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.
Inspections of the main cables on the Forth Bridge, followed by inspections on the Severn and the Humber, were undertaken in accordance with guidelines developed by the American authorities. This process involved removing the external wire wrapping between cable clamp positions and driving wedges into the body of the cable to enable inspection of as many wires as practically possible in eight radial positions around the circumference
The first inspection of the Severn cables took place in the summer of 2006, some 40 years after opening and approximately one-third through the original 120 year design life of the bridge. Eight sections between cable clamps were opened up. A number of broken wires were discovered and the condition of the remainder were carefully logged on a scale of 1 to 4 (1 being ‘very little corrosion’ and 4 being ‘widespread corrosion with rust over most of the surface of the wires’). Locations for inspection had been chosen at high, medium and low levels of both the main span and the side spans. It was no surprise that the sections at the lowest level were the worse for wear, because any moisture entering the cable would tend to run down and collect at the bottom.
A statistical analysis of the potential loss of strength was undertaken. This indicated a significant reduction in the factor of safety against failure compared to the original design intent. The only part of the loading that was controllable to any extent was the weight imposed by the traffic, so a regime of traffic monitoring and control was imposed, having particular regard to heavy goods vehicles. The main control measure was to prohibit HGVs from using Lanes 2 of the carriageway; this prohibition is still in force. The monitoring by Weigh-in-Motion (WiM) sensors, assisted by video camera monitoring of flows, is used to confirm that the traffic volumes and mixes remain within the assessment design loadings.
Most of the broken wires that had been found were repaired and the opened sections re-wrapped and re-painted. In addition a system of acoustic monitoring was installed to record on-going wire breaks. Measures were also put in hand to completely seal the cables in a membrane and, in time, to dry and de-humidify the interior of the cables by forcing dry air through them.
By 2011, when a second intrusive inspection was carried out, it was apparent that the measures imposed were taking effect. The humidity in the voids between individual wires was below 40% relative humidity, at which no corrosion of steel takes place, and the rate of wire breaks had been reduced to a very low level.
The monitoring and control of traffic and the de-humidification and acoustic monitoring is on-going and these measures are likely to be needed for the remaining life of the bridge. At the time of writing, the situation appears to have stabilised and, although the condition of the cables is considered unlikely to deteriorate significantly further, another examination is scheduled for 2016. Should this inspection confirm that a stable state has been reached there is scope for increasing the time between periodic inspections, inspecting only the most vulnerable (lower) panels, and allowing the relative humidity of the injected air to rise somewhat thereby saving some cost of the energy required to dry and warm the air.
L,M&N. Tower Extensions and New Cable Stays
The most obvious work was to the cable stay system. The height of the central towers was extended and additional cable stays were introduced to support the box-girder deck at closer centres so that it could carry the increased design traffic load.
This re-configuration required eight new cable anchorages inside the deck sections. The existing towers were strengthened and given new cable anchorages and saddles for the new cable-stays. New tower extension pieces were attached to the tops of the tower. This was the only part of the strengthening work that required the complete closure of the Crossing, on only two occasions.
The original cable stays, with 20 strands in a tight triangular configuration, were replaced by two open arrays of 12 strands each of slightly larger cross-section. The new configuration has the added benefit that any strand can, if necessary, be replaced without closing the bridge.
O,T,U&X. Steel Deck
The trough stiffener to deck welds under the wheel tracks were replaced, as on Severn Bridge, and stiffening was provided between the deck box and the cantilevers.
New under-deck travelling maintenance gantries were provided and joints on the underside of the deck were inspected and repaired.
New folding gantries on the upstream side provided access between new manholes in the cycle track and new manholes in the sloping side of the box deck.
P. Central Reserve Safety Barrier
The original wire rope safety barrier in the central reserve of Wye Bridge was replaced by higher and stronger steel barriers to provide greater protection to the towers and cable stays.
Q,R,S,V,W&Y Other Work
The towers of Wye Bridge were strengthened internally and externally and the viaduct pier legs were strengthened externally.
New compressed air and water mains were provided for future maintenance and the parapet was realigned.
A new tieback system was installed, connecting the top of Beachley Viaduct to the Severn Bridge anchorage, in order to resist longitudinal forces.
The base of the pier in the river was enlarged to provide protection from shipping collision.
Modification, refurbishment and strengthening of Aust Viaduct
The steel box girders of Aust viaduct were strengthened and new access and strengthening was provided to the steel box columns. New maintenance gantries and platform were installed to improve access.
Although distant from the navigation channel, it was thought prudent to construct new concrete collars to protect the Aust Viaduct piers against shipping collision. The risk was felt to be slight but was enhanced by the habit of pilots to line up ships to head directly for the piers when guiding them up the channel.
A. Tower Legs
The towers supporting the main cables were originally constructed as hollow steel boxes, cantilevered from the foundation. They carry both the vertical load from the weight of the structure and the bending moments that arise from different distributions of live loading. As a single heavy vehicle moves on to the bridge, the main cable will sag a small fraction towards the nearer tower (like a garment on the clothes-line), accompanied by similar reactions from all the other vehicles on the bridge at that time. The top of the tower will be pulled backwards and forwards, depending upon the aggregate of all the individual reactions, and it must be capable of resisting the impact of the worst case scenario, including any transverse wind loading.
Single sheets of mild steel had been used in the construction of the tower walls, each sheet being stiffened, off-site, by the addition of vertical steel flanges on what would become the interior face of the tower wall.
See picture of a new section of tower wall being lifted into place.
To relieve the box walls of some of their vertical stress, four tubular steel columns were installed in each tower leg, one in each corner.
These columns are about 120 metres high and 400 mm diameter with a wall thickness of 30 mm. Each column has 19 sections, which were threaded, one by one, through a small access door at walkway level.
The door could not be enlarged because of the high stresses in the tower walls. A grillage of steel beams and columns at the inside at the top of the tower legs transfers the jacking load from the columns to the underside of the main cable as it passes over the tower.
The columns were jacked upwards from the bottom with a force of 2000 tonnes per tower leg. Special sliding bearings locate the columns at nineteen positions within the tower to prevent buckling. The jacking lifted the bottoms of the columns by 100mm. The columns shortened in compression by about 75mm and the saddles that carry the main cables over the tops of the towers were raised by some 25mm. The Severn Bridge is an inch taller than it was before the strengthening!
B. Middle Portal Connections
The connections between the horizontal portal beams and the vertical tower legs were strengthened, both in internally and externally.
C. Tower Saddle
The main cables pass over the top of each tower leg in a saddle. During construction, the individual strands of the main cable were laid in this saddle as the spinning continued. As the main cables carry the whole weight of the bridge deck and the traffic, the stresses in the saddle are very considerable, especially the splitting effect that tends to separate the two side. The saddles were upgraded with C-clamps – similar in principle to a G-clamp used in carpentry – to help resist the splitting force.
New access platforms were also installed at the tops of the towers to assist with inspection and maintenance.
New handstrand ropes and posts were provided above the main cables. As well as providing access, these support new maintenance cradles for inspecting and painting the main cable. The main cable itself did not need strengthening but the bolts in the cable clamps, at the tops of the hangers, were renewed.
All 376 hangers were replaced, one by one, at night while traffic was restricted to one lane in each direction. The new hangers are slightly thicker that the originals and the design of the end sockets is improved to reduce the risks of kinking and corrosion
The original hangers were terminated in metal sockets filled with white metal. However, there was no hole in the socket for drainage which led to some speculative concern about the condition of some of the wires making up the hanger. In addition, just before the Royal Opening a “man with a file” was sent around to tidy up the white metal protruding from the tops of the sockets. Unfortunately he also broke through the anti-corrosion coating in several places and a number of the outside peripheral wires failed. This posed the question about the condition of the inner wires. Ultimately all the hangers were replaced, although mainly for reasons concerned with the bridge loading.
G, J & K. Steel Box Deck
The top of the steel box girder deck, over which the traffic drives, is 12mm plate stiffened by steel troughs welded to the underside. Repeated loads from the wheels of heavy goods vehicles had tended to cause local weld damage and it was decided to grind out and replace the welds immediately under the wheel tracks. The original fillet welds were replaced with a partial penetration weld requiring three passes. Changing just this one weld detail required 48 miles of welding (3 passes/weld, 2 welds/wheel track, 2 wheel tracks/carriageway, 2 carriageways, length of crossing 2 miles).
New under-deck travelling maintenance gantries were provided for inspection and painting. Also, new manholes were installed in the deck boxes to assist access for the strengthening work and for future inspection and maintenance. When the bridge was originally built, it was intended that the box girder deck would be sealed and kept dry with silica gel. For this reason the original access manholes were very small. Hence it is rumoured that a one-time Chief Highway Engineer of generous proportions was unable to visit the inside of the bridge deck.
Prior to strengthening, there was only a thin layer of asphalt surfacing on the steel deck units. A thicker layer was desirable but had to be rejected because of the effect of the extra weight on the cables.
H. Deck Bearings
Lateral bearings resist sideways forces. Rocker bearings resist vertical forces at the towers to keep the deck boxes at the correct level.
It was discovered at an early stage that the two rocker bearings at each of the towers had been designed to carry half the maximum reaction from a fully loaded deck. However, when the deck was eccentrically loaded, one of the rockers went into tension and the other carried a greater load than it had been designed for. All were replaced.
New buffers were provided to transfer longitudinal force if deck movement becomes excessive.
I. New Air and Water Mains
To help with future maintenance, a compressed air main and a Rater main were installed over the full length of the crossing. The water main is also available to be used to fight any substantial fire on the bridge.
Other Work on Severn Bridge
Movements of the deck due to thermal expansion and loading are catered for by “Demag” expansion joints in the carriageway at each tower. Increases in the required longitudinal expansion and contraction temperature allowances had to be catered for. Also it had not been recognised that the joint would need to cope with extreme racking movements due to the side sway of the main span from increased high wind design speeds. For these reasons, the joints had to be modified significantly.
It was also discovered that the large chambers in the concrete anchorages, where the individual strands of the main cable were splayed out and anchored separately, had a damp atmosphere and that the wires were beginning to show early signs of corrosion. The decision was that the chambers had to be de-humidified.
The failure of four box girders during construction caused quite a furore. The use of such girders in bridge construction had been a fairly recent phenomenon, partly because the mathematical analysis was quite daunting before the advent of electronic computers. Their popularity came at a time when the post-war surge in bridge building was getting underway. They required less steel for the construction of a bridge deck than the truss or plate girders that had been the popular choice previously.
A cantilever built using box girders poses problems for the designer. The stresses will be distributed between the various constituent plates of the boxes depending upon the accuracy of the individual elements and the precision of the fixings involved. Distortions of an individual plate could cause havoc especially during the construction of a cantilever and they are difficult to spot and to eliminate. Try to imagine how a box girder would crumple if tested to destruction by the continual adding of new units, i.e. more weight, to the unsupported end of the cantilever.
The government’s response.
In 1970, the government responded to the box girder crisis by setting up an advisory committee under the chairmanship of mathematician, Dr. (later Sir) Alec Merrison, then Vice Chancellor of Bristol University, to investigate the circumstances of the previous failures and to make recommendations for the safe re-construction of the Milford Haven bridge and the safe completion of the Avonmouth Bridge (under construction at the time). The committee also advised changes to the design rules and on the procedures to be adopted for future applications. As part of this process, a major programme of research was initiated to study the behaviour of the various components of a box girder and to consider how the mathematical analyses, then in circulation, might be improved.
The committee delivered its final report in 1974 after the results of the research studies had become available. It concluded that the primary causes for two out of the four failures were inadequate organisational procedures and poor communications. It provided an improved set of rules for the design of box girders, incorporating the findings of the latest research. It also stressed the need for the introduction of an independent check on methods of erection and on the design of temporary works, together with the need for clearer distinctions to be drawn between the respective responsibilities of the Engineer and the Contractor,
It is true to say that only one of the four box girder failures was strictly the result of lack of technical knowledge of steel box girder behaviour, although this was not apparent when the failures originally occurred. In three cases, the primary cause of failure was human error during construction and can arguably be attributed to the rapid growth of the design and construction programme out-stripping the supply of suitably qualified and experienced supervising engineers.
At the time of the design of the first Severn Crossing, the relevant live loading for such a bridge would be based on the assumption that the proportion of heavy vehicles in the traffic flow would be about 15%. This figure was based on observations undertaken in the late 1950s. It was also assumed that the heaviest loading would occur when traffic became halted on the bridge, forming a stationary queue with minimal distances between adjacent vehicles. In such circumstances, the impact on the bridge from heavy vehicles would be softened by the presence of lighter vehicles interspersed between them.
By the early 1980s, traffic counts were showing a much higher percentage of HGVs on the majority of the country’s strategic road network. It was then averaging between 30% and 33%. This was due to the increased value that operators were then attributing to two of the advantages that the motorways and improved trunk roads provided. The first was the actual savings in journey times. The second, possibly more welcome, was the increased reliability of those journey times, which is so important when operators are trying to work to tight schedules.
The widespread availability of computers by the 1970s led to the development of mathematical traffic models which provided data to inform decisions on many problems in the landuse/transportation field. Typically, these models would consist of a representation of the road network covering the area under investigation, a breakdown of that area into a number of discreet zones, a data base containing information on the traffic flows from each zone to every other zone (compiled from roadside interviews) and various tools for assigning traffic from the data base to the road network, etc. These models could be validated by comparing their forecasts for “the present day” against an independent set of data.
A model of the type described above was used to forecast the worst case loading on the Severn Bridge, prior to the strengthening work. An assessment was required of the aggregate weight of a stationary queue of traffic on the bridge, with minimal distances between adjacent vehicles. The traffic model provided forecasts of the proportions of the various types of vehicle that would use the bridge during a particular period. A separate application then selected vehicles from this mix, randomly, to generate many different queues, each conforming to the vehicle type proportions and spacing requirements, and then it calculated the load represented by each queue in terms of kN per lane meter. The figure taken forward was the load exceeded by 5% of the queues. For the Severn Bridge in the mid 1980s, this figure proved to be 9 kN per lane metre, which happened to be twice that for which the bridge had originally been designed.
The weights of the actual vehicles crossing the Severn Bridge were also measured by placing a weighbridge in the carriageway. This showed that the flows were not entirely random. Many operators of heavy lorries favoured an early morning start, when there would be relatively few cars and other light vehicles on the road to dilute concentrations of lorries. Unless that kind of situation were managed, it could result in queues being formed on the bridge with more than 30% of heavy vehicles, generating a 5-percentile loading of up to 12 kN per lane metre. Fortunately, the timely arrival of the Second Severn Crossing took the sting out of such speculations.
The main text left the construction of the Severn Railway tunnel with the departure of Charles Richardson in the aftermath of the ingress into the tunnel of vast quantities of water from the Great Spring. The story concludes below.
Dealing with the Flood Water
In January 1880, while Walker was setting up and organising pumping operations, Hawkshaw decided to lower the position of the tunnel under the Shoots Channel by 15 ft, preserving the previous gradient of 1 in 100, eastwards towards Bristol, but increasing the gradient westward to 1 in 90. It was possible to make this change without incurring large additional costs because the almost completed heading would still be wholly within the template of the amended tunnel. The greater security that this represented, against the threat of the river inundating the works, allowed him to authorise Walker to commence work at other points. Walker was then able to put in hand arrangements for recruiting and housing a very large work force (the maximum number employed, at any one time, eventually being over 3,900).
The immediate task was to seal off the heading into which the spring water had entered. Although the heading itself was quite short and wholly under dry land, it interconnected with the main heading in the Sudbrook pumping shaft. This meant that a shield had to be positioned in the shaft to seal off the problematic heading on both sides of the shaft. Like so many other operations involved in the construction of the tunnel, the task was bedevilled by practical problems and pump breakdowns. Divers were called upon to perform superhuman feats at what were extreme depths for the best survival equipment available at that period.
A good example of this came about when Walker realised that, in order to clear all the water from the Great Spring out of the tunnel, it would be necessary to close an iron door in a headwall that had been left open, inadvertently, when the men evacuated tunnel on the day the inundation had occured. The job was given to his most experienced diver, who had to negotiate all kinds of obstacles in walking for 1,000 ft though the flooded heading, all in darkness. The diver’s air pipe was connected to a static pump at the start point which meant that a very long length of pipe had to be dragged out behind him. Because the heading was filled with water under pressure, the air pipe pressed against the top of the heading, producing a great deal of friction. Even with several assistants to help convey the air pipe forward, the diver was unable to complete the assignment. The problem was only overcome when the engineers discovered that a Wiltshire man, named Henry Fleuss, had just developed a new type of diving helmet that connected directly to a cylinder of compressed oxygen strapped to the diver’s back – a very fortuitous piece of timing.
The Great Spring was eventually contained behind a long ‘cement’ headwall, 8 ft thick, into which a door was inserted. At a later stage, pumps were installed to take the water away to prevent it causing further damage – and to take advantage of the commercial potential thus presented.
Other unexpected difficulties
At the end of April 1881, when the process of ‘breaking up’ from the initial 7ft by 7ft heading to the required dimensions of the tunnel was underway on the Gloucestershire side, water suddenly burst in from the roof of the tunnel near the Sea Wall shaft. This time it was sea water; the river had broken in! Fortunately the hole, which was close to the bank on a stretch of water known as the Salmon Pool, was not very large. The shape of the surrounding rocks was such that the Pool retained a minimum of 3 ft of water even at low tide, making it impossible to locate the hole by sight. A number of men were called upon to walk slowly across the area, holding hands as they went, until one fell into the hole and had to be pulled out by his colleagues. The water was allowed to rise in the tunnel to the same level as the river itself and then, at high tide, a schooner was loaded with puddle clay and moved into position so that the hole could be plugged, using alternate layers of bagged and loose clay.
In December 1882, a major incident occurred, causing more than three hundred men to abandon their possessions and stampede out of the tunnel, shouting “The River is in”. When the initial panic subsided and it was possible to carry out a preliminary assessment, it became apparent that the situation was not as bad as first feared. In fact, the expected flooding of the tunnel did not occur. After a thorough investigation, it became clear that the panic was caused by a sudden surge in water from the Great Spring. The water had been damned behind a berm in an upper heading and, after rising and over-topping the berm, it had washed away the berm and surged into the lower heading. The men’s reaction was hardly surprising, given the conditions under which they were working. They naturally assumed that their worst nightmares were about to be acted out.
On the 9th of February 1883, a terrible accident occurred as men on the night shift gathered round the bottom of a lift shaft, to come up for supper. Four or five men had just entered the cage at the bottom, when an iron skip at the top of the shaft was inadvertently allowed to move over the lip and fall 140 ft on to the cage below, killing three men before bouncing into another group, standing by, where another man was killed and two seriously injured.
There was another emergency on 10 October 1883, when a great surge of water entered the works from the Great Spring, very much exceeding the capacity of available pumps. It swept the men and their iron skips, like so many chips, through the door leading out of the heading and into the finished tunnel, where the men were able to recover. None of the men were seriously injured but three colts were drowned. Headwalls were rapidly built to contain the flood, as water rose up against the pumps to a height of 52 ft. At first, it seemed that the works might have to be abandoned as all the available pumps, working to capacity, failed to make any headway against the floodwaters for several days. But eventually, the pumps slowly but perceptibly started to gain the upper hand, much to the relief of everyone involved. It seems that water from a nearby subterranean reservoir had suddenly been released but fortunately proved to be more manageable than seemed likely. It has been calculated that the maximum flow during this incident must have been about 27,000 gallons a minute, 16,000 more than the available pumping power could lift. Nevertheless, the Great Spring had been safely imprisoned again by the 3rd November.
Then, on the night of the 17th of October 1883, only a week after the Great Spring had renewed its assault, a tidal wave flooded the low lying area on the Sudbrook side. On a night when one of the highest tides of the year was expected and, during an exceptional storm, a tidal wave described as a solid wall of water, 5 or 6 ft high, swept in and entered the living accommodation provided for the workforce. About 90 men had just descended the Marsh shaft to continue opening up the heading into a full tunnel, near the western portal, when flood water overtopped the mouth of the shaft and fell into the workings. Three men attempted to climb a ladder out of the shaft in order to escape from the flooded works beneath; two emerged unscathed but the third was thrown off the ladder by the weight of falling water and killed. The water rapidly flooded the area at the base of the shaft, driving the men up the completed tunnel which rose at a gradient of 1 in 90. In the meantime, a workforce was rapidly gathered on the surface to build a make-shift wall around the top of the shaft in order to prevent additional water entering the works. Then a small boat was lowered down the shaft, to ferry men back to the base of the shaft where the water had by then risen to within 8 ft of the crown of the tunnel. Some had to wait hours for their turn in the boat but, by the morning of the 18th, the last remaining survivors emerged.
Modern cable-stayed bridges were only starting to be considered in the UK at the time that the first Severn road crossing was being built. And it was not until later, when electronic computers became widely available for engineering calculations, that the load bearing capacity of the later and more complicated cable stayed bridges could be analysed with any real confidence. In the earlier circumstances, engineers were reluctant to break new ground. In fact, 25 years later, when construction of the Second Crossing got underway, no cable stayed bridge in the world had been completed with a longer span than the one that was then under construction on the Severn. However, before the Second Severn Crossing was completed, the French had taken a great leap forward with the opening of the Pont de Normandie (main span 856 m). And since then, with developments in design and stronger materials, further great strides have been made with main spans now 1,000m and longer. Having said all that, the Wye Bridge is a simple but very important example of a cable stayed bridge.
The deck units chosen for the viaduct enabled the contractor, Cleveland Bridge of Darlington to complete each full span of 64 m using a simple cantilever method of construction.
Calculations confirmed that the same method could be used to erect the side spans of the bridge. However, some additional strengthening of the deck would be required if the same design of deck unit were to be used for the main span of 235 m. The obvious way to proceed with this span would be to work from both sides, building two half span cantilevers that would meet each in the middle. Some method would then be needed to lift both the leading edges to the correct height (even when stiffened, both sections of unsupported deck would have drooped, to a significant extent, under their own weights).
The addition of two simple elements of “cable-staying” would kill both birds with one stone, stiffening the whole of the bridge deck sufficiently, and providing means of lifting the leading edges of the cantilevers when the need arises. It was achieved through the addition of two posts or pylons and two long lengths of high tensile steel cable. This well targeted response provided a very effect and very efficient solution, enabling the Wye Bridge to be completed successfully and in accordance with the loading criteria of that period.
It is interesting to note that when the nearby M4 Avonmouth Bridge came to be strengthened in the 1990s (for the same reasons as the Severn and Wye bridges that will be described later), the method chosen to upgrade its load carrying capacity, was, in principle, exactly the same as the device used to build the Wye Bridge. However, in that case, the original box girders were deeper than on the Wye Bridge and part of the strengthening was effected using high tensile steel cables, pre-stressed and fixed entirely within the box structures.
The cable stays on the Wye Bridge do not carry the whole weight of the bridge deck and the traffic, but they effectively take part of the load by adding a new set of supports. The tension in each cable stay acts diagonally. This has a component of upward force, which gives support to the bridge deck, and a component of compressive horizontal force, which is absorbed into the bridge deck.
Cable stays are usually anchored to the deck symmetrically to either side of each pylon. This means that the horizontal forces on the bridge deck tend to balance out, although in doing so, they cause the bridge deck to be put into compression, especially close to the base of the pylon. And more significantly, the horizontal components of the tensions in the cables also tend to balance out where the cables are fixed to the pylons. The vertical components of the tensions act downwards on the pylons and are carried down to the foundations.
The designers of the Severn Bridge were well aware of the potential problems in the design of the long Severn crossing. In the late 1950s, using an anemometer mounted on a mast 110 feet (33 m) high (the height of the proposed Severn deck) and an array of instruments on the then existing Severn Railway Bridge some 3 miles (5 km) upstream, a spectrum of local wind speed, direction and inclination over a front of about 300 feet (90 m) was determined.
With this information, in order to avoid the risk of aerodynamic instability so graphically illustrated at Tacoma Narrows, a 1/100-scale model of the Severn Bridge deck was tested in a wind tunnel at Bedford. At that stage the proposed design was to follow American practice with the stiffening girder made of deep, about 33 feet (10 m), open-lattice truss construction as had been proposed in the 1930s.
However, by the late 1950s, the Government had stipulated that the new Forth Road Bridge should take priority over Severn. The Forth was, not surprisingly, designed to the same proposals as those being developed for the Severn but with a slightly longer main span of 3300 feet (1008 m). Mott Hay and Anderson were appointed in association with Freeman Fox and Partners to design the Forth crossing using a conventional design with a stiffening girder of latticed (truss) construction.
There was a new development in that the road deck was made of stiffened steel panels resting on the top chords of the trusses. These panels were light in themselves and the overall dead weight was also kept to a minimum by using a thin layer (c 1 ½ inch or 40 mm) of mastic asphalt as the running surface. Nevertheless, the total dead weight of the Forth deck was some 18,000 tons which required the towers, main cables and anchorages to be designed to cope with this load plus the weight of the traffic on the deck. The construction of the Forth proceeded about two years ahead of Severn.
In the meantime, scale models of newer options for the deck of the Severn Bridge were being tested in a wind tunnel. One of these options was a shallower truss configuration, it became loose and was destroyed in rather spectacular fashion. Nevertheless, and despite the disappointment, the delays in rebuilding the model provided Freeman Fox with time to develop and then test a completely novel design for the deck.
This featured an “aerodynamic” hollow box cross section only about 10 feet (3 m) deep for the stiffening girder. This increased the flexural and torsional stiffness enormously and was considerably lighter than a conventional truss. The Severn deck, as built, was only about two-thirds of the weight of the Forth deck at about 12,000 tons. Also, the suspension system for the Severn Bridge incorporated inclined hangers, which were intended to increase the aerodynamic damping.
The fate that befell the bridge over the Tacoma Narrows showed the necessity for wind tunnel testing of the design for the Severn Bridge. The first 1/100 model was destroyed during testing. However, these two disasters led to a simple and revolutionary concept that showed British ingenuity in a most favourable light, internationally, and the Severn Bridge was a great boost to British engineering.
Mott Hay and Anderson continued in their role to design and supervise the construction of the crossing but now jointly with Freeman Fox and Partners. Motts led overall and designed all the foundations. Freemans were responsible for the steel superstructures. The superstructure of the main suspension bridge was a major technical advance on all previous long-span road bridge designs.
At the time the Severn Bridge was hailed not only for its majestic appearance but also for its technical excellence. The all-up weight of steel in the Severn Bridge, 3,240 feet (988 m) span, was 19,000 tons for the deck, cable suspension system and towers compared with 39,000 tons at the Forth which had a main span only 60 feet (20 m) longer.
As early as 1943, Gloucestershire County Council started lobbying to avoid piecemeal improvements and, for the Severn Bridge, they now favoured the long span, high level, option on the Aust-Beachley line. In 1945, the Minister of Transport accepted responsibility for the crossing of the Severn and, in 1946, published a National Plan showing the first Severn crossing on the Aust-Beachley-Newhouse line with a high-speed road link from the crossing to the A48 at Tredegar Park, west of Newport.
The design of the Crossing was under way by the late 1950s. It was a time unlike any decade since. The war was over. There was hope and aspiration that a better society would arise from the carnage of the war. It was a time of enthusiasm. The proposal to build a motorway network caught the imagination and gained considerable public acceptance. It was to be pursued by the Ministry of Transport as a public programme and a team effort. Some of the personalities involved, engineers and civil servants, attracted public attention not unlike the engineers of Victorian times.
The design of a road suspension bridge needs to incorporate a stiffening girder on which the road deck is carried. This stiff girder maintains the shape of the cable as live traffic loads cross the bridge. Without the stiffening girder distributing the live loads through the hangers to the cable the behaviour would be like the rope bridges of the South American Indians, in which the rope changes shape and the walkway distorts as people cross it.
There had been a collapse of the Tacoma Narrows Bridge in America in November 1940, only some four months after opening. The spectacular failure of that bridge has been shown many times and can be seen here http://www.youtube.com/watch?v=3mclp9QmCGs.
The stiffening girder for the Tacoma Narrows Bridge was novel for the time and was constructed from two deep I-sectioned girders, one beneath each of the main cables, which were joined by cross-girders that carried the road deck.
It so happened that the natural frequency of dead load, wind-induced, oscillations (of half-span wave-length along the bridge and a half-span twist transversely) were in phase with the frequency of the vortices shed from the deck by the wind blowing up the Tacoma Narrows Channel. In a period of relatively low but steady wind, this led to increasing deformations longitudinally and transversely. These galloping deflections ultimately led to massive failure of the main girders and more or less total collapse of the deck, although the towers and main cables survived. At the point of failure, the torsional and bending deflections were enormous. Torsionally, the deck was rotating by nearly 90º and the vertical deflections were several times greater than the depth of the girders.
The designers had overlooked the effect of strong and steady crosswinds. Even Telford, who aimed at lightness in his structures, underestimated this effect in his pioneer bridge across the Menai Straights, opened in 1824. At the end of January 1826, the Menai Suspension Bridge was swept by major gales inducing extreme torsional oscillations in the deck and twenty-six suspension rods were broken. Telford’s modifications included stiffening the deck and strengthening the suspension rods.
The failure of the Tacoma Narrows made designers all over the world mindful of the wind-induced effects. Research showed plate girders were the worst form for resisting wind-induced deformations by comparison with the traditional deep truss-girder decks of earlier American and European suspension bridge designs. If the cross-wind vortex-shedding frequency is in phase with the natural frequency of the stiffening girder, deformations would build-up. However, if they could be designed to be out of phase, any potential deformations would be dampened.