Figure 1: River Torridge Bridge

There were no glued (or epoxied) segmental bridges in the UK when the design for the Torridge Bridge was started in the 1970s. This was not the first glued segmental bridge to be built in the UK, however, due to delays with changing design standards, public consultation and public inquiry. It was still something of a pioneering project, and the three 90-m (295-foot) river spans were the longest yet constructed by this method in the UK (Figure 1).

The bridge was designed by MRM partnership — now the UK arm of PB. The main contractor was Edmund Nuttall and the launching girder was designed by Tony Gee and Partners.

The River Torridge starts among the high tors of Dartmoor and meanders quietly northwards through the Devonshire countryside until it reaches the historic port of Bideford where, quite suddenly, it is transformed into a major tidal estuary. The old multi-span stone arch bridge at Bideford is one of the very few medieval bridges in the country, but as 700 years of scour and 20th century traffic were taking their toll, a decision had been made to construct a new bridge one mile downstream of the existing crossing where the estuary was 300-m (984-feet) wide and flanked on both sides by land which flooded at spring tides.

Factors Affecting the Choice of the Design

The town of Bideford owes much of its character to the shipping that visits it, so it was essential that the new bridge not adversely affect the somewhat fragile commercial viability of the port. For this reason, a high level bridge was essential. Aesthetics were at a premium because such a bridge would dominate the view from the town, especially as the estuary was a site of great natural beauty. The other main factors that affected the choice of the design were the:

• Scour in the river bed
• 7.5-m (25-foot) tidal range
• Rock profile under the river
• Danger from collision with shipping
• Height of the approach embankments
• Railway and the roads which and relative position of the river walls the bridge had to cross.

Figure 2: South Elevation

After considering the usual range of options — from cable-stayed spans to tunnels and a tidal barrage consisting of a dam across the estuary with lock gates and a lifting or swing bridge to accommodate shipping — it was decided that the most appropriate and elegant solution would be a 650-m (2,133-foot) -long, eight-span bridge that crossed the valley in a single entity of roughly equal, variable depth spans — the three river spans each being 90-m (295-feet) long and the approach spans reducing only slightly in length towards the abutments (Figure 2).

Deck Design

It was clear from the outset that there were special problems involved in building a series of large spans high above a tidal estuary that dried out at low tide and had to be kept open to commercial shipping at high tide. A balanced cantilevered construction offered the most obvious solution, and the glued segmental precast technique that had been developed in Europe appeared to be ideally suited for the situation at Bideford.

Figure 3: Deck Sections

The bridge was required to carry a single 9.3-m (30-foot) wide, two-lane carriageway with 1.5-m (5-foot) -wide footpaths. The loading was to be the British HA loading, which is a standard UK truck load, together with the 180 tonne (198 ton) HB vehicle, which is a UK permit load. After some preliminary work, it was found that the best solution for the superstructure was a 6-m (20-foot) wide single-cell spine beam varying in depth from 5.65 m (18.5 feet) at the piers to 3 m (9.8 feet) at mid-span. The cantilever wings were each 3.65 m (12 feet). Design details are illustrated in Figure 3.

A 19/15 “compact” strand system was adopted for the permanent prestressing because, at the time, this was considered to be the largest tendon that could be stressed conveniently in a structure of this nature. The tendons were generally tensioned to 4330-kN (973-kp) force, which is 76 percent of their ultimate tensile strength. For simplicity, it was decided to have only one pair of anchors at the top and/or bottom of each segment. This led to an optimum segment length of 2.575 m (8.45 feet). The segments varied in weight from 56 metric tonnes to 80 metric tonnes (62 tons to 88 tons) except for the segment on the pier (the “pierhead” segment or “pier table”), which weighed 105 metric tonnes (116 tons).

The primary moments and shears were obtained from grillage and continuous beam theory, taking stage-by-stage construction into account. Additional effects of warping, torsion, distortion and creep redistribution were assessed initially using hand methods. These were subsequently verified and adjusted using an early and somewhat primitive finite element program.

Shear keys on the faces of the segments were designed primarily to carry the newly erected segment before the epoxy glue had hardened. They consist of a series of 400-mm (16-inch) serrations projecting 100-mm (4-inches) from the face of the web. The serrations are concealed on the outer face by a masking nib so that the joint appears as a clean vertical line on the finished structure.

Figure 4: Substructure

Substructures

The deck is supported on two spill-through abutments and seven vertical columns. The columns are coffin-shaped in cross section, tapering from 2.5-m (8.2-feet) wide on the centre line to 1.7-m (5.6-feet) wide at the edge (Figure 4). This shape has both structural and aesthetic advantages. The narrow leading edge catches the eye and gives the column a slender appearance and the inclined facets add interest by reflecting different intensities of light and shadow.

Foundations

All foundations were on sandstone or mudstone. At the two abutments and piers 1, 2 and 7, the rock was sufficiently close to the surface for conventional spread footings to be used. For piers 5 and 6 (which were protected from shipping impact by the river wall) large 2.1-m (67-inch) diameter bored piles (drilled shafts) could be used.

In the middle of the bridge, however, the rock level dropped to some 16 m (52.5 feet) below bed level, so the situation for piers 3 and 4 was radically different. These piers were vulnerable to shipping impact, and a study of the alluvial deposits indicated that 10-m (33-foot) deep scour holes might develop during spring tides. It was decided, therefore, to found the two main river piers on 18.5 m by 12.6 m by 16 m high (61 feet by 41 feet by 52 feet) compressed air caissons that would be sunk to rock level where they could resist shipping impact without support from any fill around them and the foundation stratum could be inspected in the dry.

The bottom 3 m (10 feet) of the caissons were cast in Appledore dry dock, made buoyant with compressed air, towed up the river on spring tides and positioned on the sand of the river bed as the tide ebbed. The contractor elected to construct the next 9 m (30 feet) of caisson at low tide without any excavation. During this stage, scouring of the river bed caused the caissons to list alternately to the east and west on the ebb and flow tides, but any out-of-plumb was subsequently corrected by differential digging through cells that had been provided in the design for this purpose, and the caissons ended up vertical and correctly positioned. When mechanical excavation through the cells was no longer possible, the caisson roof was cast and sinking was continued to rock level by hand, digging under compressed air.

Concrete fenders were cast on top of the caissons to protect the more vulnerable columns from shipping impact. Rubber units were used to ward off glancing blows, but when designing for head-on impact it was found that the energy of such a blow would mainly be absorbed by the collapse of the impacting vessel. The impact force used for the design was therefore assessed by investigating the structure of typical vessels that visited Bideford, using Gerrards Empirical Design Method, which was developed to determine the crippling strength of the frames of aircraft.

Precast Segment Construction

The deck segments were match-cast in a specially constructed casting shed on site. Only one casting bed was provided, although the complicated “pier table” units were cast in a separate form so that there was no delay to the match-casting operation. The bridge required 251 segments, and a rate of one segment per day, had to be achieved ultimately if the bridge was to be completed on program. The original concrete mix contained Portland blast furnace cement to control alkali-silica reaction, but difficulty was found in achieving adequate early strength with this mix and in the end the contractor decided to blend the blast furnace cement with low-alkali ordinary Portland cement.

Stringent geometry control was applied to the match-casting operation and the form for each segment was set taking into account the future prestress, dead load and creep deflections and the accumulated errors that had developed in casting the previous segments in the cantilever.

Figure 5: Figure 5: Photo showing the first 105 metric tonne pier segment about to be placed. The launching girder is supported on temporary steel towers. Once the segment was stressed to the column, the girder was launched forward on rollers located on top of the pier segment.

Figure 5 shows the first 105 metric tonne (116-ton) pier segment about to be placed. At this stage, the launching girder was supported on temporary steel towers, but once the segment was stressed to the column, the girder was launched forward on rollers located on top of the pier segment.

Deck Erection

The method of erection assumed in the design was that the “pier table” would be stressed vertically to the column so that it could resist the eccentric effects of two segments being erected on either side of it. Thereafter, it was assumed that outriggers on a launching girder would be used to counteract any out of balance moments that arose during erection.

This is what, in essence, was done, except that the contractor choose to use an innovative launching girder weighing only about 150 metric tonnes (165 tons). This girder was not only significantly lighter (and cheaper) than that assumed in the design, it was also much shorter, and every stage of the erection had to be carefully re-checked, particularly the uplift effects on the last completed span. These uplift effects often had to be counteracted by the strategic placing of other precast units on the “back span” to act as a counter-balance.

The units were delivered to the launching girder by a low-loader reversing over the completed section of deck and, at one critical stage, it was even necessary to instruct the low-loader to park overnight at a specified location on the bridge to counteract the adverse differential temperature stresses which would develop at night. The uplift problems were compounded by the fact that the program did not allow for any delay while the previous in-situ mid-span closure pour reached strength, and a temporary steel hinge was used to connect the ends of the penultimate two cantilevers.

Figure 6: Deck Erection

Dywidag post-tensioning bars were used as temporary stressing to allow two pairs of precast units to be erected ahead of the permanent prestressing. These were left stressed in the completed cantilever to help control the tensions across the joints when the next balanced cantilever was under construction.

The deck was erected in eight months. Figure 6 shows construction in progress. Note the low-tide dry riverbed.

Accolades for the Completed Bridge

The bridge was formally opened in May 1987 and received many accolades. It won the “Concrete Society Award — 1988” together with a commendation from the Institute of Structural Engineers, and it was short-listed for the prestigious British Construction Industry Award the same year. Perhaps, though, the most satisfying accolade came from some of those who had vociferously opposed the scheme in the early days, who said (after the bridge was completed): “If we’d known what we were going to get, we would never have objected to the scheme in the first place”!