Dawlish, a seaside town on the south coast of Devon about 12 miles fromExeter, was originally a fishing port which grew into a well-known resort in the eighteenth century. In 1830, Isambard Kingdom Brunel designed a pneumatic railway which ran along the seafront of the town. The wide-gauge ‘atmospheric railway’ opened on 30 May 1846 and ran between Exeter St. Davids and Newton Abbot. The first passenger train ran in September 1847 but after technical problems, the Directors abandoned the project in favour of conventional trains and the last atmospheric train ran in September 1848.
Today, while the line is a particularly memorable and scenic route, it is one of the most exposed in the country and the continual battle with sea erosion and corrosion makes it expensive to maintain. Furthermore, the railway station at Dawlish is in the town centre immediately adjacent to the beach and, although most of the station is not the original Brunel buildings, it is all Grade II listed – including the footbridge which links the station platforms. However, the station is so close to the sea that in storm conditions this bridge is drenched by the spray from breaking waves and blasted by wind-born debris (sand) from the beach.
The station was originally only a single platform (on the inland side), but a second platform was added in 1858. The existing station buildings were opened in 1875 after the previous wooden buildings burned down in 1873. However, the footbridge was reconstructed in 1937 using serviceable girders that were taken from Park Royal & Twyford Abbey tube Station (a disused station on what is now the Piccadilly Line) after that station had closed in 1931. The bridge had a single square span of 17.5 metres, being supported on padstones built into the masonry of the station buildings (the staircases to access the deck being partly stone masonry within the buildings,
partly timber suspended from the deck). The walkway was approximately 1.8 metres wide. The bridge had a roof with wide overhanging eaves, though the nature of the exposure (sea spray coming in horizontally) is that these had not significantly protected the structure. Additionally, one half of the span had timber cladding to somewhat protect bridge users from spray and sand.
The girders were riveted built-up sections of early steel. The webs had a clearly visible X-brace detail, the visibility of which was exaggerated by the corrosion patterning. This detail, together with the riveted construction, was identified as being a key part of the ‘character’ of the structure and early discussion with the planning and listed building authorities identified that, if replacement was to be adopted, then these features would need to be carried forward into the replacement structure.
Maintain, repair or replace?
The steel bridge was in very poor condition with extensive, well-established and very visible corrosion. Detailed inspection in 2004 had categorised the condition as ‘fair’, though this conclusion was somewhat questionable since, even then, many holes and significant corrosion points were identified. The next detailed inspection in 2010 identified that the defects reported in 2004 had deteriorated significantly and made a less positive assessment of the condition.
A like-for-like repair option was developed, but it required replacement of a large proportion of the structure – eight out of 20 web panels, nine out of 22 web stiffeners and the full length of both flanges on both girders were to be replaced. All the repairs would be carried out with HSFG bolts replacing the existing rivets. Thus, although the structure would look superficially unchanged, most of it would be new.
A further study was carried out by Tony Gee and Partners in 2011. In addition to the known defects, severe corrosion to the girder / cross girder connections was also identified. The condition of the structure had deteriorated to such an extent that some holes in the web had been patched temporarily with hardboard just to remove the risk of public injury on a sharp corroded edge. Detailed analysis identified that not only could the structure not carry the specified imposed load due to corrosion of the members, but even in an ‘as new’ condition the bridge had been under- strength due to a lack of strength and stiffness in the U-frames providing lateral stability to the top flange of the plate girders.
Thus, although like-for-like repairs were estimated to cost approximately £600,000, Network Rail’s preferred option was a replacement structure. A new steel footbridge was considered, but while this could be detailed to reduce the susceptibility to corrosion, the location was such that it could only restart the continued (and probably unwinnable) fight. The station was listed and a simple ‘off-the-shelf’solution would probably not be acceptable.
Accordingly, a wholly fibre-reinforced polymer (FRP) structure was considered, both to simplify installation (by reduced weight) but also, more critically, to reduce ongoing maintenance costs and requirements in the extremely hostile environment. Although this was identified as being initially more expensive than steel, the whole life costs for the structure should be much reduced.
The bridge was required to withstand ‘normal’ Eurocode footbridge loading and criteria were agreed between the designers and Network Rail. In recognition that the bridge deck may fill with pedestrians (when a train disgorges a large number of passengers at once) the full ‘load model 4’ loading of 5 kN/m2 distributed live load was applied.
Parapet loading in Eurocodes was not well resolved at the time of the design, so this was taken from older standards such as Highways Agency document TD19/06 ‘Requirements for Road Restraint Systems’.
Wind loading is also a conceptually simple code- compliant situation, although the location is exposed and the wind loads are accordingly relatively high.
Aerodynamic stability had to be considered, though this is a variation from the standard as that document does not strictly apply to bridges which have a roof, and the material is not in the list the standard covers.
Lightweight bridges are potentially prone to dynamic response from the aerodynamic loads from passing trains. It was agreed that this effect would be analysed during detail design based on criteria developed during the design of the Bradkirk footbridge (issue 57, July 2009).
Initial concept studies considered various truss arrangements. However, due to concerns regarding listed building consent, it was decided to revert to a plain girder design which closely followed the geometry and aesthetic of the original bridge.
To assist with design development and also to obtain planning approval and listed building consent, several computer models and rendered visualisations were generated. A full-scale sample section of girder was also produced to assist the
planners and conservation officer to visualise the FRP structure. The conservation officer insisted that the bridge replicated the aesthetic of the original riveted structure, so imitation rivet heads were bonded to the structure. In some locations, structural bolts are included to provide a backup to the bonded joints and prevent peel stresses in the bonds. These bolts were stainless steel with dome heads to blend in the rivet heads and fastened with tamper-proof shear nuts.
The complete structure excluding the stair units was predicted to have a mass of only five tonnes, which is probably around one third of the mass of an equivalent steel structure.
The structure of the bridge was analysed using computer models and finite element analysis (FEA). Analyses carried out included static, buckling, eigen-value and dynamic response. The roof panels were found to be beneficial in increasing torsional stiffness and vibration frequency.
Simple aerodynamic stability checks indicated that the critical wind speeds for vertical or torsional vortex shedding induced vibration were above 1.25 x design mean wind speed and therefore did not require more detailed investigation.
Design and manufacture
Both the primary structure and the parapet were made up from 1.66 metre deep side girders, each formed from foam cored shear webs, moulded by film infusion using fire retardant epoxy resin and biaxial glass fibre reinforcement, capped top and bottom with pultruded angles and plates to form the girder flanges. Web stiffeners made from pultruded plate provide additional lateral support to the girders, connected to transverse angles below the deck. The girder includes a camber of 120mm along the length of the bridge, which improves the aesthetics and provides drainage to the deck.
The deck was formed from ‘Composolite’ pultruded panels, spanning transversely between the girders. These panels are very lightweight with a skin thickness of only 3mm. To ensure adequate robustness and resistance to local concentrated loads, an additional 3mm thick pultruded plate with a gritted non- slip finish was bonded to the top surface of the deck.
The deck was bonded to the girders and also forms a shear panel to resist horizontal wind loading, removing the need for diagonal bracing below the deck. Unfortunately, the deck has to terminate 2.7 metres from the ends of the bridge to leave room for the stairs. This creates a long length of girder acting as a cantilever and unable to resist the large wind side load. To strengthen these cantilevered areas, additional lateral support plates were fitted to the flanges external to the girder.
The roof transverse frames were fabricated from back-to- back pultruded angles to form T- sections with bonded and bolted joints. The roof frames support longitudinal purlins made from pultruded box section to support the roof panels. These frames also provide lateral restraint to the top of the girder.
To further increase the lateral and torsional stiffness, a much stiffer transverse frame is provided at each end of the bridge. Roof panels are made from standard corrugated fibre-cement panels, and the stair units at either end of the bridge are made from a single FRP moulding, including the stair treads, risers and side panels, hanging from the bottom flange of the bridge girder.
The entire bridge structure was manufactured from FRP materials. The majority of parts are pultruded with glass fibre reinforcements and fire retardant polyester resins to achieve the required structural properties. The pultruded parts used to fabricate the main girder flanges were pultruded in 17.5 metre lengths for the span of the bridge to avoid the need for joints.
Parapet panels have PET foam cores and were moulded from fire retardant epoxy resin reinforced with biaxial glass fabrics using film infusion. Each parapet is in three sections with simple, bonded, butt-strap joints. The final structure was painted to achieve the aesthetic requirements and to provide environmental protection to the composite structure.
Since the materials are still considered ‘novel’ by Network Rail, a rigorous design and checking process was implemented. Tony Gee and Partners was appointed to prepare the Form A (Approval in Principle document), complete the design and the Form B (Design / Checking certificate).
Design work undertaken by subconsultant Optima Projects was validated by Tony Gee. In addition, the structure received a full ‘Category 3’ independent check by Parsons Brinckerhoff.
Network Rail managed the process of obtaining listed building consent. This required the production of reports and options studies justifying replacement rather than repair, and was driven through by Network Rail’s own planning and listed building specialists with support from the designers and the initial repair study. Consent was eventually obtained, although the requirements of the process resulted in various detail changes to the configuration from what would be structurally necessary.
The final design mimics the form of the existing structure in order to minimise the visible changes to the various views that take in the station. The new bridge remains part of the listing, and therefore has become probably the first listed FRP bridge in the UK.
Many thanks to David Kendall, Optima Projects Ltd; Ian Smith, Tony Gee & Partners; and Wendy Gough, Network Rail for help with this article.