The Trust was established as a charity in 1999 to promote public education in the engineering, environmental and associated disciplines involved in the planning, design, construction and maintenance of the Severn Crossings. The current Trustees are all Chartered Civil Engineers who have had close involvement in senior positions with the design, construction and maintenance of one or both of the two bridges.
The Trust’s objectives are:
To provide a permanent record of the many professions and disciplines involved in construction using as illustration the First Severn Bridge, the Second Severn Crossing, together with their associated Approach Roads, the Severn Tunnel and the former ferry crossings of the Estuary;
To promote the web site for the general public, schools, colleges and universities each side of the Estuary and the national and international engineering professions;
To explain to the general public the broad spectrum of engineering disciplines and other professions involved
The Trust operated the Severn Bridges Visitor Centre. After opening in December 1998 it attracted thousands of visitors a year. It housed an exhibition showing the history of the river crossings using interactive displays, video films, pictures, models and descriptions until closure in 2008.
Why have you published this site?
This website offers a permanent public display for the exhibition material from the Visitors Centre, expanded with engineering detail for those who are interested. Our goal is to provide and share information on the background to the need for and the construction of the Severn Bridge, opened in 1966, the Second Severn Crossing, opened in 1996, and information on earlier crossings of the Estuary. It celebrates the broad spectrum of engineering disciplines and other professions involved, and the environmental and construction achievements of these two crossings.
When was it created?
The main story of the crossings presented on this site is taken from the exhibition materials created ready for the Visitors Centre opening in 1998.
In 2009, when funds for a new exhibition building in the M48 Service Area at Magor, which the Trust was contemplating using for its Exhibition, were withdrawn, the Trust chose to create a virtual exhibition instead. Then in 2012, they recognised that publishing a more detailed website might also attract engineering professionals and academics, and chose to add materials of deeper interest and more technical value. The very detailed papers available in the Proceedings of the Institution of Civil Engineers, that describe the works covered by the web site and which were written by the engineers who had been responsible for doing the work, helped the trustees greatly when writing about the engineering detail behind the main pages.
The web site was completed and published in 2016, to coincide with the 50th Anniversary of the opening of the Severn Bridge, and the 20th Anniversary of the completion of the Second Severn Crossing.
Who else has supported this effort?
The Trust has received financial support from sponsors across the construction industry, including consultants, contractors, suppliers, professional bodies, other trusts and Government Transport Departments. The Trustees are most grateful for the support received from these many organisations to establish the Trust and support its work.
The trustees are also particular grateful for the assistance of Stephen Jones for his agreement to the inclusion of material from his published books, relating to the previous history of Estuary Crossings within the new web site, and to Neil Thomas of Photographic Engineering Services for his agreement to the inclusion of a number of his excellent photographs that are to be found in the more detailed passages of the text that relate to the Second Crossing and can be viewed using the appropriate “more information” links embedded in the Main Text.
The 12 minute video on the construction of the Second Road Crossing, which is located at the end of the penultimate page of that chapterand entitled “Building the Second Bridge”, was made with financial support from the Chartered Institution of Highways and Transportation using original material made available to the Trust by the John Laing Trust and the British Film Institute. Copies of the DVD may be obtained from the Trust at a cost of only £5 each, including postage and packing, by contacting firstname.lastname@example.org
On the Severn Bridge, the two main cables act a bit like a washing line. The tension in a washing line supports the weight of the clothes that are pegged to it. In the same way, the tension in the main cables supports the weight of the deck and traffic. The bridge deck is hung from the main cables using wire hangers (rather than clothes pegs). And because the main cables are held up by the towers, the weight of the whole bridge is carried down through the towers, on to the underlying foundations.
If you put something heavy on a washing line, it will sag at that point. With a suspension bridge, the road is supported by a stiffening girder, which spreads out the weight of the traffic, so avoiding excessive sag under an exceptional load. If you hang something on a washing line away from the centre, the point will not only sag but it will also move towards the nearest end (try it!). Similarly, as a heavy load travels over a suspension bridge, it will not only dip downwards at the point of the load, it will also move longitudinally towards the nearest tower.
If you stand on the walkway of the Severn Bridge, you can feel it moving as the traffic travels over it. If you stand by one of the towers and watch the expansion joint, you can sometimes see the whole bridge moving as the weight of the traffic travels across. We should not worry that the bridge moves. It is meant to do this. This is how it absorbs the weight of the traffic and transfers it into the main cables.
The tension in the main cables carries the whole weight of the bridge deck and the traffic. This tension is resisted by the anchorages at each end, just as the tension in a washing line is resisted by whatever it is tied to at each end. And because the main cables are held up by the towers, the weight of the whole bridge is transferred through the towers to the ground.
Readers seeking further information on the historical context to the modern estuary crossings are advised to consult the two sources from which most of the material in the previous chapters has been taken.
Stephen Jones generously invited the Trust to make use of relevant material from his book, “Links with Leviathans” the third volume of his trilogy; “Brunel in South Wales” now published by The History Press of Stroud, Gloucestershire. Chapter 4 of the book, entitled “Severn Gateway”, covers all aspects of early transport links across and around the lower Estuary in great detail and was an invaluable aid to the Trust in covering the story.
The essential source for a real understanding of the construction of the Severn Railway Tunnel was written by Thomas Walker, the contractor who assumed responsibility for the project in 1877. His comprehensive book entitled “The Severn Tunnel. Its construction and difficulties (1872-1887” was published in 1888. Facsimile editions by Kingsmead Press of Weston-super-Mare were published in 1969 and 1990. The website’s ‘page’ on the tunnel is based almost entirely on information obtained from that book.
In terms of engineering achievement, the 19th century railway tunnel beneath the Estuary undoubtedly ranks alongside the two great twentieth century bridges. However, with the passage of time and the virtual absence of any visible indication of its presence, its importance and stature can easily be forgotten or overlooked. To help restore the balance, a fairly lengthy coverage of its construction has been provided on the website to allow a wider public appreciate the scale of that achievement. The story deserves to be better known.
The standard procedure for erecting the deck of a cable-stayed bridge would be followed, starting at one of the pylons and adding single deck units, alternately, first to the centre span and then to the back span, so that the growing section of deck remains in balance over the pylon, in order to minimise the bending forces in the pylon legs. A production yard was established in the main Avon construction complex for the assembly and casting of the deck units, each 7.3 m long. It should be noted that, unlike the twin pairs of deck units used on the adjacent viaducts, those adopted for the bridge were of full deck width.
The structural steel work for the deck was all fabricated in Italy and trial-erected, there, to check the geometry. It was then shipped to Avonmouth and brought to the construction yard with each deck unit in six separate pieces. The steelwork had previously received its first two coats of epoxy primer and micaceous iron oxide. It was then assembled and positioned under the form-work on which the deck slab would be cast. While still in the construction yard, the slab reinforcement was then securely fixed in position and the concrete was poured, leaving room for a 2 m wide lateral concrete stitch to be added later, to link each new unit to the previously erected unit.
The maturity of the concrete was carefully monitored so that the false work could be struck as soon as the concrete was capable of supporting the required loading. Each deck unit slab was cast while its steelwork was still attached to the previously completed unit, in front, and the assembled steelwork of the next unit, behind. The previously completed unit was then removed and carefully weighed before being sent to the painting shed. This was because the deck alignment control system needed to know the actual weight within 2%. The final coat of polyurethane paint was then applied to the steelwork in an enclosed and heated shed.
The special deck units that would sit on the lower cross beams of the pylons, on the back span piers and against the viaducts, were all prepared separately so that they could be introduced into the casting sequence at the appropriate time. The deck units were transported to the pylons in the correct sequence, using a motorised barge. The barge was fitted with laser and computer technology that enabled it to maintain its intended position within 500 mm, so long as the tide was running at no more than 2-3 km/hour. As the tide in the estuary could run at up to 10 km/hour, lifting operations were restricted to one hour each side of high tide.
Erection of the deck started at the Avon pylon with the special unit earmarked for the lower cross beam of the pylon. This unit was lifted by a crane mounted on a jack-up barge, into position on the lower crossbeam and temporarily tied down. The next two, standard, units were then lifted by the same crane, one to either side of the central unit to provide sufficient room for two cranes on the deck. After cable stays had been attached to the second and third units and their stresses adjusted, two DSL cranes were lifted onto the embryo deck, one to deal with the centre span, the other with the back span.
Successful erection of the deck units was achieved through the use of two DSL cranes, each comprising a pair of luffing jibs (see photographs/ illustrations) mounted on a common body, with a lifting frame permanently attached. For operational purposes, the DSL cranes would stand on the leading edge of the previously erected deck unit, anchored through the top of that unit, to the structural steelwork below the deck slab. When a new deck unit had been lifted into position, it would be held by the crane while it was fixed to its predecessor and hooked up to the pylon by its own pair of cable stays. The crane would then be moved forward by hydraulic rams that were fitted to its body, as far as the leading edge of the new unit, so that the process could be repeated. Unable to swivel under their own power, the DSL cranes had to be positioned accurately on the deck so that the operation of lifting a unit, from a barge stationed 50 m below, could be kept under careful control.
Each deck unit would be transported to the bridge site on a motorised barge as part of a pair, the first for inclusion in the centre span, the next for the back span – always in that sequence. The appropriate DSL crane would pick up the unit from the barge and hoist it to deck level. The new unit would then be bolted to its predecessor in the sequence, using the bolt locations that had been determined during the preliminary assembly of adjacent units in the construction yard. The alignments of the deck units would be checked after bolting. Then, in situ concrete would be poured to provide the lateral 2 m concrete stitch between the new unit and its predecessor. The weight of this concrete was required to balance the deck, but the cable stay installation could get underway as soon as the stitch had been poured.
The weight of the special deck unit that would sit above the first back span pier proved to be beyond the capability of the DSL crane. Also, the presence of the caisson under that pier prevented the transporter barge from approaching close enough to allow the DSL crane lift the adjacent centre span unit from the barge. It was therefore decided that the two units should be cast together and a crane, mounted on a jack-up barge, was used to lift them into place. Then, when the first back span deck unit had been raised and bolted to the adjacent pier unit, eight cables were used to tie down the embryo deck to the pier.
Erection of Cable Stays
All cable installations were undertaken by international specialists, PSC Freyssinet. Each completed cable consists of a number of parallel seven-wire galvanised strands locked in position by tapering wedge anchorages. The number of strands per cable varied from 19 to 75. The first strand was threaded through the cable sheath and one end of the combination was picked up by a tower crane. The other end was threaded through the lower anchorage and inserted into a mono-strand stressing jack. The upper end would then be threaded through the appropriate anchorage aperture in the pylon to be wedged and locked into place against a tapered hole that had been machined into the anchorage plate. The mono-strand jack at the lower end then applied a tension equivalent to about 60% of the required final tension, forcing the wedges into the anchorage.
A second strand was then threaded through the lower anchorage, up the cable stay sheath and through the upper anchorage where it was wedged into place. This second strand was tensioned using the mono strand Jack until the load measured by the load cell was equal to that measured on the first strand. The process was continued until all the strands in the cable had been installed and stressed to 60% of the required cable load. The additional extension of the cable required to increase the load to 100% of the required value was then calculated and each strand was stressed to achieve this extension. This was because the load applied to the strand could be controlled more accurately by extension than by load measurement. A similar cycle was followed for the corresponding back span cable stay, keeping the lengths of deck, either side of the pylon, in balance.
Some cable oscillations were experienced during the final months of construction and it was decided that the secondary cables, or aiguilles, should be added to provide a lateral link between the cables to dampen these oscillations. Five aiguilles were fitted to each group of 30 main cables. These aiguilles comprise 27 wire strands passing each side of the cables and they are connected to the cable-stay ducts by purpose-made clamps. At the lower end, they are fixed to the deck and tensioned. The damping characteristics of these aiguilles proved to be disappointing so, at a late stage, two dummy longitudinal ‘girders’, made of pressed steel panels, were introduced below the concrete deck between the main outer steel girders. These provided additional aerodynamic damping by breaking up the airflow under the deck and the final configuration has proved effective under all wind conditions experienced to date.
When all the deck units had been erected from the Avon pylon, the DSL cranes were moved to the Gwent pylon, to begin a similar sequence there. The final deck unit of the main span was lifted from the Gwent side. Before this operation could go ahead, the size of the gap between the two cantilevers had to be checked. It was found to be correct within 20 mm but with different gaps between the pairs of plate girders. The splice plates that would connect the pairs of plate girders were drilled according to the measurements obtained.
Adding the final deck unit
In order to insert the final unit into the centre of the main span, it would be necessary to find a means of making small adjustments in the positioning of the existing sections of deck. First, the decks were effectively fixed, longitudinally, to the pylons by modifying the hydraulic circuit on the shock transmission units (described in a paragraph below) to prevent movement. It was then possible to move the two sections of deck apart by pumping hydraulic fluid into one side of each unit, so providing space to erect and secure the last unit from the Gwent side in the normal manner. However, when the last cable stays were stressed, there was a difference in level between the two sections of deck, due to the presence of the DSL crane on the Gwent side. The crane was moved back, past the pylon, to equalise the levels. The shock transmission units were then used again, this time by removing some fluid to move the two sections back towards each other, to enable the girders to be spliced together. The modifications to the hydraulics of the shock transition units were then removed to allow the unit to perform as originally intended, i.e. as combined springs and shock absorbers. The final two in-situ concrete stitches, one either side of the last deck unit, were then poured to complete the main span.
Joints and bearings
The movement joints between the bridge and the viaduct, in effect, allow the two elements to move independently of each other and so erection of the final deck units, at the extremities of the bridge, were straightforward. Details of those joints are described under “Construction of Viaduct deck”. Vertical and lateral shear forces are transferred from the ends of the bridge to the viaduct, respectively, by pot bearings and sliding elastomeric bearings on steel beams fixed to the viaduct deck. The completed bridge deck is a single entity, without any internal movement joints.
Hydraulic shock transmission units were fitted to each pylon to transfer live load forces from braking and other minor incidents, and from seismic action, from the deck – equally – to both sets of pylons. These units incorporate elastomeric springs to keep the deck central about the pylons. Vertical forces from the deck are resisted by sliding pot bearings situated on the lower cross beams of the pylons. Lateral wind forces are transferred to the pylons by elastomeric bearings, fixed to the sides of the deck.
On each back span pier, guided sliding pot bearings accommodate longitudinal deck movements. These bearings also transmit vertical loads and wind loads to the pier on which they sit. Four vertical tendons were used at each back span pier to tie down the deck to the pier.
The Bridge was eventually opened on the 5th of June 1996 but, on the 12th of May a charity Event was arranged. Members of the public were invited to walk across the Bridge, and back again, to raise money for a charity of their own choice.
The two largest caissons, 13 m wide by 53 m long, were used to construct the foundations for the bridge pylons. These caissons weighed 2,000 t each which was the maximum weight that the construction equipment could manage. The design for the shells of these caissons had been dominated by load effects during their construction in the casting yard. Special bracing was needed, both for the manufacturing process and for the journey from the construction yard to the bridge site.
The process of producing and transporting these very large 2000 ton caissons was identical to that used for the viaduct caissons (see Construction of the Viaduct Foundations). Special care was taken in choosing their precise locations in the estuary because of the proximity of the main navigation channel, with its extreme tidal range and fast currents. The caissons were placed in the lead up to high tide, using the very large double sheer-legged cranes, mounted on a jack up barge (a large barge with special supports in each corner that could be jacked down to the river bed to provide a firm foundation for deploying the cranes). Additional temporary restraint was provided by hydraulic jacks fixed to steel beams that had been concreted, at low tide, into holes in the rock around the outside of the caisson. As in the case of viaduct foundations that were considered to be at risk of ship impact, the pylon caissons were filled with in-situ concrete as soon as they had been made fast.
The hollow pylon legs each rise to a height of 137.2 m above the caissons, 149 m above ordnance datum. With the exception of the crossbeams and the anchorage tie beams, they were constructed entirely of reinforced concrete. Each pylon required 38 separate lifts of concrete, of between 3.5 m and 4.48 m in height, starting at the top of the caisson and using self-climbing formwork. Concrete for the pylons was produced in a batching plant located on the caisson and it was placed by skip, using tower cranes anchored to the pylon. The top of the caisson was also used to store materials and provide temporary accommodation for the workforce.
The crossbeams and anchorage tie beams were prefabricated on shore. The crossbeams were transported to the pylons, in the same manner as the caissons, being lifted off the transporter vessel using cranes mounted on a jack-up barge to provide a stable platform. The lower crossbeams weighed 1400 t each. Each was lifted on to the partly constructed pylon and was then cast, in situ, into the pylon itself, during the next concrete pour. Later, the upper crossbeams, weighing 900 t each, were lifted to rest, temporarily, on to the previously fixed lower beams and, when the pylons were nearing completion, they were raised to their eventual positions, using strand jacks mounted on brackets fixed to the pylons. They were post tensioned to the pylon using seven pairs of stressing cables which were then grouted into the structure. The cable anchorage tie beams were lifted into position by a tower crane located on the completed lower cross beam.
The task of placing the concrete was complicated by the complexity of the steel reinforcement, especially in the vicinity of the cable anchorage locations. Because of this difficulty, and the fact that each anchorage has its own unique orientation, the anchorage tie beams were cast on-shore where the position, and the vertical and horizontal alignments of the anchorages, could be more easily controlled in a purpose-made formwork jig on-shore.
During the design process, it became clear that plain reinforced concrete pylons would not be able to cope with the design bending moments and so each pylon was pre-stressed vertically. Following their completion, 16 vertical tendons were positioned in ducts within the hollow interior of each, and then tensioned. The cables were anchored beneath the level of the lowest cable anchorage and, again, into a slab at the top of the pylon. The program dictated that both pylon legs should be under construction at the same time, necessitating two complete sets of construction equipment.
Access to both pylons required careful planning. The Avon pylon was connected to shore at low tide, by the temporary causeway, while the Gwent pylon had to be accessed from the Avon shore by boat – across the Shoots Channel. At high tide, a boat or barge was the only means of access to either pylon. A small jetty was constructed on the Gwent shore to assist the transport of personnel to the Gwent pylon.
The key to the successful erection of the viaduct was the acquisition of a bespoke launching gantry. It was 234 m long, designed to carry a maximum load of 200 t, and it was used to build the viaduct deck. It consisted of a pair of parallel steel trusses that stretched its entire length and two transverse primary supports. It also had secondary supports at the front and rear of the trusses, and a pair of crabs. It operated above the deck and was able to move itself forward. Also, for a reason that is described below, it was able to move itself from one side of the deck to the other, as the work progressed.
The gantry was used to build the viaduct deck using the balanced cantilever method. In outline, this involves building out two sections of deck from each pier, in the form of cantilevers, one to the front of the pier, the other behind, so that the new length of deck remains in balance over the pier. There are 27 segments in each span, excluding those fixed to the pier, and when 13 have been erected on each side of the pier, the gantry would be moved forward to the next pier and the whole process would be repeated. After the first and subsequent repetitions, a further complete span would have been constructed, except for a final segment in the centre, and that segment would be placed before the next repetition commenced.
In this particular case, there was a further complexity because the decision had been taken to construct the deck using twin box girders, one under either side of the deck, in order to halve the weight of the box segments. Both box girders had to be built at the same pace in order to keep the growing length of deck in balance over the pier, laterally as well as longitudinally. The effect of this was to require the gantry to be moved from one side of the bridge to the other, after placing just four deck segments (one in front and one behind the pier to catch up with the second pair on the previous cycle, and then one more of each, to surge ahead again). In anticipation of all this, the gantry had been constructed to deal with just half the deck at any one time.
All deck segments were carried from the construction yard to the viaduct by a transporter that would travel to, and along, the completed section of deck to deliver the segments. Construction started at the Avon side, with the placing of the two special segments that would sit, side by side, on the first pier. Pier segments were more robust and complicated than others and they exceeded the weight limit of 200 t imposed by the use of the gantry. SRC’s solution to this was to construct these segments in two parts, the first in the usual manner, the second by adding a further 180 t of in-situ concrete to each, after the segments had been placed on the piers.
The gantry was initially positioned to one side of the viaduct location, with its front primary support on the Avon abutment, the rear support behind the abutment and the front leg on a temporary bracket which cantilevered forward from the front face of the pier. It was then able to place the first special deck segment in the correct position on the pier but on temporary supports. The gantry was then winched backwards a short distance, before being moved laterally to deal in similar fashion with the special segment for the other side of the pier. With the second pier segment in place, the 180 t of additional concrete, which had been omitted from the casting of the segments because of the weight restriction mentioned above, was added.
The primary supports of the gantry were then moved forward until the forward support was above the pier that had just received both its deck segments, and the rear support was on the abutment. The gantry was then fixed, temporally, before other deck segments were added to those on the first pier, as described above, using the principle of balanced cantilevers. Working out from the pier, new segments were placed, the first to the front of the pier and then one behind it, to keep the growing sections of deck in balance, longitudinally. As soon as a segment was lowered into position, it would be stressed tightly against its predecessor, using temporary tendons. After delivering a single pair of ‘balanced’ units, the gantry would be moved across to the other side of the viaduct, to repeat the process, so that the balanced cantilevers on both sides of the viaduct could be brought forward at the same pace, to minimise the lateral loading on the pier.
It is important to appreciate that a single move of the gantry did not produce two whole spans of box girders but four half spans, centred on the pier, i.e. two pairs of “balanced cantilevers”. Not counting the special pier units, 27 segments of equal length went into the construction of each completed box girder span, a total of 54 per span.
The segments themselves consisted of reinforced concrete box shells, open at both ends. They were manufactured on both sides of the estuary, approximately half each side, using five casting bays on both sides. The segments were constructed with a single pour of concrete into a mould that already contained the necessary cage of steel reinforcing bars and ducts for the tendons. Each segment would be match-cast against its future neighbour. It was possible to manufacture 20 units from five moulds in a six-day week, in all but the worst of weather conditions.
In order to ensure the correct and secure positioning of the tendons, complicated internal diaphragms were installed in all the deck segments that were located over piers. Two other, more simple, diaphragms were added to each span, one in each of the deck segments located at each end of the low-height central section of the span. The purpose of the additional diaphragms was again to accommodate locking devices and deflectors in order to prevent any slippage of tendons (in compliance with the requirement that the free length of a tendon must not exceed 40% of the span) and to effect a change in the direction of some tendons. The locking devices and deviators were part of purpose-made steel assemblies, bolted into the appropriate segments after they had been removed from the casting shed and were no longer on the critical path. When tendons had been stressed, they would be wedged in place to prevent any longitudinal movement or loss of tension.
As each deck segment was delivered and lowered into position using the gantry crab, epoxy glue was applied by gloved hand to the joint surface, and the new segment was temporarily pre-stressed to its predecessor. Once a balanced pair of segments had been erected, the permanent cantilever tendons at the top of the segments were stressed so that the temporary tendons could be de-stressed and released, allowing these tendons to be reused.
When all 27 segments had been erected on both carriageways, the permanent bearings were fixed in position and the vertical load transferred from the temporary jacks on the pier segments to the bearings. The nominal 2 m stitch in the centre of the deck between the twin box girders was concreted and the permanent continuity tendons above the bottom flange were stressed to enable the gantry to be moved forward again to start the next cycle. Further continuity stressing was installed at a later stage in the erection process.
The permanent post-tensioned tendons within the box girders were all enclosed in high density polyethylene ducts that were temporarily supported on frames suspended from the soffit of each unit. The ducts were made continuous between the anchorages and they were passed through holes in the deviator plates into which pre-bent galvanised steel ducting had been fitted. The ducts were fixed both at the anchorages and at the deviator plates. Strands were introduced into the ducts and all tendons, even the 250m long continuity tendons, were stressed from one end, using a multi-strand jack, and were then secured. Finally, the ducts were injected with hot wax to complete the protection system.
A special construction cycle had to be devised for the final half spans connecting to both the Avon and Gwent abutments where the balanced cantilever method was not available. The launching gantry was located behind each abutment in turn and used to deliver the deck units, placing them on to a pair of beams supported from the ground by a military trestle. When they were in position, the units were glued and temporarily stressed together before being connected to the completed balanced cantilever span that was perched on the first pier. Tendons were then introduced, stretching between the abutment and the first pier, and stressed to form a coherent beam out of the 26 separate units. Eventually, the first span was stressed to the rest of the viaduct as far as the first movement joint.
The ends of the slip roads on the west side of the M 49 junction extended 240 m past the Avon abutment, causing an additional complication. The two box girders had to deviate from each other to accommodate the road geometry so that the concrete stitch between them had to be made progressively wider. Precast concrete beams were constructed in the yard, to fit transversely across the gap between the box girders. Then assisted by a false-work frame from the ground below, the beams were used to support construction of the concrete stitch.
To complete the roadway surface, the top face of each pair of viaduct units and the in-situ concrete stitch between, was sprayed with a polyurethane waterproof membrane. Then a 20 mm thick sand asphalt carpet was laid to protect the waterproof membrane and this was followed by 100 mm of hot-rolled asphalt.
Mechanical and Electrical Works.
The following features were among those added to the completed viaduct as part of the provision for its future maintenance:-
• A monorail access train, and stations, suspended beneath the viaduct and bridge decks.
• An access gantry for the main span of the Shoots Bridge.
• An access gantry for each stretch of viaduct (plus the adjacent back-span of the Shoots Bridge).
• Access lifts in each of the four pylon legs of the main bridge.
• Lighting in all the accessible internal parts of the bridge deck, pylons, viaduct decks and piers.
The monorail train operates over the whole length of the crossing between the east and west abutments and may be accessed from both ends. It is designed to carry personnel, with a trailer for carrying equipment and materials. Intermediate steel platforms, that were suspended from the bridge deck or fixed to the pylons, allow access to the cable stayed bridge.
Access to the main span of the bridge was provided by a conventional under-slung steel gantry supported on rails beneath the deck and at the edges of the deck. However, the under-slung access gantries which service the viaduct spans and the back-spans of the bridge are of a novel design, comprising a central rigid platform from which a rotating steel truss access beam is suspended. This is to allow the gantry to pass through the space between the twin legs of the viaduct piers by turning the beam through 90 degrees. Hydraulically operated telescopic platforms permit access to a number of different levels beneath the deck.
Good marine access to the location of all pier foundations was important. The better ground conditions and accessibility are found on the east side of the river and so the consortium decided to concentrate activities on that bank. To ensure access across the English Stones at low tide, a 2 km long causeway was constructed from the Avon shore to the east pylon of the main bridge. Over a 500 m section, the bed was 2.5 metres lower than elsewhere and it was necessary to build it up to the same height as other parts of the causeway so as to maximise usable time. Precast concrete culverts were installed under this part of the causeway to allow the water to ebb and flow without overtopping the roadway prematurely.
A protective earth bank was used to assist work on foundations on the west side of the estuary. Large diameter bored piles, suitable for use in soft ground (alluvium and peat), were used for the first eight foundations on this side. SRC decided to use pre-cast concrete caissons in areas where the rock was sound, or where problems were caused by the strength, or the rise and fall, of the tides. Caissons could be relied upon to spread the load, resist overturning, and, particularly, to resist horizontal sliding caused by ship impact. They also provided good protection against strong currents and exceptionally high tides. All caissons were built and launched from the main construction yard on the Avon side, with its special launching ramps and towing equipment.
On the Gwent side, a protective earth bund was constructed out from the shore, over the alluvium, and large 2 metre diameter bored piles sunk to support the viaduct pier bases (Ref Figure 8).
Near the Avon bank, large diameter bored and de-bonded piles were used to support the caisson that would form part of the special foundation in the vicinity of the Severn Bridge Tunnel (shown in Figure 11 and 16 of paper 1141 Kitchener and Ellison), see “Design of Viaduct Foundations”. The piles bear onto rock below, well clear of the tunnel. The caisson is, in effect, cantilevered out from the pile cap. One end of it is located close to the tunnel but all the loads pass into the ground at a safe distance from the tunnel (Figures 4 and 5 refer).
The hollow caissons were all specially braced to resist stress during transit. They were transported in holding frames on huge crawler tractors, and delivered down the approach ramp. From there, they were loaded onto a special transport barge that was controlled by 4 separate engines, one on each corner, to ensure that the vessel would be able to hold its position, at high tide, to allow the caisson to be lifted off the barge and lowered into its final position. (Figure 13 SAR 3 and Figure 14 Lisa A refer).
Sequence of operations involved in moving a Caisson from the casting yard to its location in the Estuary.
Copyright; Neil Thomas of Photographic Engineering Services.
Each caisson was placed onto a prepared bed of bags filled with grout, where they would be held by the crane until the grout had gained sufficient strength. In some cases, breaks would be created in this supporting arrangement to ensure that river water could flow freely through the base of the caisson. This is because of the possibility that the huge structure might otherwise sink into the underlying formation and effectively seal its base. Large and heavy, it may be, but a caisson with a sealed base would be treated by the rising tide as a de-facto vessel and so it might be caused to lift momentarily and so be dislodged from its desired location. To avoid this, the tide was therefore allowed to enter until a sufficient window of opportunity was available for the base of the structure to be properly sealed with concrete, with no further ballast needed to prevent any possibility of floating.
The pier units were precast in the Avon compound, in purpose-made steel moulds, using vertical match-casting to ensure an accurate, tight and secure fit between adjacent units in the completed pier. High density polystyrene (HDPE) ducts for the pre-stressing tendons were cast into the walls. The steel U-tubes, to link between the two HDPE ducts, had previously been incorporated into the steel reinforcement cage for the last pour of in-situ concrete infill at the top of the caisson. Great precision was required to ensure that the tendon ducts were properly positioned right up through the pier. Refer to diagram (? Ref: ). Epoxy glue was applied by gloved hand to the upper surfaces of the precast pier units before the first unit was placed in position. This process was repeated to build the pier to full height, using additional thicknesses of epoxy glue to maintain the alignment.
A temporary platform was provided across the top of each pair of ducts, to assist with the stressing of the tendons within the piers. The tendons, each comprising 19 strands, were made upon on this platform and then winched through the ducts. They were stressed from both ends simultaneously and hot wax was pumped up from the bottom of each duct to protect against corrosion. On completion, the profiled chapeaux were cast on top of the caissons to protect against ship impact. The more vulnerable piers were further protected by the pouring of mass concrete into lower parts of the internal void within the caisson.
The bridge deck is constructed from individual full width units, each 7.3 m long and containing four separate components:-
• a reinforced concrete slab at the top;
• two 2.15 m deep steel plate girders, set well apart to support the slab;
• transverse members, at 3.6 m centres, between the plate girders;
• open steel trusses, below the items mentioned above, to add stiffness to the girders.
The reinforced concrete slab varies in thickness between 200 mm and 350 mm. It is 470 mm thick in the vicinity of the anchorages.
The cable spacing of 7.3 m at deck level, dictates the transverse girder spacing, each girder being located at the mid point between adjacent pairs of deck anchorages. The anchorages are contained in special steel assemblies on the outside of the deck. The main girder web thickness varies between 20 mm and 25 mm, increasing to between 35 and 40 mm in the regions of the pylons and the back-spans. The design of the stiffening truss beneath the bridge deck is dictated by the need for a rectangular space to accommodate the monorail-suspended access train. The truss is constructed out of steel T-beams, I-beams and cruciform sections. Solid plate-girders are needed at the back-span piers, and at the connections to the segments of viaduct on either side of the bridge.
The government stipulated that the cable-stayed bridge must be able to function with any one cable removed for replacement – and that, when combined with 10% reduction in live loading, safety must not be jeopardised by the accidental removal of any two cables. This, and the desire to avoid using large cables, explains the relatively close spacing, at 7.3 m, of the cable anchorages along the deck.
The whole structure is stiffened by the presence of a pair of back-span piers at each end of the bridge. They are all built in a similar manner to the viaduct piers. This is to counter the fact that the long cables supporting the end of back span distant from the pylon, are far from vertical and so less efficient in dealing with vertical loads, making that part of the deck susceptible to vertical movement caused by loading on the centre span. The deck needs to be tied down to the back span piers, to avoid the possibility that a sufficient part of its weight might otherwise be lifted off the isometric bearings on the piers to allow it to move laterally, relative to the piers. The stressing cables are located within the hollow piers and they are made up of wire strands that are housed in wax-filled ducts to protect against corrosion.
The deck of the bridge is slightly wider than the viaduct deck because the cable stays pass through the deck, to be anchored to the steelwork below. The cables consist of between 19 and 75 parallel seven-wire strands, held together in a sheath. The actual number of strands depends upon the position of the cable within the fan; the greater the length of a cable, the more inclined it would be and so the less efficient in dealing with a vertical load. In other words, the longer the cable, the greater the number of strands it would contain.
The Royal Fine Arts Commission recommended the use of a light colour for the cables and a pale green was chosen. The black sheathing of the cables was covered in adhesive plastic sheeting of that colour, prior to installation. The pale green colour enhances the bridge appearance. It is carried through on the wind shielding and street lighting, blending well with the wide expanse of estuary, sky, and Welsh mountains in the background.
Erection of the bridge on the basis of the geometry of the completed, but unstressed, structure would inevitably have resulted in distortions. It was necessary therefore to take full account of the stresses that would be experienced when the bridge structure is complete and fully dead- loaded, and the effect that these stresses would have on the ultimate dimensions of the connected elements. Two independent methods were used for these calculations and the results compared. The calculations were then repeated at each erection stage to allow any necessary adjustments to be made. In the event, very good agreement with the design geometry was achieved.
The main pier foundations for the cable stayed bridge were positioned on the English Stones, near a rocky outcrop known as Gruggy. They had to avoid nearby fault lines and keep clear of the deep and steep-sided channel that had been carved out over time and was now the main navigation channel, known as The Shoots, with its peak currents and turbulence.
The foundations had to withstand the dead and live loads from the bridge and be able to cope with possible ship impacts. They were carefully positioned following extensive investigations. A number of special navigation aids for the use of pilots on ships using the estuary were agreed with the Gloucester Harbour Trust, including a harbour radar system which replaced the previous unsightly proposals for ship protection islands in front of those piers that would be at relatively high risk.
Once again, caissons were deemed to be the most suitable choice for foundations in these conditions. They would be larger than those used for the viaduct, measuring 53 metres in length by 13 m wide and weighing 2000 tons each. They were designed to withstand impact from ships weighing 6500 t. The caissons would be heavy enough to prevent overturning and shear keys, in the form of 2 m dia. tubular steel piles cast into 2 m deep holes but allowing the upper parts of the piles to be encased in the concrete infill that was then poured into the bottom of the caissons, would resist any sliding which might otherwise result from ship impact.
The main piers of the bridge were designed to look similar to those of the viaduct. And although the caissons used in the main bridge foundations were larger than those for the viaduct, similar design criteria were applied. Back span piers were added to prevent the back spans from being lifted by loading on the central span. (see Design of the Shoots Bridge Deck).
The navigational channel in the vicinity of the crossing, known as The Shoots, is located in a deep but relatively narrow natural trench, little more than 300 m wide at river bed level. A symmetrical cable-stayed bridge, the centre-piece of the new crossing, would be constructed across this channel, with a main span of 456 m and a total length of 900 m. It is linked to the shore by major viaducts on either side, both 2 km in length, bringing the total length of crossing to 5 km. Following consultations with shipping interests, the clearance of the bridge was set at 37 m above mean high water level. This allowed for a notional rise in sea level (as a result of global warming) of 1 metre.
Two twin-legged portal-frame pylons, rising 149 m above the river bed (101m above deck level) provide the main features of the bridge. Each pylon leg includes the upper anchorages for 60 cables, thirty of which radiate down on to the main span and 30 to the back span, making a total of 240 cables in all. The cables pass through openings in the concrete bridge deck, which is 34.6 m wide, but they are actually anchored on top of the deck steelwork.
The 456 m main span reflects the need to locate the pylons at a sufficient distance from the steep side slopes of the main channel to ensure the stability of the bridge foundations. Additional stiffness for each back-span is provided by two piers that are similar, in both appearance and spacing, to the piers supporting the viaduct.
The base of the pylon legs are 10.2 m wide, along the line of the bridge, reducing to 5.4 m at the top.
Transversely, they are a standard 4.0 m wide. The legs of the pylons are constructed in reinforced concrete, using the “lift over lift” method, with the aid of self-climbing formwork. The legs are hollow, with lifts and steel stairs installed to provide access from the caissons to the deck level and to the top of the pylons. The importance of the pylon proportions to the appearance of the bridge was widely appreciated and was subject to much scrutiny from the architect, the engineers – and the Royal Fine Arts Commission.
Each pair of pylons is provided with two hollow, precast reinforced concrete cross beams for transverse stiffness. In order to reduce the amount of steel reinforcement required, these upper cross beams are post-tensioned. To cope with the tensile, splitting forces caused by the cables anchored on either side of the pylons, small precast beams have been fitted across the internal space between the front and rear faces of the pylons and post-tensioned. The tendons for this operation are contained within ducts and anchored outside the pylons. Hot wax is injected into the ducts to exclude air and water and so protect the tendons. The anchorages for these tendons are visible between the exit points of the cables.
There is a tendency, under certain loading conditions, for the weight of the deck to be lifted off the back span piers. This can produce significant bending moments in the pylons and, to cope with these moments, the pylon legs need to be post-tensioned vertically. 16 vertical tendons, each comprising 19 strands, are installed within HDPE (high density poly-ethylene) ducts in the front and rear walls of the hollow pylons, and tensioned. The tendons stretch from a point below the first cable anchorage, to the slab at the top of each pylon leg.
The crossing includes a total of 4.2 km of viaduct in two sections, one on either side of the cable-stayed bridge, each comprising 20 standard spans of 98.12 m. Twin box girders were used in preference to a single, wider box, because the smaller, separate units would be easier to handle and would not need to be post-tensioned transversely. Also, there would be space between the twin box girders for an under-slung train to provide access for maintenance, along the whole length of the structure. Both box girders, in every span, contain 27 separate viaduct segments, each 3.5 m in length. In addition, a special segment of similar length was placed on every pier. Each of the twin box girder flanges would be 15.6 m wide and there would be a 2 m concrete stitch between them, bringing the total width of the viaduct deck up to 33.2 m.
The key to the load bearing strength of the viaduct spans is the use of the high tensile steel tendons that would be added during the actual construction of the viaduct deck. The tension would be introduced into the tendons and then locked in, using wedges or anchors in a process known as “post-tensioning” (because the stress is introduced after the individual deck segments have been constructed separately, and then positioned). Each new deck segment would be introduced and held in position while temporary tendons would be used to tension it sufficiently hard against its predecessor to ensure that it would remain secure when the lifting tackle was removed. Later, the temporary tendons would be removed and replaced by the permanent set.
The soffit of the viaduct deck shown on the Illustrative Design was curved. However, the consortium rejected this idea in favour of the tri-planar form illustrated on the diagram below. This would avoid having to use a large number of blocking devices around the upper surface of the lower flange to insure that the lower tendons followed the line of the curve fairly closely. With this change, the number of blockers or deflector diaphragms within each span was reduced to two, thus increasing the speed of production and reducing cost.
The Consortium had to comply with special government requirements relating to the use of tendons. When used for post-tensioning the precast concrete units together, tendons must be “external”, i.e. they must remain accessible and replaceable, even when enclosed within a hollow structure – and the free length of any such tendon must not exceed 40% of the span. SRC’s solution was to fit internal restraining devices, mounted on what could be described as a vertical diaphragm, at each of two locations within the span, to grip and hold all tendons passing those points and, in some cases, to act as a deflector to change the direction of the tendon.
In total, 2,400 separate deck segments, each 3.5 m long, would be installed in the viaduct. They would be constructed on shore, in the form of a hollow reinforced concrete box open at both ends and weighing up to 200 t. In elevation, the viaduct segments are 7.0 m deep at the piers and 3.95 m deep through the central section of the span. See illustration. Care had to be taken to optimise this design because the tri-planar form is less pleasing to the eye than the curved option. The addition of a 2 m wide in-situ concrete stitch between the top flanges of the twin segments of the viaduct box girders, would enable the in-situ concrete deck slab to be laid right across the width of the deck in a single operation.
Access to the deck for maintenance is provided by a train that runs on rails under the central reservation, in the void between the two boxes. A number of conventional gantries with working platforms underneath the deck would also be provided and they would also allow for access up on to the top of the deck. These gantries would be suspended from both outer edges of the deck and they would be able to move horizontally along the deck.
Longitudinal movement joints are provided at the centre of every fifth span, i.e. 392.48 m apart, to restrict progressive collapse of the structure. This means that there are five separate, homogeneous sections of deck on each side of the cable-stayed bridge, each capable of small longitudinal movements, relative to its neighbours. The actual joints are inserted at mid-span to ease arrangements for maintenance. They comprise stainless steel/PTFE sliding surfaces which allow longitudinal deck movements, while restricting movements in other directions and, in doing so, they transmit vertical and horizontal shear forces from one side of the joint to the other. Similar joints are provided at the junctions between the viaduct and the main bridge, and at the two end abutments.
The viaduct deck is supported by elastomeric bearings (essentially, large thick rubber pads) located on each pier. These bearings transmit vertical and horizontal forces from the deck to the piers. Each section of deck therefore rests, or “floats”, on elastomeric bearings. No point of the deck is actually fixed and so each section can move a short distance, independently of its immediate neighbours. This enables temperature length changes, together with braking and seismic forces to be distributed over several piers.
Seismic activity provides the critical loading case for the bearings which are designed to restrict the pier’s horizontal acceleration. The bearings must be large; 1.3 m x 1.3 m and 500 mm deep, to cope with the seismic loading. Lateral deformations for the seismic design case are estimated at between 150 mm and 200 mm, while the maximum deformations for more normal loading conditions are expected to be in the region of 80 mm. A fail-safe shear key device is incorporated into the pier head to restrict movement in the case of a seismic event. Every bearing was shear-tested for normal in-service loads, and prototype bearings were subject to a full range of tests, including a shear load representing the full ‘design’ seismic load.
Every deck segment located over a pier must be vertically post-tensioned to that pier to provide the required stability against out-of-balance loadings that will occur during erection of the deck, especially during the construction of the balanced cantilevers, described later under “Construction of Viaduct Decks”.
The viaduct widens out by 13 m over the last 240 m at the Avon end, to accommodate the slip roads from the M49 junction. In this location, the central part of the deck between the viaduct’s two concrete box girders, comprises an in-situ reinforced concrete slab supported on 2 m deep precast concrete beams which span between the inner webs of the box girders.