An informative and productive industry workshop about track-bridge interaction was held recently at the Somerset County Cricket ground in Taunton. The workshop brought together over 40 track and bridge experts from the across the UK and Europe and also attracted delegates from Japan.

Track-bridge interaction is one of those interfaces where the bridge engineers tend to assume that it is a ‘track problem’, the track engineers do not necessarily understand the bridge design process, and the specialist supplier, when brought in at a late stage, is left with an insoluble problem.

The aim of this workshop was to promote a joint understanding of the issues around trackbridge interaction, highlight requirements in current and proposed codes and standards and discuss the need for communication between bridge engineers and track engineers.

Do I really need rail expansion joints? David Rhodes of D R Squared is the chair of the European Standards (CEN) task group on track-bridge interaction, which is currently developing updated requirements for inclusion in European standards. He started the workshop by summarising the fundamentals of trackbridge interaction.

This topic has become of increasing interest in recent years due to the change in relative stiffnesses of track and structure. He contrasted the ‘old’ way, exemplified by a masonry viaduct with jointed rail, where there is a ‘rigid’ bridge structure with ‘flexible’ track. In comparison the ‘new’ approach, demonstrated on high-speed rail lines, has more-flexible long span structures and more-rigid trackforms such as slab track.

The phenomenon of track-bridge interaction relates to forces which occur due to relative movement of the track and the bridge. This can occur as the temperature changes, due to traction and braking of the trains and due to bridge deformation under vertical loads.

Bridges tend to be articulated using bearings. The bridge will expand and contract as the temperature rises and falls. In contrast, continuous welded rail remains the same length but the stresses in the rail become more compressive or tensile as the temperature increases and decreases.

The relative movement of the bridge under the static rail tends to ‘drag’ the rail with the bridge movement, and the forces transmitted through the fasteners give rise to additional stresses within the rail and additional forces imposed on the bridge deck.

The ‘give’ between track and deck is described by a force/displacement curve. This is typically modelled by two parameters. u0 is the maximum elastic displacement. Up to this movement, the fastening behaves as an elastic spring and is considered to obey Hooke’s law, with force proportional to displacement.

Beyond this point, the fastening becomes nonlinear and is modelled as behaving plastically and applying a constant force, k. This constant k is given as force per unit length for two rails, hence has units N/m. The choice of letter k is somewhat confusing since this is not a stiffness, but this is now the industry-accepted terminology.

To close his presentation, David Rhodes considered the options available to the track engineer. There are two fundamental alternatives.

The first option is to install a rail expansion device where there is a discontinuity in the structure in order to avoid any additional rail stresses.

Alternatively, the track engineer can opt to accept some locally increased rail stresses. This increase can be limited, either by specifying a limit on the expansion length of the bridge, or by putting a limitation directly on the additional rail stress.

Why can’t we design a highway bridge and put a railway track on top?

John Lane of RSSB presented a view from the bridge engineering side. He noted the difference between highway and rail loads, with typically heavier and faster vehicles present in the rail environment. He also noted the need for compliance with track geometry quality requirements, such as those given in the Infrastructure Technical Specification for Interoperability.

The deformation requirements for bridges, which had originally been developed as guidance in leaflet UIC 774-3, have now been incorporated into the bridge design standards BS EN 1990 and BS EN 1991-2. These are now mandated through the Technical Specifications for Interoperability. Bridge designers are therefore required to demonstrate, amongst other criteria, that acceptable values are achieved for bridge vertical deflection, vertical acceleration, twist, deck end rotation and longitudinal displacement at the deck ends due to traction and braking.

The bridge design standards also contain codified limits on the additional rail stresses, of 72N/mm2 in compression and 92N/mm2 under tension. The work of Mr Rhodes’ task group suggests that these limits should be relaxed in certain cases, for example where rails are continuously restrained laterally such as in slab track.

The background to the current additional rail stress limits of 72N/mm2 and 92N/mm2 has been located and placed in the public domain. It was also noted that, strictly, these limits only apply to UIC60 rail on ballasted track with sleepers spaced at 650mm. The background report describes the derivation of the stress limits and provides the methodology which could be applied to other sleeper and rail configurations.

Selection of track system components Steve Cox of Pandrol Track Systems highlighted some of the analytical techniques available to calculate the rail stresses. These are referred to as ‘additional’ rail stresses, since they are over-and-above the stresses which would be present in continuous welded rail subject to the same temperature and stressing history, but without the presence of a bridge.

The additional rail stresses can be calculated using finite element models, but these need to take into account the non-linear behaviour of the fastenings. This is typically achieved by iteration until a valid solution is achieved.

The model also needs to take account of the fastening force generated due to thermal effects under the ‘unloaded’ condition, and the variation due to traction and braking under the ‘loaded’ condition when the fastenings are
considered to have greater stiffness.

Next, Steve Cox outlined some of the options available to track engineers to control the additional rail stresses generated by track-bridge interaction. Changes to the fastening components can be used to adjust the two main parameters of the force/displacement curve, u0 and k.

He considered the possible extremes, ranging from rail being fully welded to the structure giving a rigid non-slip connection, to the opposite extreme of a friction-free connection. In practice, neither of these extremes would give a practical engineering solution. However, a spectrum of options is available between these extremes, including standard fastenings, low toe load fastenings, very elastic fastenings and zero longitudinal restraint fastenings.

Presenting results obtained from tests of different fastening arrangements, Steve Cox showed that, by varying the toe load, it is possible to obtain a four-fold difference in creep resistance, k. By adjusting the vertical resilience of the
assembly it is possible to obtain differences in the slip deflection u0. Consequently, it is possible to engineer a considerable variation of more than an order of magnitude in the slope of the linear region of the graph, representing the stiffness of the fastening assembly.

Bridge case studies and monitoring Marc Wenner of Marx Krontal presented a case study illustrating the calculation of additional rail stresses due to bridge movement.

This highlighted that the structure joints are usually the decisive locations for considering the additional rail stresses (above right). It also explored the effect of the nonlinearity in the fastening system, in particular, that using linear superposition of the thermal and traction/braking effects rather than nonlinear combination, would typically result in an unconservative underestimation of the relative displacements between the track and the bridge.

Next, Marc Wenner presented some initial results from a monitoring programme of a new high-speed rail bridge on the Nürnberg to Berlin route, the Gänsebachtalbrücke (above left). This is an innovative semi-integral structure with an expansion length of 112 metres, on which the calculated rail stresses were close to the admissible limits.

The federal railway office implemented a monitoring programme to confirm the results achieved in service. The programme included comparison of rail stress over time due to temperature changes, and bridge displacements
under braking of a test train.

Pont Briwet

Jerry Barnes of Hewson Consulting and Sean Ring of Beazley Sharpe (Railwise) discussed the benefits they had achieved on the Pont Briwet bridge (below) replacement on the Cambrian Railway in Wales. The bridge is 133 metres long and integral, with track on a 465-metre radius curve.

The original Form A had included for long-blade movement switches coupled with a breather joint at each end of the bridge. Careful consideration of track-bridge interaction allowed for these to be designed out and the use of unbroken continuous welded rail to be justified, giving considerable cost savings.

Rail expansion joint system

On some occasions rail expansion joints will continue to be necessary. Johannes Rohlmann and Burkhard Zillien described the Voestalpine rail expansion joint system. The system consists of longitudinal moveable stock rails and stationary switch rails, and is available in four variants allowing for movements of up to +/- 900mm. A crossbar mechanism provides support to the rails as the structure joint size expands.

More than 600 units have been in service since the early 1990s in Germany, Spain and Sweden. More than 100 units have been installed in China since 2013 for speeds of up to 300kph on ballasted and ballastless track.

Laying down the challenge

Angus Low ended the workshop with his personal views based on many years of international bridge design with Arup. He drew attention to what appeared to be a general confusion around track-bridge interaction.

He illustrated this with numerous documents from around the world regarding track-bridge interaction which appeared to be contradictory to #the UIC leaflet 774-3.

He also highlighted deviations between accepted industry practice and the text of the UIC leaflets and queried the common practice of using increased fastening stiffness under a train, since the movement of the train may allow the vertical stresses in the fastenings to be released.

Summarising the day’s presentations, it would seem that the track-bridge interaction ‘problem’ is largely defined by the bridge configuration.

The bridge engineer should therefore be conscious of the rail systems requirements, and if necessary engage early with experienced track engineers and suppliers. However, it was noted that this could be difficult in some contracting environments, particularly if the structure is designed long in advance of the rail systems contracts being let.

Closing the workshop, David Rhodes noted the developments in standards at European level which should help clarify some of the track structure requirements. He noted that research and monitoring programmes are ongoing, such as those described by Marc Wenner, and that further development was likely.

Written by Mungo Stacy