CHARMEC Railway accidents should they be avoided? Anders Ekberg CHARMEC / Applied Mechanics anders.ekberg@chalmers.se
The current status
Railway safety performance in the European Union 2011 Rail fatalities Figure 1: Number of fatalities for different categories of railway user (2007-2009) 1 000 929 900 857 852 800 700 600 500 504 400 300 200 380 400 100 70 89 37 38 38 29 50 44 73 0 Passengers Employees Level-crossing users 2007 2008 2009 Unauthorised persons Others
Sweden 2008 SIKA Rail and road fatalities Total number of causualities 1999 2008 Road: 3422 drivers and passengers in (car, truck, bus) Rail: 4 passengers Railway traffic about 10% of person kilometers About 100 times safer per person kilometer
Expected and unexpected consequences of accidents
Expected consequences Economical effects primary vs secondary who is paying? Long-term effects Swedish winters trust in rail Improvements Improved reliability through improved technology Traffic congestions
More unexpected consequences Legal effects Eschede lid on development Environmental effects Emissions from shift to road + maintenance actions + leakage etc Safety effects Eschede worst railway accident in Europe in modern time
Train accidents > road deaths Eschede train accident 101 fatalities German roads 1998 7 792 fatalities http://www.safespeed.org.uk Increased road traffic by 2% from regulations is plausibe more secondary fatalities on road!
Some conclusions so far Travel by rail is generally much safer than road traffic Good safety reasons to shift traffic to rail (in addition to other factors) For this reason also non-fatal accidents are crucial since they cause disturbances and lack of trus in train travel
Derailments major failures
EU-project D-Rail Main causes of derailments 1. axle ruptures axle break 2. excessive track width 3. wheel failure 4. skew loading 5. excessive track twist 6. track height/cant failure 7. rail failures 8. spring & suspension failure Complications with clustering!
EU-project D-Rail The problem with statistics Reported freight derailments and average costs 2005 2010: France 30 690 k Austria 69 615 k UK 257 169 k Germany 84 730 k Switzerland 16 667 k Large influence of reporting thresholds!
Eschede wheel break nbv.atw.hu
Hatfield rail break More than 2000 potentially dangerous sites were found in UK after the derailment
Brétigny-sur-Orge crossing Picture SNCF
Picture SNCF Brétigny-sur-Orge
Brétigny-sur-Orge Picture BEA-TT Bureau d enquêtes sur les accidents de transport terrestre
Södra station axle break Picture Trafikverket Broken axle Stockholm Södra 2013-11-12 Ultra sonic inspection 2013-10-31 Foto visar hjulets lutning i förhållande till växellådan.
Haninge likely rail break in S&C Picture Seko, Stockholm
Norra stambanan rail break Picture Anders Eriksson/Strukton Rail/Scanpix
Grötingen bearing box
What to do?
1 Paradigm shift Traditional approach static infrastructure re-action fixed maintenance intervals New situation rolling process industry pro-action condition monitoring and forecasting
Catch damage early! Increased load Increases deterioration Lennart Lundfeldt, Banverket Unwanted results
Exponential life decrease Wöhler around 1850
A typical consequence Kalousek, J, E Magel, J Strasser, W N Caldwell, G Kanevsky, and Blevins B (1996) Tribological interrelationship of seasonal fluctuations of freight car wheel wear, contact fatigue shelling and composition brakeshoe consumption. Wear 191, 210 218 winter peaks
Reality behind statistics Needs for reprofilings Example of direct cost: ~50 Mkr for one line and one season
2 Monitor
3 Maintain Iman Arasteh khouy, Håkan Schunnesson, Ulla Juntti, Arne Nissen and Per-Olof Larsson-Kråik, LTU
4 Precise limits Stop only derailment prone trains
Challenges
The scatter is massive
The margins are slim Per Gustafsson
Optimisation and sensitivity
Some final remarks
Railway are safe for passengers Major failures should be avoided for overall societal safety etc Solutions that limit rail operations are sub-optimized The railway network needs to be seen as a process industry Risk management low probability / potentially major effects
Deal with outliers, but do not disrupt operations Optimize, but be aware of that this increases the sensitivity to disturbances Keep deterioration levels sufficiently low and under control the rest will follow
www.chalmers.se/charmec Many thanks for your kind attention!