The idea of a huge temporary sand bund in the sea, won the Balfour Beatty/Costain joint venture the £94M contract for the 1,000m long Cardiff barrage, believes the JV's project director Paul Neal. A Balfour man, he runs the 65%/35% combination as a single team.
The 200m long by 100m wide bund enclosed all the complex concrete and mechanical structures for the southern side of the barrage during the two year building programme, allowing operations in the dry right down to 18m deep foundations. And, economically, the sand would be re-usable, helping form the permanent rock and sand embankment being built from the north side of the bay.
The designer's suggestion in the tender was for the barrage's major structures, from locks to the fish pass, to be built inside sheet-piled cofferdams. 'But that would have meant working over water,' says Neal. This would have been inconvenient, risky and probably even dangerous in the record tidal range in Cardiff. Water rises 75mm a minute at mid-tide. 'It was doubly risky,' he says 'because we could only bring men and materials in from the north.'
Proximity of houses and flats on the Penarth head meant a severe restriction on southern access; only with special permission could machines or materials come in. 'And each time there had to be special permission,' adds Neal.
But the bund could be built at the same time as the rock core of the shore-linked northern section of the permanent embankment. This 'flank' then acted as a road for incoming trucks from the north. A temporary access bridge, 'our lifeline' says Neal, was built across the remaining 260m wide 'closure' channel. Made up from 30m span Mabey units, it had a 'bascule' lifting section manned night and day, to give boats continued sea access during the works.
The temporary bund had other advantages. Severe noise restrictions were imposed at Penarth by the Vale of Glamorgan Council - 'just where all the main structures are' recalls Neal ruefully. That would have severely constrained pile driving.
By using a bund, around 70% of the approximately 1,000 piles required for the project were cut out, says Rick Randall of Costain, deputy project director. Some of the remaining permanent sheet piles on the locks could be 'assembled' along with concreting, toed into a foundation beam and not driven.
'Day time noise limits were a 75dBa cumulative limit,' says Neal 'which effectively means around 1 hour 30 minutes of loud work, sometimes less. To complete all the piling required would have taken years.' Night time piling work was impossible, with a 65dBa absolute limit imposed, no more than a loud conversation.
Even before creating the bund, noise limits were a severe restriction, recalls Neal, who has been with the project from the first day. They meant careful programming for an extensive 18 month dredging operation which preceded construction, to make sure barge-mounted backhoe dredgers were far enough away from residences.
'You have to work with the tide,' he says 'which means at all hours.' Noise travels far over water at night. 'You could not even drop a spanner,' he remembers, adding that a computer taping system was run to distinguish working clatter from sea-gull screeches, which could trigger noise monitors. Its tape loop stored the last 20 seconds when a loud noise registered.
The backhoes diverted the bay's two river channels to direct water away from works; they more importantly formed a foundation trench for the barrage, which sits on firmer ground beneath the oozy mud banks.
Geology is relatively straightforward says Randall. The whole site is covered by alluvium up to 12m thick and then there is a varying layer of sand, gravel and cobbles between 1m and 6m. Below that is 3m to 4m of glacial till overlying Mercia Mudstone.
The main embankment was to sit on the sand and gravels, which meant removing the mud; not so easy in soupy water and twice daily 14m high tide depth. Subcontractor Dredging International used a computer system with satellite fixes and data from real-time tide level monitors to give bucket position, and the operator had a 3D display image of where he had dug. Some 2M.m3 of alluvium was removed by the 8.5m3 bucket Zenne and the 12m3 Big Boss.
An added difficulty occurred with huge greenheart timber piles found near the Penarth Marina. The 14m piles were 'very tough' according to Eric Westwood, the current project manager for Bechtel, which has had both project engineer and supervising engineer roles on the scheme. The timber formed a kind of channel protection wall in Victorian times, long since silted in. Delicate work with the buckets was needed to dig around these and eventually pull them out, 'costing some unexpected weeks' he allows.
The 18.5m reach of the backhoes proved crucial since the gravel layer in many areas was up to 3m deeper than expected. Facing an increased excavation BB-C proposed a design change to the client. Under the contract any 'value engineering' gains could be split 50:50 so a proposal to limit rockfill volumes by founding the barrage core higher up on a sand bed was a gain all round, even though some redesign to the trench was needed and careful grading of the sand, won from deposits in the Bristol deeps.
Sand was brought in 3,500t loads to form the trench base by a trailer suction dredger, the Vlaandren-xx 'the biggest split hopper unit in the world', says Neal. This then needed compacting for which the contractor devised its own vibrocompaction unit based around a ICE 815 vibration pilehammer. Sheet piling used a BSP 9t hydraulic drop hammer.
The operation identified a layer of entrapped silt 300mm to as much a 2m thick caused complications and some friction with Bechtel. Eventually it was agreed that there was time enough in the first section of the barrage, the 'flank' projecting from the Alexandra Docks, for consolidation to give the silt enough strength.
'In the closure section however, where the permanent embankment would be built up rapidly there was a sudden danger of slip,' says Westwood. A rock toe was added on the seaward side to give additional stability. This 'engineering solution' to the problem was something Bechtel favours he adds. The contractual issues are not fully resolved.
The suction dredger was also used to bring in sand for the temporary bund. This was also founded down to the sand and gravel layer, though some of the less important bayside toe areas were over silt.
The temporary bund was shaped with bulldozers and excavators, the Caterpillar D6s brought in through Penarth under escort and trucks and dumpers ferried from the north using a small landing craft - 'ideal where there is no jetty' says Neal. Barges also brought excavators and rock grabs.
Rather like 'walking up a down escalator' there was a constant fight to shape, re-shape and armour protect the sand heap between tides, only to see it partially wash down again.
Equipment, including Komatsu, Hitachi and Caterpillar, parked up at high tide on the sand-filled sheet pile double skin cofferdam that forms the permanent Penarth abutment and served as the end of the bund. As with a sand 'island' formed for the control area, the piles were tied with two layers of tie rods, up to 96mm in diameter.
Another temporary double skin cofferdam ran 80m on the inner side of the bund, because a sprawling embankment at that point would have blocked the entrance to the local Penarth Marina. The two cofferdams allowed for a scour protected gate entrance between them.
A fleet of Volvo articulated dump trucks brought local Welsh quarried limestone to armour the face and form a toe. 'This was particularly important on the seaward face,' explains Nick Platt the chief engineer for the JV, 'which was much steeper, to allow space for work on the breakwaters'.
Giant caisson units had to be floated in on the exposed sea face to form the yacht harbour (see page XIII).
Stability of the bund, a 20m high sand wall, had been subject to a major Hazops brainstorming session over two days, says Neal. 'There would be 300 workers inside, as much as 20m down,' says Randall, adding that watching a big ship pass on a high tide was an awesome experience. Designer, client, sand specialists and contractor sat down with risk specialists from the AEA among others to hammer through all the risks.
One conclusion was to go with a cut-off wall in the bund, although strictly speaking the limited permeability of the sand meant that it could survive with just a dewatering wellpoint ring inside. Plans for the cutoff to be sheet piled would have had noise disadvantages, so a 'vibwall' was installed.
'A vibration rig with a large leader guide drove a big I section down through to the glacial till,' says Randall. A thin cement bentonite slurry was injected as the section was withdrawn. 'The mix set with a putty like consistency and proved exceptionally good; two years later as the bund was removed sections of it remained standing vertically with no support' he says.
Instrumentation monitoring and trend analysis had shown the bund performing very well throughout its two year life despite having to cope with a twice daily tide cycle. Design of the structure was independently checked by Halcrow.
A cutter suction dredger came in once the concrete structures inside were completed (see page XIII) and the bund carefully re-flooded. Some 0.5M.m3 of sand was re-usable, going to form the inner side of the permanent embankment, resting against careful graded filter layers.
A 500m length of the flank permanent rock embankment had already been formed up to 7.3OD level early in the programme with the sand bund cofferdam. Rockfill was local stone up to 800mm in size. Much larger boulders formed an armour surface.
That was on the flank. Theclosure section remained. Until the sluices were complete and tidal flow could be funnelled through them, this could not be sealed. 'And the sluices had to have the carefully curved training walls complete to work,' says Neal, 'which meant all the breakwaters had to be finished.'
Closure of the final 260m long opening took place last autumn. It was an extraordinary operation, remembers Westwood, being 'one of the few dam constructions I know of done completely in the tidal flow'. There was no diversion though the sluices could take part of the flow.
The embankment was to be built up vertically, rather than working in from either end for a horizontal closure. 'That would have built up the flow rates to unsustainable levels as the gap narrowed' says Neal.
Gibb's design studies, physical modelling at Hydraulic Research in Wallingford, and mathematical modelling at Holland's Delft Laboratory, had suggested the embankment could be built up incrementally. As long as no more than a 1m difference existed along the embankment at any stage, there would be no dangerous channelling.
BB-C re-analysed the flows using maximum spring tide figures rather than average tide figures, and allowing for high fluvial input into the bay. To be certain, they decided to allow no difference in height at all, presenting a level surface to the tide as it came in and flowed out.
'There is bound to be erosion on the leading edge,' says Neal 'and that in both directions in a tide, but you can avoid any channelling'.
But forming a 1m lift across the whole embankment, especially at the lowest levels was impossible. At the lowest levels the tide allowed only a three hour window for work which had to cover a 30m width of the core.
'We laid a 5m band of rock along the seaward side,' explains Platt. Once a new horizon was presented to the tide the rest could be infilled in strips.
The operation took place last year from June to August, two months for the closure and another month to place the armour. Rockfill was 400mm to 800mm in size with armour up to 5t blocks, all stored in large stockpiles in the Cardiff Docks.
During the operation BB-C worked to military precision. A special closure committee met each morning to check off that all machines, men and materials were ready for the operation on the next tide and also to check out the latest weather forecast and river flow levels. Wind, rain and pressure surges could all influence water levels.
'It was,' recalls Neal 'potentially one of the most risky parts of the project.' But it went exceptionally well. Behind the barrage another 300,000m3 of sand had to be dropped from the estuary to complete the embankment.