Seismic isolation using conventional laminated elastomeric bearings 1

[bugbus]

I want to get a few different opinions on the following questions relating to the seismic response of bridges supported on laminated elastomeric bearings.

For a bit of context, new bridges built in Australia usually comprise prestressed concrete girders with a cast-in-place deck, and supported on conventional laminated elastomeric bearings. The girders are normally simply supported, including at the piers, although the deck slab will be continuous over the pier to avoid a joint (referred to as a link slab). The typical method of restraining the superstructure in the transverse direction is to provide restraint blocks with plastic bearing pads on the abutments and piers (to limit transverse movement and potentially dislodgement), and in the longitudinal direction by providing either buffer bearings between the ends of the girders and the curtain wall or simply by relying on the deck joint closing up at the abutments. The restraint blocks and bearing pads are typically installed with gaps anywhere between 10-50 mm to allow for some movement of the superstructure before they become engaged.

Typically, the laminated bearings are so flexible in shear that the superstructure will effectively oscillate as a single degree of freedom mass, with a relatively long period of vibration (sometimes as long as 2s). This is particularly the case for a single-span bridge with abutments that are firmly locked into the ground. This is the type of bridge I am mainly concerned with here.

My first question is this: how should we analyse the seismic response of such a structure given that there is the potential for 'pounding' between the superstructure and the restraints? The common analysis approach is to consider all the gaps closed, such that the superstructure and substructure are 'locked together' at locations of lateral and/or longitudinal restraints. This would seem to be a conservative approach, because the overall structural response will be much stiffer and thus attract greater earthquake loads (particularly for single span bridges as I mentioned above). But it also ignores the very real pounding behaviour between the superstructure and the restraints. Typically, the size of the gaps is less than the displacement demand of the superstructure, and so pounding would be expected in most cases. Is this pounding something to be concerned about? Could it generate greater earthquake loads compared to the idealised model where all the gaps are assumed to be closed? I'm not aware of any method of analysing such a situation, apart from a nonlinear transient analysis with compression-only effects. Clearly, that's probably overkill for most situations.

My second question is based on a different approach, which is simply to provide large enough gaps between the superstructure and the restraints so that the superstructure can freely displace during the design earthquake, and the displacement demand is taken up entirely in the bearings before the restraints can be engaged. (I should mention that, even if not required for earthquake effects, lateral restraints would normally be required to prevent dislodgement due to other effects such as collision.) For this approach, we would then be relying on the laminated bearings to effectively act as seismic isolators. The clear advantage of this approach is that the whole structure sees a lot less earthquake force. But, I would note that there are specific bearing types and other devices on the market that are intended to do precisely this, e.g. lead-cored elastomeric bearings, seismic dampers, etc., although these are rarely seen in Australia, being a relatively low-seismic country. So when would a specific seismic isolating device be required, such as LRBs? Can conventional laminated elastomeric bearings be relied on to fulfil the role of a seismic isolator?

Appreciate any insights in advance!!

 

[bridgebuster]

I work mainly in the downstate region on NY, which is NYC, Long Island, and the abutting counties north of NYC. Almost all new or rehabilitated bridges use elastomeric bearings with sole and masonry plates. Multi-rotational bearings are mostly limited to curved and highly skewed bridges (>30-degrees); sometimes they're used on very wide bridges.

For a single span bridge, we make the bridge seat wide enough so that one end doesn't fall off the seat and the other end doesn't crash into the backwall. For a contiguous bridge, i.e. simple spans with link slabs, it's essentially the same thing but the abutment bridge seats get wider because the span length, per AASHTO, is measured from expansion joint to expansion joints. The pier widths are also made wide enough to keep the stringers from falling off. When necessary, we add shear blocks, guide bars, or similar to restrain transverse movement. Longitudinal restrainers are avoided on new bridges; occasionally a rehabilitated bridge might have them; I only recall on bridge in the city with them. Lead core bearings and dampers would be limited to seismic retrofit of a major bridge.

 

[Lomarandil]

I can't imagine that pounding would generate larger accelerations (hence seismic loads) than the analysis with gaps closed. That seems like it would need to be the perfect storm of your structural response lining up with the seismic forcing function. If anything, the vast majority of the time the pounding should be dissipating energy -- definitely overkill to analyze.

I'd say that a conventional elastomeric bearing could have some effect as a seismic isolator when detailed as you propose... I don't know how the shear stiffness/isolation compares to a purpose-built device to know how effective it would be.

 

[BridgeSmith]

The elastomeric bearing provide the freedom of movement to accommodate ground movements within the limits of the displacements allowed by the joints, shear keys, etc., but they don't provide the damping effect of the lead core of an LRB.

 

[bugbus]

Thanks all, appreciate the responses.

As BridgeSmith mentioned, I think the main difference of the LRB is its much greater damping. If I'm not mistaken, the damping ratios are about 0.05 for the conventional bearings and 0.15~0.20 for the LRBs.

I suppose where I'm coming from is that the seismic displacement demand (as it's specified in the code) is based on an underlying damping of 0.05, so if the bridge could be detailed with wide enough gaps to accommodate that, I suppose the conventional bearings would be suitable.

So maybe the LRBs are more suitable in regions with much larger earthquakes where the additional damping would help to reduce the displacement demand.

Time to do some more reading I think...

 

[BridgeSmith]

the damping ratios are about 0.05 for the conventional bearings and 0.15~0.20 for the LRBs.

They can be higher if need be. It's all about limiting displacement vs. limiting force transferred to the substructures. If you want small movement, the forces will be larger.

The lead core is pretty amazing in itself. It deforms plastically, but recrystallizes to provide the same damping response for the next earthquake. It also creeps, so shears under 50% to 75% less force for slowly applied loads, such as thermal movements, so the Service load forces to the substructures are considerably lower than the Extreme Event forces.