INTRO: Aluminium alloy is now the preferred material for passenger coach bodyshells, but recent high-profile accidents have led to concern about its performance in high impact collisions. Both steel and aluminium bodies can meet today’s crashworthiness specifications, but in a high energy collision serious damage cannot be avoided

BYLINE: Simon Leutenegger, Alois Starlinger and Jürg Zehnder*

MODERN RAIL vehicles are designed so that a stiff and strong passenger saloon is protected by energy-absorbing crumple zones, normally located at the ends of the car body. Recent high-profile collisions have involved a much higher energy level than provided for in current specifications. These are exceptional cases in which the consequences are impossible to predict, except that serious damage is inevitable.

In any consideration of rail safety, it is important to remember that the cost of any improvement must be justified in relation to the need to run an economic and competitive service. So this article is not about the design of car bodies to comply with existing collision requirements, but about high-energy collisions and safety within the passenger area of a coach.

With high-energy collisions the structural concept is much more important than the choice of material. It should be pointed out that the first cost of aluminium is higher than steel, but its lighter weight leads to energy savings and hence lower life costs.

Aluminium alloy bodies, which normally consist of double skin extrusions, will fail by rupture in the weakest zones of the structure. Steel bodies, which are usually single skin or thin sheet double skin, will fail by local or global instability of the shell. At the moment there are no research results available to compare different body designs in terms of a possibly dangerous reduction of the passenger survival space.

Nonetheless, comparison of calculations and pictures of real collisions show that there is a high probability that a double skin aluminium passenger structure, before failing by rupture in the weakest zones, has the potential to absorb more energy than a steel design before failing through global instability.

It is widely acknowledged that rail is the safest form of land transport, and its inherent concept of guided vehicles means that it is well suited for active safety measures. As signalling and safety systems develop, major accidents are increasingly rare. This means that any serious incident attracts disproportionate media attention.

In Japan the shinkansen network has seen no passenger killed or seriously injured in a train accident since the first line opened between Tokyo and Osaka over 36 years ago. Today the shinkansen network carries more than 800000 passengers a day. In Europe, despite at least three derailments between 250 and 300 km/h, no passenger has been killed or seriously injured on any of the high speed lines built since the late 1970s. Measured in passenger-km, the risk of death in a rail vehicle is six times lower than in a civil aircraft and about 100 times less than in a private car.

On conventional lines, where the infrastructure was often built more than 150 years ago, implementation of active safety measures is more complex and also more onerous. Human error in its many forms cannot be eliminated, and it should therefore be accepted that accidents will occur from time to time. For this reason it is essential to incorporate passive safety measures in the design of passenger rail vehicles.

A glance at the data in Table I shows that complete protection of passengers up to any collision speed by passive safety measures designed into the vehicle is not possible without unreasonable effort. As an example, the energy absorbing capacity of four high performance buffers equals about 0·3 MJ.

Another example is an eight-car Class 390 Pendolino train being built for Virgin Trains’ West Coast Main Line service in Britain. If a Class 390 were to collide at 100 km/h with a stiff obstacle (154 MJ), and assuming an average crush force of 3000 kN, the total deformation length would need to be 51m.

The collision energy in three recent accidents and the installed energy absorbing capacity are shown in Table II. The installed energy absorbing capacity of an 11-car Class 390 as an example of a modern train with sophisticated crashworthiness features amounts to about 45 MJ, whereas a 60 km/h head-on collision of a Class 390 with an identical train at a standstill would consume 19 MJ per trainset.

Until about five years ago train operators did not specify any energy absorbing capability beyond the performance of the buffing and draft gear. Today it is possible to install an energy absorbing capacity of a few dozen MJ into a high speed train. That is sufficient to cope with collisions up to a closing speed of about 60 km/h.

The energy released in the cited collisions is more than one magnitude greater than the absorbing capacity. Serious damage is inevitable, and the nature and severity of structural failure shows typical patterns for the design concepts and materials of the bodyshells involved.

Targets for passive safety

Clearly it is not realistic to guarantee survival space for all passengers and crew in a head-on collision at a speed of 80 km/h or more. This also applies to private cars, where sufficient survival space is offered in collisions only up to 60 km/h, even though the car may be built for a top speed of 120 or even 250 km/h.

Thanks to modern signalling and train protection equipment, it will in future be possible to reduce further the number of head-on collisions. This means that passive safety requirements can be determined on the basis of incidents caused by events outside the rail system such as obstacles at level crossing. It is reasonable to limit speed on routes with level crossings to 140 or 160 km/h as a train driver will apply emergency braking as soon as he sees an obstacle on the track ahead, reducing the likely impact speed.

A reasonable target for passive safety could therefore be based on the assumption that in a collision between a train and an immobilised lorry at 100 to 120 km/h the passenger area and the driver’s cab should suffer minimal damage and the deceleration experienced should not be excessive. This was the basis for TGV Duplex design in France, and the predicted crash behaviour was proven in a full-scale test.

Another scenario is an object protruding into the loading gauge, such as a shifted load on a train passing on an adjacent track. In such cases damage cannot be excluded, but it is possible to formulate a degree of protection as a minimum criterion.

Crash scenarios

Historically, strength requirements were defined for freight wagons likely to suffer heavy impacts during shunting, the aim being to avoid structural damage. Compressive resistance was set at 200 tonnes in Europe, 800000lb in South Africa and 1000000lb in North America, rendering it necessary to improve the performance of the buffers or drawgear to cope with the higher mass of the vehicles without reducing shunting speed, typically 15 to 18 km/h.

For passenger vehicles the same compressive resistance was adopted without defining specific crash scenarios. As DMUs and EMUs are not usually shunted, the strength requirement was reduced in Europe to 150 tonnes.

With traditional passenger coach design, high resistance was limited to the underframe. To prevent one coach overriding the underframe of another, anticlimbers had to be fitted. End walls were also strengthened, often by fitting collision posts and corner posts. So until recently it was accepted that collisions at speeds higher than in shunting accidents would cause structural damage, ideally in vestibules or luggage areas.

Thanks to progress in non-linear finite element analysis, it is now possible to simulate the structural behaviour of vehicles beyond the elastic limit. That knowledge, combined with progress in materials technology, allows the structural engineer to design bodies with defined energy absorption properties and targets for passenger and crew safety at much higher impact speeds.

One set of requirements has been formulated in CEN TC256WG2, which is available as a ’prenorm’ (prEN12663 Part 2).

Survival space

When dealing with high speed collisions, different considerations must apply. Keeping the survival space for passengers and crew intact is, besides limiting deceleration, the most important criterion for passive safety. The goal for the design engineer is to guarantee sufficient survival space with no reduction by folding and deforming of structural elements, even at high collision energies.

Recent accidents have caused concern about the crashworthiness of aluminium bodies. In particular, an ’unzipping effect’ was seen as a new phenomenon that presented severe danger for passengers. Unzipping describes a fast progressing rupture, typically of a longitudinal weld seam, caused by the lower strength in the heat-affected zone. The term is not applied to minor ruptures in the body structure.

Leaving aside low speed crash scenarios, the question is to determine which body design offers the best protection for passengers by keeping the survival space intact in a high speed collision, and whether it depends on the choice of material.

After the controlled absorption of energy in crumple zones installed at the ends of the vehicle, the body sustains the highest forces, which are mainly longitudinal. Rupture or loss of stability depends on strength and stability phenomena. How well the strength in a structure is used depends again on stability phenomena. Local buckling is related to the thickness of the detail design, while buckling of the complete structure depends on the slenderness of subassemblies such as underframe and sidewall. Local instability may produce a global stability collapse.

Double-skin shells, whether aluminium or steel, offer lower local slenderness and therefore higher local stability than single-skin shells. Thanks to their inbuilt stiffness, they also offer higher global stability. Although these structural design factors are far more important than material characteristics, it is worth looking at the two most commonly used bodyshell materials.

Given two shells of comparable weight, one in steel, one in aluminium, the lower specific mass of aluminium means that the elements of this structure will be thicker than those of the steel body. In terms of buckling phenomena, the low and therefore better slenderness of the detail aluminium structures usually dominate the lower Young’s modulus due to the squared influence of slenderness.

So detail aluminium structures start to buckle at higher forces than detail steel structures of the same weight because of the higher buckling stress and greater cross-section.

Moreover, typical aluminium bodyshells consist of easily extrudable double skin sections where local and global stiffness are built in automatically as part of the manufacturing process.

In terms of the global stiffness of subassemblies, steel car bodies are only comparable if they are built as double skin shells. But even then the local buckling sensitivity of steel details is still a problem.

Further, owing to the high slenderness of detail steel structures, they tend to start buckling in the elastic range, and therefore sustainable compression forces may diminish after the first buckles whereas thicker aluminium details start buckling in the plastic range and there is potential for actually increasing the sustainable forces.

Another point is that slender thin sheet elements such as steel details are more susceptible to geometrical and other imperfections such as initial waves from fabrication or welded cross stiffeners.

In summary, aluminium bodyshells will sustain at least as high longitudinal forces as steel ones before buckling, and due to their greater deformation because of the lower Young’s modulus they will require even more energy before failing1.

One question must be asked. Does the reduced strength of the heat affected zone (HAZ) that produces the unzipping effect in mainly longitudinal weld seams of aluminium shells destroy these advantages? So far the answer is a clear ’no’. This is because the HAZ not only has reduced strength characteristics but better ductility compared to its parent material.

Quantifying comparisons between steel and aluminium bodyshells taking all these factors into account, either by calculation or in full-scale tests at extremely high collision speeds, has not yet been carried out. Further research will offer more guidance for the design engineer.

Choice of material

Passenger coach design engineers aim not to avoid damage to the bodyshell, but to give the greatest possible protection to passengers and crew. To this end lightweight design is essential, partly to keep the kinetic energy low and partly to permit the best use of efficient energy-absorbing zones where no passengers are present.

Aluminium alloys are today the preferred material for the bodyshells of passenger coaches. The typical materials are all medium-strength alloys of the family AlMgSi (6000 series) with an elastic limit between 150 and 260MPa and with ultimate strength between 190 and 310MPa in fully heat-treated condition.

With special heat treatment, lower strength values combined with much higher elongation are achievable. The heat input from welding has the same effect, namely lower yield strength with improved ductility.

The passenger cell is typically formed of full-length extrusions joined by longitudinal welds. As the minimum wall thickness is dictated either by the extrusion process or by the stiffness needed to guarantee acceptable natural frequencies - and so acceptable ride comfort - the body structure is normally overdesigned from a strength point of view. This was proved in various crash tests where longitudinal forces of up to 6MN were supported without permanent deformation of the extrusions and with no failure of the welds.

Weld failures occur when forces perpendicular to the weld seam are acting. If this is really the case locally, or if special requirements are specified, then the design engineer will simply increase the wall thickness of the extrusions in the HAZ to compensate for the loss in strength.

After 25 years of experience with aluminium rail vehicles, the latest designs use AA6005A alloy for underframe and sidewall extrusions, AA6106 for roof extrusions and AA6082 for heavy gauge sections such as bolsters and headstocks.

The new energy management requirements are best met with AA6008, a special alloy in the same family. This has a slightly improved ductility, both as base material in T6 (fully heat-treated) condition as well as in the HAZ of weld seams due to a finer metallurgic structure than AA6005A, even if the standardised values are the same. It is therefore used in T6 condition for many longitudinal extrusions.

In a specially developed temper for extremely high ductility and medium strength the same AA6008 alloy is used for energy absorbing devices such as crash tubes, and welded collapsible structures, for example in the entrance vestibule at the car end.

Dealing with unzipping

During conventional fusion welding of aluminium a HAZ forms on each side of the weld bead. Here the mechanical properties are influenced by the thermal effect of the welding on the metallurgical structure of the aluminium. If artificially aged alloys such as AA6008 are welded, some softening occurs. This is due to local resolution of hardening precipitates near the weld bead and over-aging in the zone near the parent metal.

In most cases the mechanical properties of welds are characterised by tensile tests transverse to the welding direction. In this case, unaffected parent metal, the softened HAZ, and the weld bead are within the gauge length of the tensile test. Consequently, plastic deformation will concentrate on the soft zones, whereas the parent metal will remain practically unaffected. Thus the ductility in the HAZ is underestimated if tests are performed with a specimen transverse to the welding direction. If tensile test specimens are taken parallel to the weld bead, a better characterisation of local mechanical properties can be obtained.

Results of tests with welded AA6008 sections show a yield strength of about 120MPa and elongation of around 20% in the HAZ; the ultimate strength is 220MPa.

The last 20 years have led to a better understanding of aluminium alloy metallurgy, and the effect of the silicon content in the parent material on the ductility of the HAZ is now well known. It is therefore kept at the lower limit of the AlMgSi0.7 permitted range.

Knowledge of the effect of the filler wire on ductility suggests that AA5183 (AlMg4.5Mn) should be used as this gives a slightly better ductility in the weld bead than AA5356 (AlMg5). AA4043 (AlSi5) filler wire, especially in combination with high silicon content base material, leads to a lower ductility of the HAZ.

Finally, it is worth commenting on the question of the strain rate sensitivity of steel and aluminium: It seems that under impact loading the yield strength of both steel and aluminium alloys increases slightly. On the other hand the elongation of steel is slightly reduced, whereas the elongation and consequently the ductility of aluminium alloys increases thanks to the face-centred cube lattice microstructure. Quantifying tests to evaluate these effects under strain rates corresponding to high speed collisions are in progress.

TABLE: Table I. Energy of different types of moving vehicle

Vehicle mass speed speed energy max decel- braking type kg km/h m/s MJ eration distance m/s2 m

LRV 30 000 55 15·3 3·38 2·7 43

5-car metro train 150 000 80 22·2 36·3 1·3 190

8-car Class 390 400 000 225 63·9 820 1·3 1570

Private car 1 500 100 27·8 0·58 2·5 154

TABLE: Table II. Energy absorption in high impact collisions

Accident Date Mass* Collision Collision Design tonnes speed energy capacity km/h MJ (est) MJ (est)

Eschede, Germany June 3 1998 600 200 926 3

Ladbroke Grove, Britain October 5 1999 115 210 195 1

Årsta, Norway January 4 2000 110 180 130 1

*Mass of aluminium-bodied train

* Simon Leutenegger is Head of Engineering, Alois Starlinger is Head of Structural Analysis, Computer-Aided Engineering &Testing, and Jürg Zehnder is President of Alusuisse Road & Rail Ltd, a member of the Alcan group of companies

CAPTION: Full-scale quasi-static collision tests were carried out on a Pendolino Britannico bodyshell. Good ductility of the weld seams in the collapsible aluminium tubes was shown, with the heat-affected zone folding without rupture

CAPTION: Typical high energy deformation pattern of a double skin aluminium bodyshell after the Eschede accident in June 1998, showing failure in the weakest or most highly stressed zones


CAPTION: Typical high energy deformation pattern of single skin steel car bodies showing global instability


CAPTION: Fig 1: Alusuisse Road & Rail Ltd was involved in developing the crashworthiness features of Virgin Trains’ Pendolino Britannico trainsets for use on Britain’s West Coast Main Line. The front and rear of the body structure has been optimised for crashworthiness using non-linear, dynamic finite element simulations focused specifically on the crumple zones. Structural integrity of the passenger survival space has been verified by an elasto-plastic, quasi-static finite element calculation taking into account the maximum collision peak loads of the end zones. Special attention has been paid to the modelling of the major welds, and the passenger survival space has been designed to include a large margin against structural collapse. Under the specified collision conditions the plastic strains within the critical welds of the passenger cell reach about 30% of the allowable value. The safety factor against global buckling effects is 2·7


1. Jaccard, R. Aluminium - ein Werkstoff für die rationelle Konstruktionspraxis. Schweizerische Aluminium Rundschau 3.80

Lightweight structures meet tougher crashworthiness standards.

Recent developments in the design of lightweight aluminium alloy structures for rolling stock allow suppliers to meet more demanding crashworthiness standards. Better designs of structure that deform in a controlled way at the car ends have improved the ability of coach structures to withstand high velocity impact. At the same time experience following recent high-profile accidents in Europe has led to a better understanding of metal behaviour, in particular in areas close to weld lines. With aluminium alloy now chosen for 80% of new orders for rolling stock in Europe, the most recent designs can meet the much higher safety standards imposed in several countries.

Les structures légères répondent aux normes sévères anti-collision

Les récents développements en matière d’élaboration de structures allégées destinées au matériel roulant, en alliages d’aluminium, permettent aux fournisseurs de répondre à des normes anti-collision de plus en plus exigeantes. De meilleures conceptions de structures se déformant de manière contr