Crashworthiness

Southall rail crash - reproduced from the London Emergency Services Liaison Panel (LESLP) Website [1], which states: You may use the information and pictures contained herein for educational purposes in a non-profit making and non-commercial environment.
A letter in the latest issue of Rail questioned the crashworthiness of mark 3 stock. The most serious example of failure of this type of stock is the incident at Southall in 1997 (above), where a signal passed at danger resulted in an HST travelling at high speed receiving an impact from a freight locomotive, with 7 fatalities. But is there any stock would have performed any better in such an event? As in any accident, there is always an element of luck, and the circumstances at Southall were a very particular and unfortunate combination; they are analysed in detail in this informative video.

One aspect of crashworthiness is the strength of the bodyshell structure, and despite Southall, mark 3 stock may be as good as it gets. The subject is discussed at length in Whole Train Dynamic Behaviour in Collisions and Improving Crashworthiness, publication TS118, issued by the Rail Safety and Standards Board in 2006. This is of course not the last word on the subject, but tellingly, the author of TS118, Dr Gordon Taylor, states that "It should be noted that none of the standards, either UK or European, comment on the desired crashworthiness performance of a vehicle once the dedicated crashworthiness features have been overwhelmed. For example, it may be useful to provide guidance on the preferred mode of structural collapse."

It is argued that a shortcoming of mark 3 stock is the lack of collapse zones at the ends of the vehicles, to enable the controlled absorption of energy in the event of end-on collisions. But how much value would collapse zones have been at Southall?

The concept of collapse zones came to notice following a the serious accident at Clapham Junction in 1988, involving trains formed of mark1 stock and resulting in 35 fatalities. Previously, this stock had been regarded as providing a good standard of protection, but expectations had risen in the 40 years since it was designed. TS118 notes that "Following the collision at Clapham in 1988, it was required, as an interim measure, that all new vehicles be designed to absorb 1MJ in a collapse zone at each vehicle end. Subsequently, theoretical calculations were undertaken in which impacts between commuter trains of varying length were undertaken. These showed that, for two such trains each comprising 35 tonne vehicles, the collapse zone energy absorption requirement for a 60km/h head-on collision was 1.5MJ for the impacting end and 0.75MJ for each intermediate end. Collapse zones of 1m at the impacting end and 0.5m at each intermediate end and a peak collapse force of 3MN were also specified as being readily and economically achievable."

Steel versus aluminium
During the last 20 years, large numbers of vehicles have been introduced with bodyshells manufactured from aluminium alloy extrusions. Structural assessments comprising quasi-static crush tests, behaviour in accidents and mathematical modelling have been undertaken to allow comparisons with low carbon steel bodyshells to be made. The principal conclusions were:

• Vehicles built from long aluminium alloy extrusions are generally able to sustain higher loads before collapsing than equivalent steel vehicles. This results in higher decelerations when aluminium vehicles collide.
• Typical aluminium alloy extrusions tend to be thicker and stiffer than equivalent fabricated steel sections and hence are more prone to fail by tearing rather than bending, with the result that less energy is absorbed.
• Welding of alloys typically used for aluminium vehicles (6000 series alloys) significantly reduces the tensile strength relative to the unwelded alloy. This concentrates all the structural deformation in the weaker heat affected zone of member joints, producing the effect of low ductility.

Quasi-static tests undertaken on aluminium alloy structures during their early development highlighted difficulties with their collapse behaviour, especially in respect of the very high force necessary to initiate collapse (almost 1000 tonnes in one example) and the subsequent low force sustained as the welds failed by tearing along the heat affected zones. As part of the Crashworthiness Development Programme, BR Research designed, built and tested an aluminium alloy cab end using a bonded and riveted construction to eliminate the welds. This was only partially effective and parallel developments by vehicle manufacturers, using 5000 series alloys and adopting different construction techniques, proved more successful.

In terms of resulting passenger casualties, a study undertaken in 2000 following the collision at Ladbroke Grove, concluded that there was little difference in the crashworthy performance of modern steel and aluminium alloy vehicles. The same study also concluded that there remained a number of improvements that could further enhance the performance of aluminium alloy vehicles.

Causes of fatalities
There are four principal causes of injury in rail accidents, being

• ejection from the vehicle,
• penetration of crash debris into the vehicle,
• crush of the passenger survival space, and
• impact of the vehicle occupants with internal fixtures and fittings (termed secondary impact).

Injuries due to crushing or penetration are usually grouped under the term ‘loss of survival space’. If a train can be designed to remain upright and in-line in an end-on collision, the likelihood of sustaining fatalities or serious injuries due to uncontrolled crush, penetration and ejection will be substantially reduced. Therefore in the following discussion, which assesses the likely benefits of this work, the term ‘loss of survival space’ has been expanded to also include injuries resulting from ejection.

Of the 162 fatalities recorded during the period 1984-2005, 89% were as a result of crush, penetration or ejection. These categories also accounted for 32% of serious injuries, but just 3.5% of minor injuries. Hence the majority of both minor (96.5%) and major (68%) injuries were attributable to secondary impacts. During the 1973-1983 period, 99% of the 67 recorded fatalities, and 56% of the serious injuries, were attributed to loss of survival space or ejection.

Most fatalities over the two periods for which data is available can therefore be attributed to loss of survival space of one form or another. However, the data show a reduction in the number of fatalities and serious injuries attributable to loss of survival space between the two periods. This is most likely due to the reduced number of accidents involving Mark 1 rolling stock in the latter period... the data therefore demonstrate that phasing-out of Mark 1 rolling stock has coincided with a reduction in the number of fatalities resulting from loss of survival space. However, the loss of survival space remains the principal cause of fatalities on the UK network, as well as accounting for a significant proportion of the major injuries.

The data for the two periods also indicates a clear trend towards ejection as a major cause of fatalities or serious injury. Since ejection is usually the result of trains coming out of line during an accident, this indicates that there is benefit in improving the stability of trains in end-on collisions. This also supports other efforts to reduce the likelihood of passenger ejection such as the strengthening of windows.

...in the same period the number of casualties per accident is actually greatest for side-on impacts. This may be reasoned by considering that in the event of a side-on collision the primary crashworthiness features of the vehicles are bypassed, and there is much less survival space protection.

Discussion
The study from which the above extracts are drawn can be regarded as close to exhaustive. Amongst the conclusions that can be drawn are that

• the entire issue must remain under review in the light of ongoing experience
• since the advent of mark 3 stock, safety has been brought towards a plateau, since accidents are of their nature unpredictable and that designing for one set of conditions can mean that performance in other circumstances is sub-optimal.
• the change from steel to aluminium construction in recent years has introduced a new set of problems, and there are other risks unrelated to crashworthiness, including fire.
• collapse zones can only dissipate a fraction of the energy involved in a high-speed incident and can themselves become a hazard unless passengers are excluded from those areas of the vehicle
• there are many types of possible incidents in addition to end-on collisions and as safety standards improve, it becomes increasingly difficult to design in anticipation of a variety of events of low probability
• if the compliance costs of safety regulations place an unnecessary burden on the the railway, people will transfer to cheaper, less safe modes of transport, making the safety measures counter-productive.

Survivability in the event of accidents has also been increased by changes unrelated to the basic design of vehicles, such as improvements in interior design and the use of laminated glass windows to reduce or eliminate the possibility of ejection from vehicles, which had become a major cause of fatalities in accidents.

It could reasonably be argued that the performance of mark 3 stock as currently operating could be accepted as providing a performance standard for the crashworthiness of future rolling stock.

Accidents involving mark 3 stock since 1997
1999 Ladbroke Grove (7 fatalities in mark 3 vehicle)
2004 Ufton Nervet (7 fatalities)

Accidents 1967-1997

Date	Location	Nature of accident	ATP-preventable?	Rolling stock	Fatalities

19.9.97	Southall	Train collision	—1	Post-Mark 1	7
8.8.96	Watford Junction	Train collision	Yes: SPAD2	Post-Mark 1	1
8.3.96	Rickerscote	Derailment, then collision	No	Non-passenger	1
31.1.95	Ais Gill	Derailment, then collision	No	Post-Mark 1	1
15.10.94	Cowden	Train collision	Yes: SPAD	Mark 1	5
25.6.94	Branchton	Derailment	No	Mark 1	2
27.7.91	Newton	Train collision	Yes: SPAD	Mark 1	4
8.1.91	Cannon Street	Buffer stop collision	Yes: Overrun	Mark 1	2
4.8.90	Stafford	Train collision	No	Post-Mark 1	1
20.4.89	Holton Heath	Train collision	No	Non-passenger	1
6.3.89	Bellgrove Junction	Train collision	Yes: SPAD	Mark 1	2
4.3.89	Purley	Train collision	Yes: SPAD	Mark 1	5
12.12.88	Clapham Junction	Train collision	No	Mark 1	35
11.11.88	St Helens	Derailment	No	Post-Mark 1	1
19.9.86	Colwich	Train collision	Yes: SPAD	Post-Mark 1	1
11.9.86	Bridgeton Depot	Train collision	No	Mark 1	2
27.4.86	Micheldever	Train collision	No	Non-passenger	1
9.3.86	Chinley	Train collision	No	Post-Mark 1	1
4.12.84	Eccles	Train collision	Yes: SPAD	Post-Mark 1	3
3.12.84	Longsight	Train collision	No	Mark 1	1
11.10.84	Wembley Central	Train collision	Yes: SPAD	Post-Mark 1	3
30.7.84	Polmont	Derailment	No	Post-Mark 1	13
3.2.84	Wigan	Train collision	No	Non-passenger	2
9.12.83	Wrawby Junction	Train collision	No	Mark 1	1
3.2.83	Elgin	Derailment	No	Post-Mark 1	1
9.12.82	Linslade	Derailment	No	Mark 1	1
11.12.81	Seer Green	Train collision	No	Mark 1	4
8.12.81	Ulleskelf	Derailment	No	Post-Mark 1	1
7.11.80	Crewe	Train collision	Yes: Excess speed	Non-passenger	2
14.3.80	Appledore	Derailment	Yes: Excess speed	Mark 1	1
22.10.79	Invergowrie	Train collision	Yes: SPAD	Mark 1	5
16.4.79	Paisley Gilmour St	Train collision	Yes: SPAD	Mark 1	7
25.2.79	Fratton	Train collision	No	Mark 1	1
19.12.78	Hassocks-Brighton	Train collision	Yes: SPAD	Mark 1	3
5.9.77	Farnley Junction	Train collision	No	Mark 1	2
3.1.76	Worcester Tunnel Jc	Train collision	No	Non-passenger	2
26.10.75	Lunan Bay	Train collision	Yes: Excess speed	Post-Mark 1	1
6.6.75	Nuneaton	Derailment	Yes: Excess speed	Mark 1	6
23.1.75	Watford Junction	Derailment, then collision	No	Post-Mark 1	1
23.10.74	Bridgwater	Train collision	No	Non-passenger	1
11.6.74	Pollokshields E Jc	Train collision	Yes: SPAD	Mark 1	1
19.12.73	West Ealing	Derailment	No	Mark 1	10
30.8.73	Shields Junction	Train collision	Yes: SPAD	Mark 1	5
27.4.73	Kidsgrove	Train collision	Yes: SPAD	Non-passenger	1
6.9.72	Leicester	Train collision	No	Non-passenger	1
11.6.72	Eltham Well Hall	Derailment	Yes: Excess speed	Mark 1	6
16.12.71	Nottingham	Train collision	Yes: SPAD	Non-passenger	3
6.10.71	Beattock	Train collision	No	Non-passenger	1
2.7.71	Tattenhall Jc	Derailment	No	Mark 1	2
26.2.71	Sheerness	Buffer stop collision	Yes: Overrun	Mark 1	1
20.5.70	Guide bridge	Derailment	No	Mark 1	2
31.12.69	Roade Junction	Derailment, then collision	No	Post-Mark 1	1
18.5.69	Beattock	Train collision	No	Mark 1	1
7.5.69	Morpeth	Derailment	Yes: Excess speed	Mark 1	6
8.4.69	Monmore Green	Train collision	Yes: SPAD	Post-Mark 1	2
8.3.69	Ashchurch	Derailment, then collision	Yes: Excess speed	Mark 1	2
4.1.69	Paddock Wd—Marden	Train collision	Yes: SPAD	Mark 1	4
9.9.68	Castlecary	Train collision	No	Mark 1	2
5.11.67	Hither Green	Derailment	No	Mark 1	49
27.9.67	Didcot	Derailment	Yes: Excess speed	Mark 1	1
31.7.67	Thirsk	Derailment, then collision	No	Mark 1	7
5.3.67	Connington South	Derailment	No	Mark 1	5
28.2.67	Stechford	Train collision	No	Mark 1	9

Source: Memorandum by University of London Centre for Transport Studies (RS 24):
TRENDS IN TRAIN ACCIDENTS ON BRITAIN'S MAIN LINE RAILWAYS (Andrew W Evans LT Professor of Transport Safety, University College and Imperial College London June 1998)