory factors are set out in the remainder of this section of
the report. For each of these factors the relationship between cost has been measured
and regression analysis undertaken to determine its significance.
Regression
23 Due to the limited number of data points in this study multivariate regressions were not
used to assess the impact of several structural factors simultaneously, therefore
effects of single drivers were measured instead.
24 This approach is limited in that maintenance costs are typically influenced by a large
number of factors with complex interactions and it is not always possible to isolate the
impact of a single factor. In other words, a particular structural factor may have a
relatively minor cost impact; however this impact may be masked by other factors and
not show through in a correlation against total maintenance cost. Equally, a factor that
shows a correlation with total maintenance cost may not necessarily be a strong
influence. Therefore the analysis has been backed up with a good measure of
engineering judgement that could explain any correlation.
Asset Base and system characteristics

UNDERGROUND

25 The table below shows an overview of the signalling systems on the lines. A full
diagrammatic illustration of the asset base for each metro is shown in Appendix 4.

UNDERGROUND

26 In a modern CBTC system, the signalling asset base is significantly different from
older, traditional colour light signals, with a higher proportion of the equipment being
centrally located in control or equipment rooms. These locations provide more
protective operating environments and will allow easier access and reduced time for
maintenance. Nevertheless there are a significant number of trackside assets that
need to be maintained.
27 In order to compare the trackside asset base of modern CBTC systems and to account
for the varying complexities of different systems and lines, LU has developed a
methodology for categorising the assets into a unit of measure called CBTC
Equivalent Units (CEUs). An example of CBTC architecture is set out below:
28 Based broadly on the concept of Signalling Equivalent units (SEUs), which is already
used in the industry, the trackside infrastructure of CEUs can be defined as:
(a) a radio base station or a loop base station (counting as a single unit
regardless of the number of antennas or leaky feeder sections fed from the
base station)
(b) a signal head (counting as a single unit regardless of the number of aspects,
subsidiary signals, route indicators and point position indicators);
(c) a controlled point end (counting as one for each machine in complex point
arrangements, but ignoring back drives. Derailers are included and wide to
gauge points count as one point end);

UNDERGROUND

(d) plungers / staff protection key switches (counting as 1 CEU even where
operated in pairs)
(e) any other attribute that requires a particular control function.
29 The following graph shows density of CEUs by track km for the metros:
30 As the graph illustrates, the CEU densities of the signalling systems maintained on the
Victoria and Jubilee lines are broadly comparable to the external peers. Guangzhou
has both the highest and lowest density with Line 1 at just over 4.7 CEUs per km and
Line group 4&5 at just under 2.5 CEUs per track km.
31 Plotting the CEU density against cost per track km (in $k, PPP normalised) and
conducting the regression analysis shows a positive correlation between the two (R2 =
0.263 or 26%) as illustrated in the graph below:

UNDERGROUND

32 In other words, the higher the density of CEUs the higher the cost. This is consistent
with engineering judgement, since more assets result in more maintenance and a
greater likelihood of failure. Furthermore the CEU count is also a good indicator of the
complexity of the line, so a line with a greater proportion of junctions and crossovers
will typically have a higher CEU count than a simple straight line of equal length.
Key finding: Regression analysis shows a positive correlation between CEU density
and signalling maintenance cost per track km.
33 Next, the analysis of the asset base turned to the relationship between track km and
cost. The graph below simply shows the mainline track km plotted against the cost per
track (in $k, PPP normalised).

UNDERGROUND

34 As illustrated in the graph above, the trend and regression analysis shows an
appreciable negative correlation (R2 = 0.225 or 23%) between track length and unit
cost; the longer the length of the line, the lower the unit cost. This is consistent with
engineering judgement to a certain degree as there will invariably be an element of
fixed cost associated with managing any line. However the element of cost involved in
the signalling maintenance that can be classed as fixed has not been explored in this
study and therefore cannot be quantified.
Key finding: Regression analysis shows a negative correlation between line length
and signalling maintenance cost per track km.
Performance
35 Without much of the electro-mechanical track-side equipment (signals, train stops,
etc.) historically responsible for many signalling failures and through in-built
redundancy, improved detection of system deterioration, greater diagnostics and

UNDERGROUND

remote intervention capabilities; modern CBTC systems will generally have a higher
level of reliability.
36 However, failures of signalling assets do still remain a significant cause of incidents,
with metros attributing 12% of all failures to this asset group. Therefore performance of
the assets and the level of response required is a key explanatory factor for total
maintenance cost.
37 It is important to be aware that performance can be measured in a number of ways
and that metros will have different areas of focus for this. For example, at LU lost
customer hours (LCH) are used as the main measure of impact of incidents on
passengers. Although this is the most passenger-focused measure of delay it is also
the most difficult to measure and therefore not currently used by other metros.
38 One measure which is more readily available is the number of Service Affecting
Failures (SAFs) which is defined as:
Incidents where the passenger train service is delayed
Service Affecting Failures
by 5 minutes or more due to the failure of a signalling
(SAFs):
asset.
39 The graph below shows the number of SAFs per track km for signalling assets, per
annum for the study group:
Performance - SAFs per track km

1.2 BETTER WORSE
1
0.8
0.6
0.4
0.2

UNDERGROUND

40 The higher the number of incidents per track km, the worse the performance of the line
is. As shown from the graph above, the LU lines are toward the worse end of the
scale, with only Barcelona having more incidents per track km. Line 3 in Guangzhou
and Taipei’s Wenhu line have the fewest faults per track km.
41 Taipei’s success in minimising the number of SAFs is partly a result of their use of a
“predict and prevent” approach as opposed to “find and fix”. This has been provided
through the increased use of remote condition monitoring technology. For example, a
signalling operation recording system has been installed at critical stations that
highlight error messages that provide early warning of faults.
42 As a result there has been a 70% reduction in the number of SAFs since 2009.
Key finding: The use of predict and prevent and in particular remote condition
monitoring technology has greatly reduced the number of incidents on the Wenhu
Line.
43 It should be noted that the introduction of train attendants has also helped reduce the
number of incidents in Taipei. There had been many system instability incidents since
its launch in July 2009. Therefore, it was decided to staff the trains with train
attendants for constant real-time monitoring, to report train status, and provide
immediate troubleshooting when necessary, and in order to reduce reaction time in the
event of failure. The presence of attendants has improved system stability
considerably.
44 Although the system has gradually stabilised since, train attendants are still able to
support the operational control centre during incidents, help monitor track status and
provide necessary passenger service. Therefore, their presence on the trains is still
necessary.
45 In Guangzhou, Line 3 is its best performing line, primarily due to its newer northern
section which has seen performance improve dramatically since it opened in 2010.
The main source of failures is the train-borne signalling equipment (VOBC) which for
some fault types is protected by redundancy, so if one VOBC fails it can switch to a
back-up device. Guangzhou estimates that c 10% of failures are prevented in this way
Key finding: Using built in redundancy in the system has proven to reduce the
number of service failures for train borne equipment on Guangzhou’s line 3.
46 Similarly in Hong Kong, the WRL has similar redundancy built into its VOBC systems.
In addition, the wayside station control systems (for vital point control) in the running

UNDERGROUND

line have a dual redundancy design (redundant Intersig) for availability. Each Intersig
has dual processors which must be in agreement for safety.
47 However, SAFs do not necessarily reflect the impact of incidents on passengers and it
is important to understand how metros deal with asset failures. One way to look at this
is to analyse the time taken to fix the issue, or the mean time to repair (MTTR), defined
as:
Total hours that assets are unavailable following

Mean Time To Repair (MTTR) = SAFs
Number of SAFs
48 The graph below shows the MTTR for signalling assets across the study group:
49 In this measure, the LU lines perform better than the rest of the peer group, with only
Hong Kong delivering a comparable MTTR. This suggests that although these lines
may have a higher number of incidents to deal with, they are handling them better and
minimising the disruption caused to passengers.

UNDERGROUND

50 On the Jubilee Line, in order to minimise downtime during failure of train-borne

signalling equipment (VOBCs), a complete ‘sub-rack’ is removed rather than spending
time trying to diagnose the particular faulty electronic card within the sub-rack. This is
then sent to the second Line maintenance team for testing. This has helped reduce
significantly the duration of incidents involving VOBCs.
Key finding: During failure it is better to replace a subset of assets rather than target
the specific component in order to resolve the issue as fast as possible.
51 On Hong Kong’s WRL, the trackside inductive loops have been a source of delays.
Although they do not have redundancy built in, they do have self monitoring control
signals injected into them to allow the maintainers to be advised if the loop is
damaged. This can often be a problem with loops particularly during planned track
work.
52 In Barcelona, L9/10 station staff (TOLA's) are trained to resolve train and signalling
incidents as swiftly as possible. For example, TOLAs can secure points, and if a train
is stranded they can board the train and drive it manually. This can greatly reduce the
duration of the failure.
Key finding: The use of multi-skilled operational staff in responding to and resolving
failures is a key contributing factor to reducing MTTR, particularly on UTO lines.
53 Generally some incidents are quicker to fix than others, particularly in modern systems
where a higher proportion of the equipment is centrally located in control or equipment
rooms. So a different mix of incidents could be another reason for the variation
between lines.
54 The necessity for a quick response may be dependent on the intensity of the service in
the area and the number of passengers likely to be affected. To understand this
relationship better, the number of journeys made per track km and the maximum
number of trains per hour (tph) were analysed by line:

UNDERGROUND

Passenger journeys (million journeys/ track km)

5.0
4.5
4.0
3.5
3.0
2.5
2.0
1.5
1.0
0.5
0.0
55 Guangzhou’s Line 1 and the Victoria line are the busiest in terms of passengers per
track km, therefore any failure of the signalling assets would impact a significant
number of passengers, even if localised. The least busy lines are Guangzhou’s
Guangfo line and Barcelona’s line 9/10. Both have recently opened and only partially
so will have increased patronage as the construction of the remaining sections are
completed.

UNDERGROUND

56 Both the Jubilee and Victoria Lines are toward the top end of the range, meaning that
they run a more intensive service than most of the external peers during the peak
hour. This is the critical time for dealing with performance and it is important to test
whether the MTTR shown earlier has any relationship with the maximum tph.
MTTR versus tph

35
30
25
R² = 0.1195
20
15
10
0
0.2 0.3 0.4 0.5 0.6 0.7 0.8
MTTR versus tph Linear (MTTR versus tph)
57 The regression analysis shows some relationship between MTTR and tph; the higher
the tph, the lower the expected MTTR. Although the correlation is not shown to be so

UNDERGROUND

strong (R2 = 0.1195 or 12%) it is consistent with engineering judgement. As discussed

earlier, in the regression analysis a particular factor may show a relatively minor
impact; however this impact may be masked by other factors and not necessarily show
through in a correlation when tested.
58 The more pertinent question is whether having a shorter MTTR has any impact on
maintenance cost. The relationship between the two is shown in the graph below:
Cost versus MTTR

160
140
120
100
80
R² = 0.2279
60
40
20
-
0.3 0.35 0.4 0.45 0.5 0.55 0.6 0.65 0.7
Cost versus MTTR Linear (Cost versus MTTR)
59 This clearly shows an appreciable negative correlation (R2 = 0.2279 or 23%) between
MTTR and total maintenance cost; the shorter the average time taken to repair a fault
the higher the cost. This does intuitively seem a reasonable hypothesis as shorter
MTTR is likely to be related to a larger incident response presence across the line.
Key finding: Regression analysis shows a negative correlation between MTTR and
signalling maintenance cost per track km.
Maintenance regime
60 The asset base and the required level of performance will largely drive the
maintenance regime. This section looks at how metros deliver this work, focussing
primarily on the level and frequency of activities.

UNDERGROUND

61 Metros were asked to provide data under the following categories:

(a) Inspections – Planned non-intrusive examinations of assets or sub systems;
(b) Preventative maintenance – planned activities that are part of the general
maintenance of assets. Also included in this category are programmed
replacement of assets (routine change);
(c) Corrective maintenance – the maintenance activity carried out after the
reactive maintenance of an asset following failure, or discovery of a fault after
preventative maintenance; and
(d) Reactive/Emergency maintenance – unplanned maintenance following the
failure of an asset. Thus the resource requirement pursuant to Reactive
Maintenance is driven by the number of failures and the time required for the
subsequent maintenance. This will be the incident response team required
and is discussed in the next section
62 There are many different types of signalling system components and maintenance
intervals vary widely, therefore determining a comparable set of activities is not always
easy between lines. With this in mind, the analysis compared the frequency of
inspections for line-side ATO/ATP equipment as shown in the graph below:

UNDERGROUND

Maintenance frequencies - lineside assets(weeks)

30
25
20
15
10
63 As shown from the graph, the London metros have some of the shortest cycles among
the study group although the majority of lines tend to have frequencies at 12-13
weeks. The stand out metros are Hong Kong WRL and Paris Lines 1 and 3 which
have frequencies at 26 weeks. The regression analysis showed no correlation
between inspection frequency and cost, which is to be expected as this is only a high
level representation of the level of activities.
64 It is more pertinent to drill down to a lower level of detail so we can compare
maintenance frequencies between lines for a basket of different assets and activity
types. This is only possible for lines that have the same technology and hence a
comparable set of activities. The table below compares a basket of key activities
between two lines that have the Thales SelTrac S40 signalling system; namely Jubilee
Line and Hong Kong’s WRL:

UNDERGROUND

65 The breakdown here shows how some activities on the Jubilee Line, which mirror the
frequencies stipulated in Thales maintenance manuals and that have subsequently
been absorbed into LU standards, are done less frequently on Hong Kong’s WRL line.
Since the system was installed on the line in 2009, Hong Kong’s experience has made
them aware that the system can still function well with these longer intervals, without
compromising the safety or performance of the asset.
Key finding: Hong Kong metro are performing the same maintenance activities less
frequently than on the Jubilee Line without compromising on the safety or
performance of the asset.
66 In Hong Kong, a change of maintenance regime can be triggered by a number of

things, including:
(a) Service performance review, RCM study or Cost-benefit analysis;
(b) A major incident and subsequent study recommendation; or

UNDERGROUND

(c) New practice introduced by legislation or imposed by statutory authorities.

67 Before the change, they will conduct safety and reliability validation. If this involves a
significant change of the maintenance of a safety critical item, they need to do a formal
review of safety impacts and submit this to the Safety Technical Committee and
Technical Standards & System Integration Committee for approval.
68 They will then first perform a trial run and implement the change if the trial run result is
successful.
69 At LU, any proposed change must go through the Directors Risk, Assurance and
Change Control Team’s (DRACCT) process. The DRAACT meet every 4 weeks and
has delegated authority from the LU executive to consider, agree or authorise any
changes to plans. The criteria for the route through the DRACCT process is set out
below:

UNDERGROUND

Key finding: Hong Kong have a similar system of approval for changes to the
maintenance regime as LU but test the benefit of the change first as a trial before
rolling out to the whole line.
70 An issue that exists on some of the older lines is that there are legacy assets which
remain after a signalling upgrade that require a disproportionate level of maintenance.
Indeed, one of the most challenging aspects of setting the maintenance regime is
managing the requirements of these assets alongside the modern systems. For
example, critical points which were not renewed after the signalling upgrade on the
Jubilee and Victoria lines are amongst the assets that require the most amount of
attention and resource.
71 We’ve discussed earlier how metros consider points as a major source of signalling
incidents; therefore it’s important that where possible the maintenance regime is used
to limit these failures.
72 On the WRL in Hong Kong, points in critical locations are inspected and maintained
more frequently than all other points. This targeted maintenance has helped reduce
the failures at these locations, which often cause the longest delays. This is also the
case on the Jubilee and Victoria lines.
Key finding: Metros can reduce failures by adapting the maintenance regime so that
the frequencies of inspections/maintenance are based on the criticality of asset
locations.
Maintenance organisation
73 Metros were asked to provide details of their organisation structure and how they were
set up to deal with signalling maintenance of modern systems.
74 Firstly, analysis was undertaken on the relative size of the organisations set up to deal
with signalling. The graph below shows the number of Full Time Equivalent (FTE) staff
per track km for each line:

UNDERGROUND

FTEs per track km

2.5
1.5
0.5
75 As shown, the London metros are broadly in the middle of the range with Barcelona
and four of the Guangzhou lines having more FTEs per track. Taipei has the fewest,
but this is partly explained by additional maintenance support that they get from the
signalling supply contractor as part of the warranty. The headcount for these are not
included here.
76 Additional analysis was undertaken to understand the relationship between the
number of FTEs and cost but the regression did not show any strong positive
correlation.
77 What follows below is an overview, by metro, of how the maintenance organisation is
set-up, what activities it performs and how it is organised to deal with incident
response.
Jubilee Line
78 On the Jubilee line, first line maintenance is provided by its DLO signals engineers.
79 Second line support is provided by the Signals DLO TBTC support team and the
Signals computer section. Additional contracts have been let for second line support

UNDERGROUND

for the C&I assets. Third line support consists of component level repair of defective
equipment once it has been identified by either first line or second line processes.
Third line services are provided by the Equipment Workshop, Westinghouse and
Thales.
80 Jubilee line is part of the Jubilee, Northern and Piccadilly Line group (JNP) in London
that has a separate incident response team. These are grouped by geographical
location rather than by line as shown below:
81 All reported signalling faults, are recorded by the JNP Control Centre and directly
communicated to the Signal Systems Duty Manager (SSDM). The SSDM is
responsible for the fault response and safe repair of the asset. The details of the
reported incident are recorded on a fault report form and a duty Technical Officer is
dispatched to the site to commence the initial investigation.
82 Technical Officers liaise with local operational staff and ensure that the Signal
Operations Manager is fully aware of the progress in identifying and rectification of the
fault. Once the fault has been rectified the Technical Officers notify the SSDM who
notifies the Line Controller and ensures the fault is closed with the control centre.
83 For significant faults or incidents causing a 15-minute delay or more the SSDM
completes a timed chronology of events relating to actions taken by themselves and

UNDERGROUND

staff under their control during the incident. This report may be used for further
investigation. Items recorded include:
(a) Details of the original report: its source, time and content
(b) Assignment details for each supervisor or technician involved
(c) A report from the signal technician who attended the failure, including
particulars of the fault, the cause as perceived, remedial measures taken and
the key timing of events
(d) Follow-up reports from supervisors, specialist engineers and second line
maintenance or repair agencies where appropriate
(e) Final conclusions as to the underlying cause
84 Further explicit information is gathered when there has been an incident involving an
allegation against the integrity of the signalling system.
85 Significant issues arising as a result of signal failures are discussed every period
during the Signal Manager's team meeting. From the discussion any actions are
formulated and agreed and the relevant operational manager is given the task.
Victoria Line
86 On the Victoria line, first line maintenance is provided by its Direct Labour
Organisation (DLO) signals engineers. Second line support is provided by Signals
ATC support team. The ATC team is currently dedicated to Central and Victoria Lines.
Third line support consists of component level repair of defective equipment once it
has been identified by either first line or second line processes. Third line services are
provided by LUL Equipment Workshop, Bombardier and Invensys.
87 Technical Officer (TO) depots deal with Signalling and Point failures. A response Unit
is located at each TO depot at any one time however some TO depots are not manned
during weekends. There are three depots on the Victoria Line.
88 A similar process for reporting signalling faults is used on the Victoria Line as
described above for the Jubilee Line.
Hong Kong – West Rail Line

89 The signalling team are responsible for the maintenance of the Central and Wayside
equipment on the West Rail Line between Hung Hom and Tuen Mun stations and the
signalling equipment inside Pat Heung Depot.

UNDERGROUND

90 The on-board Signalling equipment (VOBC) is maintained by the Rolling Stock

Maintenance Department (RSMD).
91 For preventive maintenance and corrective maintenance of trackside and central
signalling equipment, first line maintenance is performed by the Rapid Response Unit
(RRU) and Signal maintenance team. Second line maintenance is provided by the
Signal maintenance team. Adhoc support is provided by the Signal and Control
Maintenance Support Unit if required. Third line support consists of component level
repair of defective equipment once it has been identified by either first line or second
line processes. Third line services are provided by the Central Electronic Workshop,
Point Workshop and Thales.
92 The chart below sets out the high level structure:
93 There is a separate Incident Management organisation for the West Rail Line which
consists of a Fault Report Centre (FRC) and a Rapid Response Unit (RRU).
94 They provide first response to failures of all infrastructure equipment (Signalling,
Telecom, Radio, Network, Fibre Optic Transmission System, Pway, Power
Distribution, Civil) along the whole West Rail line and also the central equipment of
other systems (MCS, PIDS, PAS, CCTV) that impact on safety or the train service.
The FRC team also mans the VCC terminals in the Operations Control Centre (OCC)
to execute the required commands to maintain the train service. The cost of these
activities are not included in the signalling maintenance unit costs.
95 WRL’s incident management process for signalling is set out as follows:

UNDERGROUND

(a) All reported faults of assets are recorded by the FRC which will then create a
reactive maintenance work order;
(b) The RRU is dispatched to site immediately for investigation and recovery
action. For equipment located inside the Central Equipment Room, FRC will
perform the initial investigation and recovery action;
(c) FRC will coordinate with RRU to liaise with the Traffic Controller and then
Station Officer. The technical officers from the relevant disciplines are also
informed to provide advice or support if required;
(d) FRC will also provide immediate engineering advice on handling incidents to
the Chief Controller and Traffic Controller of the OCC. Moreover, FRC will
execute the required SMC workstation/VCC terminal command for recovery;
(e) The supervisors of the respective disciplines will close the Reactive
Maintenance Work Order after the fault is rectified;
(f) For incidents causing a 5-minute accumulative delay or more, a Technical
Report is prepared with details of the events, maintenance history, cause of
the incident, immediate action taken, conclusion and recommendation for
follow up.
Taipei – Wenhu Line

96 The Wenhu Line (also called the Brown Line) is comprised of two sections; the
Wenshan and the Neihu. On the Wenhu Line, wayside and on-board signalling assets
are maintained by the company’s Medium Capacity Transit Division (MCTD). Central
Control Assets are maintained by the Information Technology Division. The
organisation chart below shows how the signalling engineers are organised in the
MCTD.

UNDERGROUND

Muzha Signalling group (about 12

Signaling & persons)
Neihu Signalling group (about 12
Communication
persons)
Workshop (about 55
Electrical Engineering persons) Muzha Communication group
Medium
Plant (about 110
Capacity Neihu Communication group
persons)
Transport Power Workshop
Division
(about 650 Track Workshop
persons) other
Rolling Stock Plant
Operations Center
other
97 The line is split into two teams (staff are based either at the Muzha depot or the Neihu
depot) and maintenance personnel are on duty 24 hours a day. Four people are
required for one shift, with three shifts in day. A total of 12 people are needed for a 24-
hour operation. There is no separate incident response team, and signalling failures
are dealt with by the standby maintenance personnel. In addition Wenhu Line is still
under its warranty period, therefore signalling maintenance work is done together with
the contractor’s maintenance personnel.
98 Wenhu Line’s incident Management process is as follows:
(a) After the system receives the news of a failure, the Central Control Operator
classifies the incident and asks a member of maintenance staff to deal with it.
(b) The member of maintenance staff then goes to the site of the incident to repair
the equipment immediately. If the problem cannot be resolved immediately,
other units in the company are asked to assist. During the incident, Taipei
TMRT will deploy a free bus to carry passengers if the incident is of significant
length.
(c) Meetings are held to review incidents causing delays of more than 90 seconds
every week. Any delays of greater than 5 minutes, are reviewed immediately
after the event and proposed improvements are monitored closely.
Barcelona – Line 9/10
99 Barcelona have a full time maintenance outsourced contract, with 24/7 response
availability. There is also an internal workforce in place to manage this contract and its
structure is shown below:

UNDERGROUND

Superstructure Service
Manager
Automatic line STC Maintenance

responsible
Maintenance Engineering
L9 Maintenance coordinator (T1) L9 Maintenance coordinator (T2) L9 Maintenance coordinator (T3)
100 The contractor employs 11 full time maintenance staff who also cover incident
response. As discussed earlier, L9/10 station staff are also trained to resolve train and
signalling incidents, so in many cases they may be the first line response.
Paris Lines 1 & 3

101 In Paris, signalling maintenance staff are not dedicated to a specific line but to a pool
of lines so are able to work flexibly across a number of different lines, with the
exception of line 14. In addition Line 3 has 85% of its maintenance work contracted
out.
102 Paris has a separate incident response unit, which is four teams of two or three people
in each that deal with signalling failures on all 14 of its lines across 405 km of track.
Guangzhou
103 Guangzhou metro has an internal signalling maintenance organisation, of around 600
people, that is split into four line groups. Staff are licensed to work across different
lines with different signalling systems if required, allowing a greater level of flexibility.
104 The high level structure of its signalling maintenance organisation is shown below:

UNDERGROUND

The signal
maintenance
department
Section 1 Section 2 Section 3 Section 4
being being being being

responsible for responsible for responsible for responsible for
line 1/Guangfuo line 2/8 line 3/APM line 4/5
105 There is a separate response unit that is dispersed across the network, depending on
the critical locations. Generally, staff are based at terminus stations and depots as well
as the network operations centre. During traffic hours, maintainers will undertake
inspections of signalling assets in equipment rooms whilst waiting for an incident.
Other factors
106 System type: Different signalling systems have different technologies and as a result
different levels of inherent reliability. Lineside and centralised technologies have
different maintenance requirements. Although the impact of this is minimised to a
certain extent given that this study covers only modern signalling systems, there will
still be differences. This factor has not been assessed.
107 Access: The table below shows the amount of weekly lineside hours available for
signalling maintenance by line. This shows that Jubilee and Victoria lines are at the top
of the range for lineside access, with Guangzhou, Hong Kong and Taipei at the
bottom. There does not appear to be any correlation between access and costs. This
has been tested using the regression analysis.

UNDERGROUND

108 Minor capex: defined here as a regime of minor replacement of components,

individual assets or campaign change of asset types planned to address emergent
safety, asset condition, reliability or obsolescence requirements. Metros were asked to
submit data for an 8 year period. As there were too few data points available for this
area no further analysis was undertaken.
Cost of labour
109 The next explanatory factor that was examined is labour cost. The cost of labour is
often cited as a key factor in understanding the cost gap between metros. Costs have
been normalised to account for the varying buying powers of countries using PPP
conversion factors, however this does not fully address the varying levels of labour
cost between countries; and more specifically it does not address the varying wage
rates between cities.
110 In order to isolate this specific factor, namely the impact that city wage rates have on
overall cost, LU looked at the work done in an extensive 2012 study by UBS to
determine a basket of comparable jobs and the relative cost per hour for these in
major cities across the world. The graph below shows the average city wage rates 2,
PPP normalised, for skilled labour in the five home cities of the study participants.
2
Taken from the 2012 study by UBS on Prices and Earnings and converted to $PPP

UNDERGROUND

Average city wage rates - skilled labour ($PPP/ hour)

30
25
London 2 times higher
20
15
10
0
Guangzhou Hong Kong Barcelona Paris Taipei London
111 As expected, London is the most expensive in terms of city wage rates per hour; more
than twice as expensive as Guangzhou. The other cities have comparable wage rates,
with Taipei the closest to London.
112 This data was then used to measure the relationship between city wage rates and cost
through regression analysis:

UNDERGROUND

Cost versus Labour rates

160
140
120
R² = 0.2809
Cost ($PPP/ t rack km)
100
80
60
40
20
-
10 12 14 16 18 20 22 24 26
City wage ($PPP/ hour)
Cost versus Labour rates Linear (Cost versus Labour rates)
113 As illustrated in the graph above, the trend and regression analysis shows an
appreciable positive correlation (R2 = 0.281 or 28%) between city wages and unit cost.
In other words, the higher the wage rates in the metro’s city, the higher the unit cost.
114 It should be noted that this isolates the impact of city wages only and does not account
for the labour rates of metros
Summary of structural and explanatory factors

115 We can summarise all the structural and explanatory factors discussed in this section
as follows:
R2
There is a clear positive correlation between city
Structural CITY WAGE RATES wage rates and maintenance cost. London has the 28%
highest city wage rates across the peer group.
Lines with a higher density of assets have higher
Structural CEU DENSITY 26%
maintenance costs.
Shorter lines tend to have a higher cost per track
Structural LINE LENGTH km, suggesting an element of fixed cost is 23%
involved.

UNDERGROUND

Lines with shorter MTTR generally have higher

costs. The LU lines perform better than the rest of
Explanatory MTTR 23%
the peer group, with only Hong Kong achieving a
shorter MTTR.
Some metros have third line support which is
Not
Explanatory Workshops provided by a central workshop. These activities
tested
have been excluded from the submitted costs.
Operational staff are involved in the first line
response and can often resolve signalling failures Not
Explanatory Operational Staff
on some lines. These activities are excluded from tested
the submitted costs.
The use of predict and prevent and in particular
remote condition monitoring technology has Not
Explanatory Predict and Prevent
greatly reduced the number of incidents on some tested
lines.
Using built in redundancy in the system has Not
Explanatory Redundancy
proven to reduce the number of service failures. tested
6 Cost analysis and benchmark range
116 With a better understanding of the structural and explanatory factors we can undertake
a meaningful comparison of costs. The graph below shows the cost per track km in $k
(PPP normalised) by line.

UNDERGROUND

117 Victoria and Jubilee Lines in London and lines 9/10 in Barcelona have the highest unit
costs amongst the study group, with Guangzhou L4&5 and Hong Kong’s WRL having
the lowest unit costs.
Adjusting for city wages
118 As discussed earlier, city wage rates are a key structural factor therefore it is important
to understand and isolate the impact of this. LU has developed the following approach,
in consultation with RTSC at Imperial College, for adjusting the costs on all lines.
119 Taking the ratio of wage rates between London and the rest of the cities:

UNDERGROUND

120 And using the % of costs attributable to labour3. An adjustment has been made as
follows:
Adj y = aby + (1-b)y
Where y = cost/km
a = ratio to London
b = % attributable to labour
121 In other words, the costs for lines that are attributable to labour have been adjusted to
align with London city wages. Adjusting these costs using this methodology gives the
following results:
122 The Guangzhou lines have moved the most as a result of the adjustment, with Taipei
having the smallest adjustment due to the fact that it’s city wage rates are similar to
London’s.
3
Where this has not been provided, the average of the metros (75%) has been used as a proxy

UNDERGROUND

123 The graph below shows these adjusted unit rates with the inter quartile range, which
LU has used as an external benchmark range. As the range is derived from metros
with modern signalling systems, it is only relevant to those lines on the LU network
which have automated operation.
124 This gives us an external benchmark range of between $72k and $96k per track km.
Further cost analysis - London
125 The graph below shows the evolution of cost on the Victoria line and how this
compares with the best in range:

UNDERGROUND

126 LU already has plans in place that will bring Victoria line unit costs within the
international benchmark range by 2014/15. Adjusting for the estimated impact of the
structural factors brings the 14/15 unit costs below the best metro in the range. These
have been estimated on the following basis:
(a) Average CEU density on all lines is 3.8 per km. On Vic line it is 4.6 (26%
higher). Analysis shows that the higher the density of CEUs the higher the
cost.
(b) Average MTTR on all lines is 32 min, which is 15 min longer than on Vic Line
(17min). Analysis shows that the shorter the average time taken to repair a
fault the higher the cost.
127 This analysis has also been undertaken for the Jubilee Line:

UNDERGROUND

128 Jubilee Line is currently just above the international benchmark range and will remain
there in 2014/15. Adjusting for the estimated impact of the structural factors brings the
14/15 unit costs with the range. These have been estimated on the following basis:
(a) Trackside CBTC equivalent units (CEUs) per track km. Average CEU density
on all lines is 3.8 per km. On Jub line it is 3.9, higher than the average but
lower than the best in range at 4.3. Analysis shows that the higher the density
of CEUs the higher the cost.
(b) Average MTTR on all lines is 32 min, which is 11 min longer than on Jub Line
(21min). Analysis shows that the shorter the average time taken to repair a
fault the higher the cost.
7 Conclusions
129 There are a number of structural and explanatory factors that need to be understood
before any meaningful comparison between metros can be made. The most significant
of these is the city wage rate. Metros that are in cities with a higher wage rate will
invariably face higher costs. This is something that is outside of the metro’s control;
therefore costs need to be adjusted to reflect this.

UNDERGROUND

130 Adjusting for city wages gives a more balanced picture. However, there will still remain
a number of other factors, particularly network characteristics that are considered
unchangeable by the organisation, or perhaps only changeable in the long-term.
131 Asset density, measured in this study as CEU density, is one such factor. Metros with
a higher density of lineside assets tend to have higher maintenance costs. This is not
something the organisation can change in the short term but should be considered
when making whole life cost decisions for future line upgrade projects.
132 Another such factor is the system type. Different signalling systems have different
technologies and as a result different levels of inherent reliability. Although the
variance in this factor is somewhat minimised as this study covers only modern
signalling systems, there is nonetheless some significant differences. These are hard
to quantify and beyond the scope of this study.
133 There are, of course, factors that a metro can have more influence over in the short to
medium term. These are potentially opportunities for improving efficiencies. Inspection
and maintenance frequencies fall under this category. However, given there are many
different types of signalling system components it is not always possible to compare
activities between metros at a high level. It is more pertinent to drill down to a lower
level of detail so we can compare maintenance frequencies between lines for a basket
of different assets and activity types. This is only possible for lines that have the same
technology and hence a comparable set of activities.
134 Performance is another area where a metro can have some degree of influence.
Metros that are set up to resolve faults more swiftly have higher costs. What target
time is set will be a business decision, dependent on a number of factors and there is
inevitably a trade off between performance and cost.
135 We’ve already discussed a number of ways in which a metro can reduce the time
taken for resolution of faults. The use of multi-skilled operational staff in responding to
and resolving failures is a key contributing factor to reducing MTTR, particularly on
UTO lines. Another way is to replace a subset of assets rather than target the specific
component which can often be difficult to find.
136 The most effective way to improve performance though is by reducing how often these
incidents actually occur. The use of predict and prevent and in particular remote
condition monitoring technology has been proven to cut the number of service
affecting failures. Also, in systems that have greater redundancy built-in, a significant
proportion of these incidents are protected so there is no impact on the service. On
some lines this can account for c. 10% of failures.

UNDERGROUND

8 Recommendations
137 Given the key findings and conclusions from the study, the following actions are
recommended:
(a) Discuss through teleconferences or visits, how other metro’s use of multi-
skilled operational staff to respond to and resolve failures is a key contributing
factor to reducing MTTR. Barcelona, Hong Kong and Paris all use operational
staff in resolving incidents;
(b) Discuss with Taipei how it has used predict and prevent and in particular
remote condition monitoring technology to cut the number service affecting
failures;
(c) Undertake a more detailed comparison of maintenance activities between the
Jubilee Line and Hong Kong’s West Rail Line on Thales SelTrac S40 system
assets; and
(d) Consult with Guangzhou about the level of redundancy in their systems and
how this has helped prevent a significant proportion of failures;

UNDERGROUND

Appendix 1
Maintenance of
modern signalling syst

UNDERGROUND

Appendix 2
No Question DATA COMPLETENESS

0% 100%
0.1.1 Name of Line(s)

Number of
0.1.2 stations
Mainline Track
0.1.3 km
0.1.4 Total track km
Number of
trains in service
0.1.5 in the peak
Maximum tph in
0.1.6 the peak
Number of
control sites on
0.1.7 line
Weekly Line
side access
0.1.8 hours
Are operational
staff involved in
any
maintenance
0.1.9 activities?
Number of
passenger
journeys
0.1.10 (Annual)
Type of system
1.1.1 used

UNDERGROUND

Primary system
1.1.2 supplier
System trade
1.1.3 name
Please provide
breakdown of
1.1.4 asset base:
Number of Main
line Points (i.e.
1.1.5 Switches)
Total number of
points
1.1.6 (switches)
Number of main
line Track
1.1.7 circuits if used
Total number of
track circuits if
1.1.8 used
Track-based
ATP/ATO
communication
1.1.9 systems
Number of
devices in track-
bed (e.g.
balises,
1.1.10 magnets)
Number of in-fill
loops on main
1.1.11 line (if used)

UNDERGROUND

Total length of
in-fill loops (if
1.1.12 used)
Radio-based
ATP/ATO
communication
1.1.13 systems
Number of
lineside
antennas (if
1.1.14 used)
Length of 'leaky'
feeder cable (if
1.1.15 used)
Type of control
1.2.1 system
Processing /
1.2.2 control units
Driver interface
1.2.3 units
Operator on
1.2.4 board
Please provide
breakdown of
1.3.1 asset base:
% of your
system with
fault
1.3.2 redundancy

UNDERGROUND

Please provide
details of your
maintenance
organisation
(Including
incident
response) in the
"2.1.1 Org
chart"
2.1.1 worksheet
Number of
2.1.2 FTEs
2.1.3 Hours per week
Maintenance
2.1.4 unit responsible
Maintenance
coordination
(application for
railway track
work and
placement of
personnel for
2.1.5 those)
Maintenance
2.1.6 technician
SAP
2.1.7 management
PRL preventive
2.1.8 action

UNDERGROUND

Refill pieces
and warehouse
2.1.9 management
%
2.1.10 Internal/External
Do you have a
seperate
incident
2.1.11 response team?
Lineside
ATP/ATO
2.2.1 equipment
On-train
ATP/ATO
2.2.2 equipment
2.2.3 Back-up system

Lineside
ATP/ATO
2.3.1 equipment
On-train
ATP/ATO
2.3.2 equipment
Obsolescence,
routine change
2.3.4 of components
2.4.1 Total costs
2.4.2 Inspection costs

Lineside
ATP/ATO
2.4.3 equipment

UNDERGROUND

On-train
ATP/ATO
2.4.4 equipment

2.4.6 Other
Preventive
maintenance
2.4.7 costs
Lineside
ATP/ATO
2.4.8 equipment
On-train
ATP/ATO
2.4.9 equipment

2.4.11 Other
Corrective
maintenance
2.4.12 costs
Lineside
ATP/ATO
2.4.13 equipment
On-train
ATP/ATO
2.4.14 equipment

2.4.16 Other
Reactive
maintenance
2.4.17 costs
Lineside
ATP/ATO
2.4.18 equipment
On-train
ATP/ATO
2.4.19 equipment

UNDERGROUND

2.4.21 Stand-by time

2.4.22 Other
3.1.1 Schedule
3.1.2 Cost
[Please provide
a breakdown of
failures in the
"4.1.1
Performance"
4.1.1 worksheet]
Data Year:
4.1.2 2010/2011
Train-borne
4.1.3 signalling
4.1.4 Lineside
4.1.5 Room based
4.1.6 Total Failures
When did your

new system
come into
5.1.2 service?
Was it an
5.1.3 overlay system?

UNDERGROUND

Appendix 3

UNDERGROUND

Appendix 4

UNDERGROUND

Signalling Maintenance Benchmarking Report

Uploaded by

Document Informationclick to expand document information

Copyright:

Available Formats

Signalling Maintenance Benchmarking Report

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Signalling Maintenance Benchmarking Report

Uploaded by

Copyright:

Available Formats

BENCHMARKING STUDY

MAINTENANCE OF MODERN SIGNALLING

Name Directorate Title

Signalling maintenance Page 2 of 57

Confidential and Commercially Sensitive

1 Executive Summary ..................................................................................................... 4

Signalling maintenance Page 3 of 57

Confidential and Commercially Sensitive

Maintenance of modern signalling systems Page 4 of 57

Confidential and Commercially Sensitive

6 The key findings from the study are as follows:

Maintenance of modern signalling systems Page 5 of 57

Confidential and Commercially Sensitive

Maintenance of modern signalling systems Page 6 of 57

Confidential and Commercially Sensitive

(c) Undertake a more detailed comparison of maintenance activities between the

12 The following metros participated in the study:

Maintenance of modern signalling systems Page 7 of 57

Confidential and Commercially Sensitive

Maintenance of modern signalling systems Page 8 of 57

Confidential and Commercially Sensitive

4 Approach and methodology

Maintenance of modern signalling systems Page 9 of 57

Confidential and Commercially Sensitive

Maintenance of modern signalling systems Page 10 of 57

Confidential and Commercially Sensitive

5 Structural and explanatory factors

Maintenance of modern signalling systems Page 11 of 57

Confidential and Commercially Sensitive

Maintenance of modern signalling systems Page 12 of 57

Confidential and Commercially Sensitive

Maintenance of modern signalling systems Page 13 of 57

Confidential and Commercially Sensitive

Maintenance of modern signalling systems Page 14 of 57

Confidential and Commercially Sensitive

Maintenance of modern signalling systems Page 15 of 57

Confidential and Commercially Sensitive

Maintenance of modern signalling systems Page 16 of 57

Confidential and Commercially Sensitive

Performance - SAFs per track km

Maintenance of modern signalling systems Page 17 of 57

Confidential and Commercially Sensitive

Maintenance of modern signalling systems Page 18 of 57

Confidential and Commercially Sensitive

Total hours that assets are unavailable following

Maintenance of modern signalling systems Page 19 of 57

Confidential and Commercially Sensitive

50 On the Jubilee Line, in order to minimise downtime during failure of train-borne

Maintenance of modern signalling systems Page 20 of 57

Confidential and Commercially Sensitive

Passenger journeys (million journeys/ track km)

Maintenance of modern signalling systems Page 21 of 57

Confidential and Commercially Sensitive

MTTR versus tph

MTTR versus tph Linear (MTTR versus tph)

Maintenance of modern signalling systems Page 22 of 57

Confidential and Commercially Sensitive