Irish Railway Record Society

Collapse and Reconstruction of Malahide Viaduct

LIAM MEAGHER, PROGRAMME MANAGER (STRUCTURES), IARNRÓD ÉIREANN

THE INCIDENT

At about 18:30 on Friday 21 August 2009, two spans on the 12-span viaduct at Malahide collapsed. The collapse of the structure was due to scouring of a pier by tidal action.

The line was immediately closed to rail services and placed under Engineer’s possession. Following reconstruction works, possession was handed back at 18:00 on Friday 13 November, 12-weeks after the incident and the line was re-opened to early morning traffic on Monday 16 November.

This paper will describe: the actions taken immediately after the collapse; the initial proposals for bridge reconstruction; how the uncovering of a previously written paper changed the initial proposals; why the bridge failed; the development of the final design; and execution of the works.

LOCATION

Malahide Station is located nine miles north of Connolly Station, the terminal station on the Dublin-Belfast line, and Malahide Viaduct is located half a mile north of the station. The line is double track, the track comprising 54kg/m continuously welded rails on concrete sleepers, and the maximum line speed is 90mph (145km/h). The line is electrified from Connolly Station to Malahide Station and the over-head line equipment (OHLE) extends beyond the station but stops short of the southern abutment of the viaduct.

Trains are controlled by colour light signals, and continuous automatic warning system (CAWS) and train detection is by track circuits. The signalling is controlled through solid state interlocking. Operationally the signalling is monitored and controlled from a control panel at Central Traffic Control (CTC) at Connolly Station.

The viaduct and the approach causeways to the north and south of the bridge cross the Broadmeadow Estuary at Malahide. The construction of the railway created an inner estuary, measuring approximately 2-square miles in area, which retains water permanently throughout the tidal cycle. The bridge has 12-spans, 8-spans of 15.85m and 2-spans at each end of the viaduct of 12.25m, giving a total length of 175.8m.

Broadmeadow Estuary is protected by extensive environmental legislation. It is a candidate site for designation as a Special Area of Conservation (SAC); it is also a proposed site for designation as a National Heritage Area (NHA); it is designated a Special Protection Area (SPA) site; and it is a designated site under the Ramsar Wetlands Convention. Any works in that area consequently must adhere to the environmental conservation requirements. A particular requirement for the viaduct works was that the mud-flat exposure areas at the western end of Broadmeadow estuary not be changed. This is to ensure that bird populations, in particular winter migrating birds, are not diminished through loss of feeding grounds. The mud-flat areas have a very gentle gradient towards deeper water consequently small variations in water level have a significant impact on mud-flat exposure areas.

IMMEDIATE ACTIONS

Aerial photographs taken by the Irish Coast Guard Service early on Saturday morning 22 August show the extent of the damage. The collapsed pier, Pier No. 4 counting from the northern end, was completely washed away and the 12-precast beams spanning on to the pier, six on each span Nos. 4 and 5, are seen to be collapsed into the sea. During the course of the subsequent works, no trace was ever found of the remnants of Pier No. 4. It is believed that it broke up on collapse and, such was the force of the water, the masonry blocks were scattered a considerable distance from the bridge.

On the evening of the incident, a decision was taken that a main contractor and other contractors could be engaged immediately under emergency procurement procedures and the main contractor attended site that evening. Other contractors commenced work over the next few days: for the provision of topographical surveys, hydrographical surveys and bridge monitoring; the provision of Engineer-Divers to inspect the bridge piers and scour damage; and the provision of mobile cranes and piling equipment.

Within a few days of the incident, an independent Geotechnical Consulting Engineer, Dr Eric Farrell, was engaged to advise on all geotechnical matters; Dr Eamon McKeogh, Hydrology Dept., University College Cork (UCC), was engaged to advise in regard to the hydrology of Malahide; his colleague from UCC, Mr Michael O’Sullivan, was engaged to advise in regard to the environmental issues; and Roughan O’Donovan Consulting Engineers were engaged to carry out a peer review of all design proposals prepared by Iarnród Éireann directly and the geotechnical and hydraulic designs carried out by the other independent Consultants.

It was known by the Engineers who attended site on the evening of the collapse that the deck comprised of precast post-tensioned beams had been renewed sometime in the late 1960s and that the beams were simply supported and were not attached to each other. There was ballasted track over the viaduct and each track was supported by two beams. There were also two outer beams incorporating ballast retaining upstands and steel handrails.

Former US Secretary of Defence Donald Rumsfeld is quoted as stating ‘There are known knowns; there are things we know we know. We also know there are known unknowns; that is to say we know there are some things we do not know. But there are also unknown unknowns; the ones we don’t know we don’t know’. In the months following this incident, it became clear that other railway administrations do not always have construction details of their bridges, in particular bridge foundation depth and foundation construction details. In the first few days after the incident, Iarnród Éireann engineers were operating in a position of ‘unknown unknowns’; the remaining piers and bridge spans appeared to be in a robust and stable condition and it was assumed at that time that all piers and abutments were supported on solid foundations bearing on good ground or rock.

Initial thoughts were that the reconstruction activities should be carried out working from the existing bridge, with two independent work sites; one on span No. 3, to the north of the collapsed spans; and the other on span No. 6, to the south of the collapsed spans. This was to avoid marine works which it was thought would have necessitated specialist equipment with possible delays in procuring a suitable contractor. It was also thought that marine based works would take considerably longer to complete.

The initial design concept was that 4 No. pile clusters would be installed alongside each of spans Nos. 3 and 6 in order to support crane out-riggers for cranes positioned on these spans. The cranes were initially to remove the suspended track and the collapsed beams. At that stage piling rigs were to be used to install bored-pile foundations to the front of Pier Nos. 3 and 5. The intention was that the collapsed Pier No. 4 would not be replaced but that precast pre-stressed beams with a concrete cast in-situ deck would span from Pier No. 3 to Pier No. 5.

Given the importance of the Dublin-Belfast line, a significant element of the national transport infrastructure, there was an expectation at the highest levels that the line would be re-opened as soon as possible. Having discussed the initial design concept with the main contractor, the Department of Transport was advised on the day after the incident that the line would be re-opened within three months.

In the days following the incident monitoring equipment was installed to detect possible movement and/or settlement at any point on the entire viaduct. Survey monitoring targets were attached to each end of the outer beams on each remaining span. The x, y and z position of each target was measured daily to determine the three dimensional movement at the top of every pier and abutment. Tilt monitors were installed at Pier Nos. 2, 3, 5 and 6 to measure north-south and east-west rotations and vibration monitors were installed at Pier Nos. 3 and 5 and read-outs were available continuously.

Engineer-Divers commenced an underwater survey on Sunday 23 August. Their first priority was to survey the remaining piers, Pier Nos. 1 to 3 and 5 to 11, and the abutments for scour damage. The second priority was to determine the depth of scour erosion between Pier Nos. 3 and 5.

On Monday morning, the Engineer-Divers were able to confirm that there was no immediate risk to the remaining piers and abutments. Some scouring had occurred to the east of spans 9 and 10 but the scour did not pose an immediate risk to Pier Nos. 8, 9 and 10. They also reported that almost 66% of the entire ebb flow, then in spring tide, was passing between Pier Nos. 3 and 5 with a velocity of approximately 4.5 to 5.0 m/sec.

SITE ACCESS

A problem familiar to all railway infrastructure staff is that of access and in particular access to incident sites. In this case access to the southern abutment was easily achieved by arrangement with the owners of the adjacent Malahide Marina and boat-yard, located on the Up side of the line beside the southern abutment. To the north of the viaduct, public road access was approximately 2 km from the bridge and necessitated access through private lands. It was possible to make use of an access road previously constructed across these lands to facilitate previous railway works.

By Monday 24 August, ballast was being drawn in to the site to create a ballast bed over the track in anticipation of bringing in heavy-lift cranes and piling rigs.

HISTORIC PAPER

On Monday evening a copy of a paper published in JOURNAL No. 143 of the Irish Railway Record Society, October 2000, written by Mr Oliver Doyle, Manager Resources & Central Traffic Control at that time, was turned up. It described the history of the viaduct and contained information which was of considerable significance to the works then planned to commence on site.

When constructing the railway line from Dublin to Belfast, a very direct route out of Dublin city had been taken, hugging the coast as far as Laytown Station. To achieve this route it was necessary to construct causeways across a number of estuaries, including Fairview, Malahide and Rogerstown.

The map extract, page 384 above, shows the causeway at Malahide. The inner estuary to the west of the railway measures approximately 1-mile north-south and 2-miles east-west. Prior to the construction of the railway, the tidal waters flowed in and out of the estuary unimpeded. With the construction of the causeway and viaduct, all of the flow was concentrated into a narrow channel.

The original viaduct was constructed in 1843. It had 11 spans of 52 feet (15.85m) with a total length of 572 feet (174.4m). The abutments were of masonry construction, supported on timber piles. The piers and bridge deck were of timber construction also supported on timber piles.

The obstruction to the natural tidal flow caused a very powerful current through the bridge spans and within a very short time after the railway coming into operation, it became clear that erosion of the soil into which the timber piles were driven was causing settlement. To address the problem approximately 90,000 tons of stone was placed beneath the bridge, encapsulating the timber piers. The stone formed a rip-rap weir, extending between the north and south abutments. The purpose of this weir was twofold, to protect the seabed against further erosion and to reduce the volume and velocity of water flowing through the bridge spans. The weir was 130 feet (40m) wide at its base and 30 feet (9.2m) high with steeply sloping sides.

The settlement of the deck was dealt with by packing up the rails, by up to 3 feet (0.92m) in places. By 1859 however it became clear that the timber was suffering decay and a decision was taken to renew the bridge. The construction requirements were quite onerous in that it was necessary not to disrupt rail traffic during the course of the works.

Having considered a number of options, it was decided that the new structure would have wrought iron spans supported on masonry piers. The piers were to be supported directly on top of the stone weir. Given the concerns in regard to the condition and stability of the original timber piles, it was decided not to disturb the original piers and to retain the original abutments. It was also decided to minimise obstruction of the waterway and to construct the bridge in three stages: to construct three piers at the northern end, renew the first three spans on both the up and down lines, take down the original piers for those spans and clear and level the waterway; then construct four more piers and spans in a similar manner; and finally complete the last four piers and spans.

One of the options that had been considered was to use cast-iron girders with cross-spanning brick arches (jack-arch construction) and having a track laid on ballast. It was known that the cast-iron would have been more durable in the marine environment; however the board of directors were mindful of the risk that cast-iron could fail suddenly with catastrophic consequences and it was directed that a wrought iron structure be used instead.

It is worth noting how the piers were constructed. The piers were initially built-up dry at Dublin Terminus, now known as Connolly Station. Each stone was marked up, the piers disassembled and the stone delivered to site. There the piers were again built-up dry alongside the causeway, but this time the piers were built upside down.

The weir had to be prepared in order to receive the new piers. At low tide, the water in the outer estuary can fall to a level approximately 3.0m below top of weir level. It was possible to restrict the flow of water from the inner estuary by damming one span at a time. Sections of track were laid on top of the weir, spanning from pier to pier, having convex plan-curvature towards the western face, filled with sods and pitched with stone on both faces. While the dam did not prevent all of the water from passing through it allowed the masons to construct the pier foundations under a shallow depth of water.

To prepare the pier formation level, loose stones were first removed to a depth of 15 inches (0.45m) below the general level. Around the hole thus formed round stones were laid on edge without cement. Then stone fragments, gravel and iron turnings were placed in the hole and rammed with a 5 cwt (260kg) block of iron. When a firm level base was achieved, a framework of rails, having the profile of the base of the pier, was laid down and adjusted to the correct level using blocks of stone, the intermediate rails being packed with flat stones. The interior space was then filled with flat stones and grouted-in. At this stage, the first layer of ashlar was laid and the remaining stonework for the pier carried on from there. It only took 5 months from the laying of the first stone on the first pier to the completion of the last pier, during which time seven deck spans had also been completed.

The information contained in the historic paper was of significant importance to the Engineers in drawing attention to facts not known to them at that time; that the bridge piers and abutments were founded directly on top of the weir.

THE NEW PRIORITY

Diving conditions during the spring tides were particularly hazardous between Pier Nos. 3 and 5 because of the potential for instability of the fallen beams and the velocity of water flowing around them. It was necessary to restrict diving in that area to that brief period between tides when the incoming tide just reached the same level as the out-flowing waters.

On Tuesday morning, the Engineer-Divers reported on the depth of scour between Pier Nos. 3 and 5. The maximum scour was measured to be approximately 3.6m below the general top of weir level and occurred where Pier No. 4 had been.

Knowing then the pier construction details and the importance of the weir to the stability of the entire structure, it was clear that unless the weir could be secured against further erosion and scour effects, there was a significant risk that the scour hole at Pier No. 4 would spread and undermine Pier Nos. 3 and 5 and potentially undermine others as well.

At that time, there was only a narrow strip of weir remaining intact between the scour hole and the inner estuary. There was concern that a full breach of the weir, with a scoured channel running from inner to outer estuary, could cause significant environmental damage because tidal flows would be dramatically changed; the ebb tide would flow from the inner estuary for a longer period; the head of water would be much higher; and consequently further erosion and scour could develop very quickly.

The primary concern at that stage switched from reconstructing the bridge to stabilising the weir and all efforts were directed to that end. The Hydrology Dept., University College Cork was engaged with a first priority to advise in regard to plugging the scour and reducing the volume and velocity of water discharging between Pier Nos. 3 and 5. It was decided to construct a haul road on the eastern side of the viaduct, with access through Malahide Marina, in order to fill the scour hole with stone.

THE WEIR

The haul road was constructed on top of the eastern side of the weir with a top surface about 500mm higher than the general weir level. It was recognised that the weir would need to be substantially strengthened on both the eastern and western sides and a total of 25,000 tonnes of stone was laid to construct the haul road. The contractor placed 1.2m diameter steel pipes east-west in the middle of each span, a metre below haul road level, to assist in accommodating tidal flows. Heavy boulders, 15 to 20-tonne weight, were placed along the eastern slope of the weir with smaller boulders and stone placed on top. Special permission was granted by the local authorities and the Gardaí (police) to permit the contracted quarries to operate continuously and for the boulders and stone to be delivered to site throughout day-light hours seven days a week. By Sunday 30^, the flow through Pier Nos. 3 and 5 had been arrested. The haul road acted as a buffer, preventing the water from flowing through the scour hole, and the high tide flow over the haul road was seen to be equalised through each span.

TIDES

The high and low spring tides extend from +2.22m OD to -2.38m OD, a range of 4.6m. The weir level prior to collapse was approximately +0.5m OD.

Spring tides at high water cause a significant volume of water to enter the inner estuary where it is retained. On the ebb tide, the water in the outer estuary falls below to a low water level almost 3.0m below top of weir level. The high water retained in the inner estuary cannot fully discharge before the tide turns again and the incoming tide replenishes the inner estuary once more. It was found that during spring tides, there was always a flow of water across the weir to a depth of at least 600mm. In order to allow works to continue unimpeded during spring tides, it was necessary for elevated stone platforms to be formed to allow construction equipment operate in dry conditions.

The high water during neap tides sometimes only exceeds the weir height by approximately 400mm. This causes a lower volume of water to enter the inner estuary and the water retained can fully discharge to top of weir level during the ebb tide before the tide turns and the incoming tide replenishes the inner estuary once more. Some of the works had to be restricted to neap tide periods: pouring concrete to the pile cap on Pier No. 4; installing the micro-piles; and some other works.

THE CAUSE OF THE COLLAPSE

On Sunday 23 August, a chance encounter with a Scout Leader of the Malahide Sea Scouts led to copies of their photographic records of flow conditions through the weir being made available to the Engineers. The Sea Scouts have for many years used the weir as a training ground on their kayak courses. The ebb flow over the weir created conditions suitable for training those who then progressed on to white water rapids elsewhere. The Scout Leaders had often taken photographs of participants and had thereby captured the weir flow conditions in the background.

It is now clear that general erosion had been taking place for some time over the entire length of the weir. In the past stone had been placed on the weir following storms but no replenishment had taken place since the mid-1990s. It appears that the top level of the weir had been slowly eroding with a loss of stone over time.

From the Sea Scout records we could establish that the scour which caused the collapse of Pier No. 4 was not evident in March 2009. By July however it is clear that spans Nos. 4 and 5 had suffered significant scouring with deep channels having been formed in the middle of each span. As the erosion developed these channels became deeper and wider and consequently more water was drawn into the channels exacerbating the problem. In the days leading up to the collapse, more stones were scoured away, deepening the channel until finally the pier foundation was undermined and the pier collapsed.

The sketch by the Engineer-Divers shows the development of two areas of scour to the east of spans 9 and 10. The eastern crest of the weir was locally eroded away progressively until in plan the weir crest was ‘horse-shoe’ shaped. On the ebb flow, the water passing over the weir crest gathers momentum and the flow becomes turbulent. This can cause erosion along the crest and the side slopes of the weir. In the case of the ‘horse-shoe’ shaped crest the turbulence occurs within the weir, around the ‘horse-shoe’ crest, and causes erosion which draws more water into that area exacerbating the problem.

In spans Nos. 4 and 5 similar erosion had occurred but in that case the weir crest was eroded away until the ‘horse-shoe’ shaped crest progressively grew and eroded the weir directly beneath the bridge and then beyond the western edge of the bridge until it had almost broken a channel through the entire width of the weir. At that stage the increase in water volume and velocity caused progressively more destructive forces to undermine the weir and ultimately undermine the pier foundation causing the pier to collapse. 

In 1967/68, the wrought iron spans were renewed. At that time the construction details of the bridge were known and the importance of the weir was well understood. When re-constructing the wrought iron spans it was recognised that the previous practice of removing deck timbers and unloading stone from ballast wagons to replenish the weir would no longer be possible and would be difficult by alternative means. A decision was taken to grout the top level of stone in the weir to a depth of 1.5m over the entire length of the weir. The grouting was successfully carried out and the weir was stable for many years. Over time however, the grout broke down and was itself eroded. Unfortunately having ‘fixed’ the problem, the knowledge of the weir’s importance faded, and with the passage of time there was a loss of corporate memory and the structural importance of the weir was not appreciated. This was found to be the underlying cause of the incident.

THE WORKS

Having constructed the haul road and equalised the flow through each span, it was realised that despite the initial reluctance to ‘get our feet wet’, a new opportunity presented itself. It was realised that cranes could be brought along the haul road to remove the collapsed bridge beams, and other works, including the re-construction of Pier No. 4, could now be carried out relatively easily. It was decided to abandon the design work then underway and develop a new design, to re-instate Pier No. 4 and construct two new spans in a style similar to the original but using precast pre-stressed beams and an in-situ deck.

Tracked crawler cranes were used to tandem lift-out the tracks which were then cut into 15m lengths and placed on span 3, immediately behind Pier No. 3. Then the collapsed beams were removed and transported to the railway gantry-yard in Mullingar. The signalling cables on the up side of the line were-routed to the down side and all were supported by catenary wire spanning from Pier No. 3 to Pier No. 5. The cables remained in use for the duration of the works. Having cleared the site, the scoured hole was then filled with 150mm stone, that size being chosen in order to facilitate the installation of driven steel piles.

A second haul road was constructed from Bissets Strand Road, alongside the western side of the southern causeway, as far as the southern abutment. This facilitated the importation of stone to be laid on the western side of the weir for strengthening purposes. The importation of stone from the northern side of the bridge was discontinued once the decision had been taken that Pier No. 4 could be re-instated, working from the haul road on the eastern side of the bridge.

In Mullingar, six of the eight larger bridge beams were load tested to confirm their load carrying capacity. Having completed the load tests, the beams were cut into sections in order to examine the condition of the post-tensioning tendons and to confirm that the grouting had completely filled the tendon ducts. The tests confirmed that the beams can safely take the loads to which they are subjected to in service. The level of corrosion on the tendons was quite small and no loss of grout was observed.

No site investigations had been carried out before completion of the haul road, but because there was little time available, it was decided to drive 273mm diameter tubular piles for the foundation of Pier No. 4 and thereby determine the depth to rock. Fortunately the main contractor was at that time constructing the Clonmore Bridge on the M52 Link Road near Mullingar where such piles were being used and a sufficient number was immediately made available for this project. 22 No. piles were driven at Pier No. 4 and rock was found at approximately 25m below top of weir level.

Site investigations did proceed as soon as the haul road was completed. Boreholes were taken to the east and west of each pier and thereby informed decisions in regard to the piling and micro-piling.

A decision was taken that while the works were underway, micro-piles would be installed below all the other piers and the abutments. 15 No. raking micro-piles, using 50mm diameter Dywidag hollow bars, were installed at every pier, and an additional 8 No. vertical piles, using 75mm diameter Dywidag hollow bars, were installed at Pier Nos. 3 and 5. The micro-piles were designed as friction piles and were not taken to rock level but were terminated approximately 15m below top of weir level. These micro-piles do not carry the entire applied load but were installed to give supplemental support to the piers.

When the micro-piling started and the vertical piles were being installed at Pier No. 5, it was noted that some settlement was taking place. At that time, the piling contractor was drilling a steel casing through the stone rip-rap beneath the pier, the rig sitting on the bridge deck and the drill passing through vertical holes previously cored down through the masonry pier. When the figures showed that there had been settlement of 20mm at the eastern side of the pier and 7mm on the western side, a cross-level of 13mm, the micro-piling was stopped. It was concluded that the method of installing the steel casing by percussive drilling, using high air pressure at the toe of the casing to drive the drill head, had led to the disturbance of the sand/silt layer beneath the weir rip-rap which caused voids to develop, with the consequent effect of pier settlement.

At that stage the installation of the 76mm diameter Dywidag hollow bar was to be carried out in three stages: core a 200 mm diameter vertical hole through the masonry pier; drill and install a 168mm diameter steel casing from the base of the masonry pier through the stone rip-rap to the sub-strata sand/silt layer; and drill the 76mm diameter Dywidag hollow bar to the required depth and grout.

Following consideration of the problem it was decided to stop the casing installation 1m before the bottom of the rip-rap layer. At that stage the casing was filled with grout which was allowed to permeate into the rip-rap and the underlying sand/silt. The grout was then allowed to set overnight before the installation of the 76mm diameter Dywidag hollow bar which was then drilled through the grout and in to the sand/silt layer.

Subsequent monitoring showed the process to have been successful.

Pier No. 4 was completed by the 9 October and the new spans were completed two weeks later. The track was re-instated by the 10 November and all micro-piling was completed by Wednesday 11 November.

At that stage, load testing of the bridge was carried out over two days using an 071-class locomotive and laden Tara mines wagons with axle weights of 18.5 tonnes. It was found that the loading had no effect on the bridge piers and no settlement occurred.

Possession was handed back to the Operating Dept at 18:00 on Friday 13 November, twelve weeks after the bridge collapse. Train testing of track circuits was carried out over the weekend and trains were positioned for passenger services which commenced on Monday morning 16 November.

Since the line re-opened, further works have been carried out to strengthen the north east causeway and finish off the weir. A total of 112,000 tonnes of stone, including the 25,000 tonnes used to form the haul road, was used to widen and strengthen the weir. The top of the weir was profiled, with V-shaped channels formed between the piers extending from the eastern to western crest. The effect of these channels is to draw the main flow of turbulent water away from the piers and thereby further mitigate the risk of scour at the piers.

Copyright © 2011 by Irish Railway Record Society Ltd.
Revised: November 18, 2011 .