23 october 2006 n° 1 r.piazza, p.mihm and c.cassir developing csts for the european railway system...

23 October 2006n° 1

R.Piazza, P.Mihm and C.CassirDeveloping CSTs for the European Railway System

IRSC 2006, Dublin

Developing Common Safety Targetsfor the European Railway System

Roberto Piazza, Peter Mihm and Christophe Cassir

IRSC, Dublin, 23 October 2006



IRSC 2006, Dublin

The legal bases for developing CSTs at EU level

Treaty establishing the European Community, in particular Article 71§1.c:“1. For the purpose of implementing Article 70 (common transport policy at EC level), and taking into account the distinctive features of transport, the Council shall (…) lay down:…

(c) measures to improve transport safety;…“

Directive 2004/49/EC, which, on the other hand, acknowledges that safety levels in the Community rail system are generally high, in particular compared to road transport, and requires:- that current safety performance of rail is not reduced in any Member State- that CSTs are developed, expressed in risk acceptance criteria (1st set of CSTs to be adopted by the European Commission by end of April 2009 - Article 7)

Mandate of the European Commission to the European Railway Agency – issued 16/12/2005(1st set of CSTs to be submitted to the European Commission by end of September 2008)



IRSC 2006, Dublin

Main aims of establishing CSTs at EU level

Limit differentiation of national policies in the field of safety targeting, as this may hinder the competitive potential of railway transport with respect to other transport modes by fragmenting the EU market

Harmonise the way safety is monitored and reduce existing differentiation in the safety performance of railway systems in Member States

Avoid that “safety arguments” are unduly used by Member States for creating barriers to the entry into the respective national markets by newcomers



IRSC 2006, Dublin

Two main options for safety targeting

Option 1: targeting the individual/collective risk of incurring into a given category of consequences of a given category of accidents.

Option 2: targeting the occurrence of possible

causes of accidents (i.e. broken rails/axles/wheels,

signal failures, n. of SPADs, etc.).

Dir.2004/49/EC (Art.7) mandates the development of CSTs

expressed in terms of risk acceptance criteria for

individuals and the society as a whole (i.e. option 1).

Option 2 may be used for complementary targets and/or

for 2nd set of CSTs.



IRSC 2006, Dublin

The approach taken by the European Railway Agency for developing CSTs

A two step approach:

First develop a quantitative baseline for understanding the level of safety performance of railway transport in the different Member States, expressed in terms of risk to individuals + societal risk (National Reference Values - NRVs)

Then derive CSTs from NRVs, by iteratively setting targets and checking the outcomes of the impact analysis, until an optimum is reached



IRSC 2006, Dublin

How many NRVs and CSTs ?

CSTs

At least 6 at MS level

(according to Art.7 SD)

Passengers

Staff

Level Crossing Users

Unauthorised persons on railway premises

Others

+Societal Risks

NRVs At least 6 (NRVs) x 23 (MSs) = 138 NRVs

One NRV for:

- each correspondent CST

- each MS, excepted MS without railway systems (Malta and Cyprus) = 23 +

NRVs for parts of the railway system (where feasible)

+CSTs for parts of the

railway system

(where feasible)



IRSC 2006, Dublin

How to measure NRVs and CSTs ?

CSTs

NRVs

Expressed in terms of RISK

ACCEPTANCE CRITERIA

Will have the same unit of measure

which is adopted for quantifying RISK

Measurement:

Dimensional Definition:

RISK of a

given event

=

Frequency of the event

[ n.of events / ]

(*) train*km, pax*km, tonn*km

X

Consequences of the event

[ Consequences / event ]

=[ Consequences /

]Unit of Time

Unit of Product (*)Unit of Time

Unit of Product



IRSC 2006, Dublin

NRVs: the list of formulae and scaling factors which has been selected by the CST WG

Category at risk Preferred complete formulaePreferred scaling

factors

Passengers

NRV 1.1

Number of pax fatalities and weighted serious injuries per year arising from significant accidents / n. of passenger train*km per year

Passenger train*km per year

NRV 1.2

Number of pax fatalities and weighted serious injuries per year arising from significant accidents / n. of passenger*km per year

Passenger*km per year

Employees NRV 2Number of employee fatalities and weighted serious injuries per year arising from significant accidents / n. of train*km per year

Train*km per year

Level crossing users

NRV 3Number of level-crossing user fatalities and weighted serious injuries per year arising from significant accidents / (n. of Train*km per year / n. of LCs)

Train*km per year per LC(To be further discussed by a specific Taskforce)

Others NRV 4Yearly number of fatalities and weighted serious injuries of persons belonging to the category “others” arising from significant accidents / n. of train*km per year

Train*km per year

Unauthorised persons on railway premises

NRV 5Number of fatalities and weighted serious injuries of unauthorised persons on railway premises per year arising from significant accidents / n. of train*km per year

Train*km per year

Wholesociety

NRV 6Number of fatalities and weighted serious injuries to persons per year arising from SIGNIFICANT ACCIDENTS / n.of train*km per year

Train*km per year



IRSC 2006, Dublin

• NRV and CST for passengers risk NRV1;CST1

• NRV and CST for staff risk NRV2;CST2

• NRV and CST for level crossing users risk NRV3;CST3

• NRV and CST for unauthorised persons risk NRV4;CST4

• NRV and CST for risk to others NRV5;CST5

• NRV and CST for societal risk NRV6;CST6

Common Safety Indicatorsfor 1st set, only data from 2006

EUROSTAT datafor 1st set, full data from 2004, 2005, 2006 + partial data (w/o production data) from 2007

OthersVoluntary time series from MS

Extra normalisation (production) data

NRVs, CSTs and data feeding

NRVs & CSTs for parts of the railway system as defined by Article 3e of Dir. 2004/49/EC

Will specific data be available ?If not, careful assumptions and

estimates will be necessary



IRSC 2006, Dublin

Data for Common Safety Indicatorsas defined by Annex I of the safety Directive

K SI K SI K SI K SI K SI

Collisions

Derailments

Level-crossing accidents

Accidents to persons caused by rolling stock in motion

Fires in rolling stock

Others

Suicides

Total (excluding suicides)

Total relative to train km (excluding suicides)

Lev

el c

ross

ing

use

rs

Un

au

tho

rise

d p

erso

ns

Oth

ers

Nu

mb

er o

f a

ccid

ents

Pa

ssen

ger

s

Em

plo

yee

s

CST1 CST2 CST3 CST4 CST5

CST6 ?

Issues: The use of national definitions

for indicators (at least for 2006 annual safety report) poses a limit on the comparability of data for different MSs

Data will be available only for 2006 onwards

K = Killed

SI = Seriously Injured



IRSC 2006, Dublin

EUROSTAT data

K SI K SI K SI

Collisions

Derailments

Level-crossing accidents

Accidents to persons caused by rolling stock in motion

Fires in rolling stock

Others

Unknown

Total

Total relative to train(or pass) km

Oth

ers

Num

ber

of a

ccid

ents

Pas

seng

ers

Em

ploy

ees

CST1 CST2

~CST3

~CST4

~CST5

CST6 ?

Issues: Uncertainty for CST3, CST4, CST5

Need to check that suicides data are not included in “ Accidents to persons caused by RS in motion”

K = Killed

SI = Seriously Injured



IRSC 2006, Dublin

Commonalities amongst Member States:the most exposed categories of individuals

(year 2004 – Source: Eurostat)

The most exposed category of individuals in 2004 was the category « others » (i.e. not passengers, nor employees)



IRSC 2006, Dublin


Most of the fatalities in 2004 were due to « accidents at level crossings » and to « individual accidents to persons caused by RS in motion »

Commonalities amongst Member States: accidents causing the majority of fatalities



IRSC 2006, Dublin

N.of fatalities per Mio train*km caused by railway accidents in countries of the European area

[(EU + EEA) Member States + Turkey] - years 2004 and 2005(source EUROSTAT - data for 2005 from DK, DE, ES, FR, IE, LU, NL, NO, PT, SE and UK not available)

Averagevalue for:

2004 = 0,922005 = 1,13(excluding

zero values)

0

1

2

3

4

5

6

BE CZ DK DE EE GR ES FR IE IT LV LT LU HU NL AT PL PT SI SK FI SE UK TR NO

2004 2005

The N.of fatalities per Mio train*km caused by railway accidents in countries of the European area [(EU + EEA) Member States] in 2004 and 2005 lied in a range 0,05 3 (510E-8 310E-6 Fatalities/train*km), which means that there was a variation of two orders of magnitude between the best and the worst performing MS.

The differentiation of statistical data amongst European Countries: a first problem to face



IRSC 2006, Dublin



Also for the number of accidents per million train*km there was a variation of two orders of magnitude between the best and the worst performing MS (0,5 40) .



IRSC 2006, Dublin

Passenger fatalities per Bln passenger*km from railway acc.in countries of the European area

[(EU + EEA) member States + Turkey] - years 2004 and 2005(source EUROST AT - data for 2004 from LT and SI n.a.; data for 2005 from DK, DE, LT , PT , SI and UK n.a.)

Averagevalue for2004 and

2005(excluding

zero values):1,00

0

1

2

3

4

5

6

7

8

9

10

BE C Z DK DE EE GR ES FR IE IT LV LT LU H U N L AT PL PT SI SK FI SE U K TR N O

2004 2005

…same order of magnitude of variation for the number of passenger fatalities per billion train*km (0,06 3).




IRSC 2006, Dublin

Coping with data insufficiency It is evident from Eurostat data for 2004 and 2005 that

there are significant differences in safety performance between MSs (two orders of magnitude variation of total FWI/train*km amongst Member States, as already shown)

There is a need to analyse why these large differences occur and also to study additional data to see how annual fluctuations might influence these results

A longer time series of national data would serve to average out some of the effects of the high-consequence low-frequence events and also to give more significance to the data for small Member States with few events

Data insufficiency: a second problem to face



IRSC 2006, Dublin

Comparability analysis of data

0

0,5

1

1,5

2

2,5

3

3,5

4

4,5

1988 1990 1992 1994 1996 1998 2000 2002 2004 2006

Year

FW

I (p

er

mil

lio

n p

as

se

ng

er

tra

in k

m)

Comparability analysis: National time series vs. Eurostat time series

Case 1: consistency of the two different time series

Case 2: inconsistency of the two different time series

0

0,5

1

1,5

2

2,5

3

3,5

4

4,5

1988 1990 1992 1994 1996 1998 2000 2002 2004 2006

Year

FW

I (p

er m

illio

n p

asse

ng

er t

rain

km

)

Eurostat

Eurostat

Case 2:Case 1:

0

0,5

1

1,5

2

2,5

3

3,5

4

4,5

1988 1990 1992 1994 1996 1998 2000 2002 2004 2006

Year

FW

I (p

er

mil

lio

n p

as

se

ng

er

tra

in k

m)



IRSC 2006, Dublin

Treatment of data

Averaging and forecasting techniques: Moving averages, exponential smoothing, trend lines etc Objective is to find NRVs less sensitive to annual random fluctuations and

influence from low frequency/high impact accidents The approach may also facilitate setting targets for annual risk reduction

0

0,5

1

1,5

2

2,5

3

3,5

4

4,5

1988 1990 1992 1994 1996 1998 2000 2002 2004 2006

Year

f

Raw data

Estimated long term

trend

Outcome of averaging and

smoothing

Ris

k



IRSC 2006, Dublin

A descending long-term trend for passenger fatalities



IRSC 2006, Dublin

MS

A

MS

B MS

C

MS

D

MS

E

MS

F

MS

G

Ris

k

NRVs = National Reference Values of safety performance for each MS

Lowest impact CST

Weighted impact CST

Highest impact CST

Common Safety Targets – how to set up 1st set of CSTs from NRVs ?

Optimal impact CST (best overall « Cost / Benefit » ratio ?)

ΔRD

ΔRF

ΔRX = range of tolerance for MS X in a given time horizon(will a range of tolerance for NRVs also have to be considered for

each MS ? It will depend on the approach finally adopted for setting CSTs)



IRSC 2006, Dublin

ERA’s CST WG: overall flowchart of activities and deliverables

Preliminary Draft

1. Progress report

5-year Workcalendar

Elaboration of the recommendation for 1st set of CSTs

09/0812/05 11/06

Pré-conditions

Development of a method for setting up 1st set of CSTs

Survey existing Safety targets

Definition of measurement units

Draft of recommen-dation for 1st

set of CSTs

Setting values for 1st set of CSTs

2. Progress report

Guidance and rules / procedures for use and enforcement of 1st set of

CSTs

Inputs from WG membersExisting safety targets; definitions, rules & prodedures for use at MS and Company level (operators, infrastructure manager, …)

Inputs

Drafting Structure and General part

Drafting Technical Part

Drafting Recommen-

dation

Recommendation laying down the 1st set of CSTs + accompanying report + draft

guidelines for use of 1st set of CSTs

11/0701/07 01/08 07/08

Consultation of social partners

Translation

Information of NSA Network

Final adjustment

Delivery of Recommenda

tion

Adoption by EC and Article 21

committee

Drafting Guideline for use of 1st set of CSTs

Annotated structure

Working plan strategy

Inputs from WG membersTime series / Pilot values at MS and Company level, additional information for benchmarking, …

WG members comments

Guidelines for use of 1st set

of CSTs

04/09



IRSC 2006, Dublin

Thank you for your attention!

23 october 2006 n° 1 r.piazza, p.mihm and c.cassir developing csts for the european railway system...

Documents

european railway system

transport safety

set of csts

european railway agency

development of csts

european commission

european community

safety levels