London Bridges: 32. Hammersmith Flyover

Here's more concrete, this time going back to roughly a decade before London Bridge.

Designed by Maunsell and Partners, Hammersmith Flyover was opened in 1961. It was intended to relieve congestion on London's Great West Road, by elevating the through traffic and isolating it from local traffic junctions. The structure rests on relatively narrow central columns, minimising the impact on the road layout below.

It has sixteen spans for a total length of 622m, and is 18.6m wide. The deck is integral with the piers and consists of a single multi-cellular variable-depth post-tensioned concrete box girder. This was built from precast box segments each 2.59m long, alternating with 0.3m long precast "ribs", which cantilever beyond the box to carry the full width of the deck slab. The precast elements are stitched together with 75mm thick in-situ concrete joints. The whole system is explained in a diagram taken from the technical paper on the bridge design (see later in this post).

There is only a single expansion joint, near the middle of the bridge, pictured here. The piers sit on roller bearings below ground level, allowing for thermal expansion.

Hammersmith Flyover was certainly one of the earliest bridges to be built as precast, segmental and prestressed: Troyano's Bridge Engineering: A Global Perspective credits the first as being Nagatinsky Bridge, in Moscow, also completed in 1961 (although Structurae dates that bridge to 1966), but his generally excellent book omits the Hammersmith Flyover entirely. The Concrete Bridge Development Group claims, incorrectly, that the technology was first applied in France in 1962.

This is essentially the same form of construction that would be used again in London Bridge some ten years later. Indeed, segmental post-tensioned concrete construction has changed very little since it was introduced - the major difference if one of these bridges were built today would be the omission of the in-situ concrete stitches.

It's interesting to compare Hammersmith Flyover to a much more widely-lauded urban viaduct of the time, Pier Luigi Nervi's Corso di Francia Viaduct, completed in Rome just one year before. The Nervi bridge boasts more aesthetically shaped piers but is otherwise far less ambitious: it uses only simply supported precast beams, of much shorter span, and thus requires two rows of T-shaped piers. It lacks the smooth lines of the Hammersmith bridge, and its footprint at ground level is much greater. Nervi never made much use of prestressed concrete and was, it has to be said, never a great designer of bridges.

Many of the choices made in the bridge design have a positive aesthetic impact, but were made for other reasons. One of the key features which distinguishes the bridge from other structures of a similar type is the presence of the cantilever ribs. There's a section in Fritz Leonhardt's famous book Bridges which discusses alternative options for concrete viaducts (and, again, which omits any mention of the Hammersmith Flyover). This shows a variety of systems, almost all united by the smooth uniformity of their surfaces - they are largely texture-free. The Hammersmith ribs not only break up what would otherwise be massively monotonous concrete surfaces, they also create a visual rhythm, and make the bridge more comprehensible seen in perspective. However, their primary purpose was not aesthetic, but was simply to facilitate rapid erection of the deck slab while remaining within a construction footprint constrained to the width of the spine beam.

The curved bridge soffits also add to the sense of rhythm, and are more attractive than the constant-depth girder which would surely have been more economic. Again, aesthetics were not the primary driver in choosing this geometry - the curvature provides the necessary depth and strength over the piers, while maximising highway headroom over the roads which pass under the middle of each span.

The box girder webs are also curved, when seen in section. They are a constant width at the top, but the intersection of the web curvature with the soffit is such that the underside of the girder varies in width, being widest at midspan and narrower at the piers, as can be seen in the very first photograph above.

As with London Bridge, the papers in the ICE Proceedings are well worth seeking out and reading, they are far better than what tends to be published today, with copious diagrams and photographs. The assembly diagram shown below is one highlight.

The bridge was built for a cost of £1m (roughly £18m at today's prices). As well as being structurally innovative, the bridge incorporated roadway heating cables, intended to avoid the need for application of road salts. These ceased to work a long time ago, and presumably road salts were applied to clear ice ever since.

One contributor to the Discussion in the ICE Proceedings stated that "It was of great interest that the structural members at Hammersmith should not cost a halfpenny to maintain over the next hundred years, which was more than could be said for any similar structure erected on the site in steel". Clearly the contributor lacked any gift for prophecy, as it was announced in December 2011 that "serious structural defects" were to force the closure of the bridge.

It was reported that significant corrosion had been found in the prestressing tendons, in the vicinity of the piers. The prestressing layout for the bridge is both unusual and simplistic: the structure was built span-by-span, rather than by cantilevering from the piers, so there is significantly less prestressing at the top surface over the piers than might be expected in a continuous viaduct of this sort. Prestressing tendons are anchored to one side of the pier in the upper surface of the top slab, passing in ducts through the slab. They continue horizontally below the slab until they pass through the pier, and are then deflected downwards, adjacent to the webs. At midspan, they pass through ducts in the bottom slab, before being deflected back upwards towards the next pier in line. Where they are external to the precast concrete, they are protected by a concrete surround cast at a later stage. The diagram below shows the typical layout.

I haven't seen a detailed description of the areas which are most seriously corroded, but the presence of anchor points on the upper surface, and the ducting through the top slab, must both have been highly vulnerable to the ingress of water and road salts.

The solution proposed by TfL's consultant Amey was to add additional prestressing into the top slab, by providing new tendons symmetrically above and below the slab. These are capable of being restressed in the future to carry more of the load if and when the original prestressing deteriorates further. New concrete barriers were installed in the central reservation of the highway to protect these strengthening cables. A first phase of strengthening work was completed by Freyssinet for the five most seriously affected spans in 2012, with an estimated cost of £10m reported. In June this year, a further £60m contract was announced, to tackle the remaining spans, address underlying problems such as rewaterproofing and surfacing the deck, refurbishing the drainage, and other works such as bearing renewal. Hammersmith and Fulham Council were reported as hoping no further money would be lavished on "this monstrosity", favouring instead development of a longer-term tunnel scheme to replace the viaduct entirely.

A tunnel had in fact been considered before the viaduct was conceived, back in the 1950s, but was rejected due to the presence below ground of the London Underground railway, a large sewer, and a high water table, felt at the time to make the cost of tunnelling prohibitive. It's difficult to see how that will have changed, now with the added complication of building a tunnel directly below a massive flyover. One proposal is for a 3.7km long deep bore tunnel (well below the level of the aforementioned obstructions). The longest road tunnel in Britain at present is the 3.2km Queensway Tunnel in Liverpool, built in 1934 at a cost which would be roughly equivalent to £500m today.

Further information:

• Google maps / Bing maps
• Wikipedia
• Engineering Timelines
• The Hammersmith Flyover (Rawlinson and Stott, ICE Proceedings, 1962)
• The Hammersmith Flyover: Site measurements of prestressing losses and temperature movement (Wroth, ICE Proceedings, 1962)
• The Hammersmith Flyover: Discussion (ICE Proceedings, 1964)
• Modern British Bridges (Henry and Jerome, 1965)
• Civil Engineering Heritage: London and the Thames Valley (Smith, ed., 2001)
• An Encyclopaedia of Britain's Bridges (McFetrich, 2010)