Bridgeweb
May 17, 2008
Email Updates
Northern exposure
Limited damage and a rapid rebuild is the design criteria for the seismic retrofit of San Francisco’s Richmond - San Rafael toll bridge
Published:  July 04, 2003

Strengthening work on the Richmond - San Rafael Bridge looks set to be the next seismic safety project on the starting blocks - bids for the estimated US$400 million scheme were due to be invited as Bd&e went to press.

The toll bridge is one of the lesser-known Bay crossings - and one of the least used, although still an essential link for North Bay commuters. It was opened in 1956 and carries two lanes of traffic in each direction on its double-deck structure. The main part of the bridge consists of constant depth 88m double deck steel trusses supported by steel towers with K-bracing, sitting on concrete substructures.

Two shipping channels are each spanned by 326m cantilever trusses, and the bridge deck shows a characteristic ‘sag’ between the two. The concrete substructures consist of two or four concrete shafts connected by concrete spandrel beams and diaphragm walls. These shafts are bell shaped and sit on a large number of steel H-piles which are placed in concentric rings. The outer rows of piles are battered at 1:4 and 1:6 and vary from 3.7m to 61m in length, reflecting the highly variable rock profile on this alignment.

Ground conditions also include a layer of sand which is subject to liquifaction during earthquakes and would amplify ground motions. At each end of the double deck bridge, there is a transfer structure - a steel plate girder bridge - which converts the traffic from its double-deck configuration to a side-by-side layout.

At the western side of the bridge the traffic then crosses a concrete trestle bridge which will be the subject of an extremely complex rebuilding project as part of the retrofitting scheme (see below).

Seismic events originating at both the Hayward and the San Andreas faults had to be considered in the design of the retrofit measures explains John Vincent, project manager for consultant Ben C Gerwick which is designing the substructure retrofit.

Gerwick is leading the joint venture consultancy group which also includes Sverdrup and DMJM. Although the bridge does not have to be usable directly after an earthquake, it must be designed so that damage is limited and can be repaired within a matter of months. The duration of the design earthquake is 40 seconds, and has been established based on experience from recent events, including Kobe in Japan, and several local earthquakes in California.

Use of these criteria highlighted a number of deficiencies in the existing structure; the existing H-piles do not have sufficient axial and flexural capacity to withstand the predicted horizontal displacements - and this is aggravated by the fact that the piles have also suffered from corrosion and loss of section over the years.

"Because they are H-piles, they are not compact and buckle more easily," says Vincent. "With the retrofit measures there will still be damage to the piles, but it will be repairable and will not threaten the stability of the bridge." Large shear forces would be imposed on the concrete substructures, as well as large uplift forces from the steel towers which are up to 44m high. The concrete shafts, which are up to 4.3m in diameter, are only lightly reinforced and the predicted ground motions would cause a shear failure in the concrete shafts and upper spandrel beam, in addition to ‘unzipping’ the lower diaphragm wall.

Controlling the behaviour of the steel towers, meanwhile, was seen as central to ensuring the seismic stability of the structure. Under earthquake motions, the designers predicted that the existing bracing would fail and the towers would buckle, causing the whole structure to collapse. "The inverted V-bracing which has been used on the existing steel towers is very poor in terms of seismic resistance," confirms Vincent.

Strengthening of the foundations involves the installation of large diameter piles and a new ‘pile cap’ element at each pier. The pile cap consists of two precast concrete units which fit between the two existing shafts and are linked to form a ‘dog bone’ shape. The pile cap will be installed first, each half supported by three 355mm diameter steel pipe piles and linked together with tremie concrete.

This pile cap contains two oversized holes which act as a template for driving the large diameter piles. Once the piles have been driven, the annulus will be grouted. Only horizontal loads will be transferred between the new piles and the existing foundations. This produces three distinct advantages; it allows contractors to avoid the need to carry out complicated connections some 18m below sea level, and means that the thickness and mass of the pile cap are reduced.

It also means that additional overturning demand on the H-piles is avoided. Semi-circular steel casings 1.2m deep will be placed around each of the existing foundation bells and connected into the precast pile cap by high strength rods. These will confine the bell piers and prevent the piles from breaking out during an earthquake. Elastomeric pads between the concrete slab and the bell piers are designed to accommodate any shear movement.

The large diameter piles will only carry their self-weight and any seismically induced lateral loads. The second part of the foundation retrofit is to put reinforced, vertically prestressed precast concrete jackets around the columns to confine them and prevent the diaphragm and spandrel from ‘unzipping’ during seismic movement.

Concrete was chosen for this application because it was possible to design a durable mix which could survive for the predicted 100 year residual life of the bridge. "The mix has a low water/cement ratio and contains fly ash and silica fume, and we will have a polyurea coating on the concrete" says Vincent.

By the use of all these measures, he explains, the designers have been able to reduce the reinforcement cover from 125mm to 63mm, saving weight on the jackets. An erection frame will be installed at the top of the concrete columns and the match-cast segments which form the jackets will be linked around the columns, then lowered down to allow the next segments to be added, until the full height is complete.

All of this work can be carried out without the need for divers, although the final installation of HS rods does require some underwater work. Strengthening of the steel towers involves what Vincent believes to be the first use on a bridge superstructure of eccentrically braced frame towers. The existing towers, which range in height from 16.5m to 41m, are very vulnerable to collapse by failure of the inverted V-bracing. Eccentrically braced frame towers are more flexible and act like a moment frame - the link beams are designed to fail in shear, not flexure, and to provide the ‘fusing’ element during strong earthquake forces. The design matches the architecture of the existing structure and the extra weight that would be the result of using concrete can be avoided.

Two frames will be installed, one on each side of the existing legs, and then the V-bracing will be removed. Full-scale tests on these frames have been carried out - although such designs have been used for several years in building frames, the ones on the Richmond-San Rafael Bridge will be much bigger than those tested so far. The interaction between the existing legs and new frames was a significant feature of the retrofit design; the legs must continue to carry gravity loads transmitted through the truss shoes at the top of each leg.

Cells at the base of each leg with be filled with concrete to prevent local buckling, however this brings problems in itself as it may stiffen the legs and cause yielding above the concrete infill. To avoid this, the concrete will be placed in layers separated by a compressible joint material. Superstructure retrofit involves an extensive amount of bracing to the existing double-deck truss - replacement of existing elements, addition of extra units in new locations and strengthening of existing bracing with additional plates.

Upper and lower decks will be reconstructed to create a seismic isolation joint between the east and west approaches and the main structure, to accommodate large relative movements. Here friction dampers and isolation bearing units will be installed on a longitudinally sliding surface and a steel deck plate will be used to span the enlarged opening. On piers without steel towers, isolation bearings will be installed, and deck expansion joints will be retrofitted to minimise damage to main structural members and limit the joint openings.

Trestle trouble

Retrofitting the existing trestle part of the bridge will provide a construction challenge for whichever contractor wins the job - it is to be completely reconstructed, on the same alignment, with only night time closures of one side of the bridge allowed.

Gerwick’s design involves replacing the existing 15m spans by 30m precast units of double-T beams - this solution will make it possible to avoid the existing foundations. The design of the new bridge is fairly standard, it is the construction which will need to be planned and executed with military precision to make sure that traffic is not disrupted.

Heavy penalties will be slapped on the contractor if it overruns the allowable lane closure period. The proposed construction sequence is as follows.

A large rig will be used to set the piles into position, then a small rig can be brought in later to complete the pile installation. The inner girders will be removed, and the inner piles installed, then the same with the outer girders and piles. The pile cap beam will then be either floated into position under the existing deck or slid in using skid beams. The existing piles are removed down to the mud line and the deck replaced. Caltrans will face severe consequences if the bridge is not open as normal in the morning, so whichever contractor wins the bid will have to carry out trial runs beforehand to prove that they can do it within the allotted time.