Page 194 - Dynamic Loading and Design of Structures

Page 169

exceedance of their shear strain capacity, or buckling of the bearings (at low strains). Special
restrainers (such as steel angles) can be provided close to the bearings to prevent them from
toppling over.
A critical point in passive control systems is that whereas isolator damping is always
reducing the displacements of the structure that are controlled by the fundamental mode, it
tends to increase floor accelerations caused by higher modes. This might be very important in
structures where protection of secondary systems (equipment and non-structural elements) is
the main reason for using seismic isolation. Seismic attack on secondary systems is frequency
selective and it is possible to design isolation systems that reduce the response of such
systems more than that of the primary structural system. A related issue is that in non-linear
isolation systems (which are used in the majority of applications), control of the amount of
base shear through the strength and the stiffness of the isolators does not guarantee control of
the storey shear distribution along the height of the building. Whenever higher mode response
is not adequately controlled, ‘bulged’ distributions of storey shear can result and in extreme
cases the shear in the upper half of the structure may exceed the base shear (Skinner et al.,
1993). The foregoing are clear indications of the need for a reliable dynamic analysis when
dealing with isolated structures.
The first design guidelines for seismic isolation were issued in California in 1986, and have
been subject to several revisions; they were incorporated first (1991) as an appendix and later
(since 1994) as a formal part of the UBC. A critical review of code provisions for seismic
isolation can be found in Naeim and Kelly (1999). The current versions of UBC (ICBO, 1997)
and NEHRP (FEMA, 1997a) contain provisions that are essentially identical, with the
exception of the definition of design earthquake (see Section 4.3.2). These provisions include
both the equivalent lateral force and the dynamic analysis procedures for seismically isolated
buildings, but the restrictions for the former are such that in most practical cases the dynamic
approach has to be applied. Two sets of verifications are required: The first one is for the
design earthquake (10 per cent/50 year probability), under which the structure is required to
remain essentially elastic. The second one is a stronger event (10 per cent/250 year
probability) for which the isolation system should be designed and tested, while all building
separations and utilities that cross the isolation interface should be designed to accommodate
the forces and displacements associated with this seismic input. Whereas simplified methods
based on the equivalent SDOF are available (see, among others, Skinner et al., 1993) and can
efficiently be used for preliminary design, most seismically isolated structures are currently
designed using time history analysis. In the current Eurocode package, provisions for seismic
isolation are only included in the bridge part EC82 (CEN, 1994c). However, currently (2000)
such provisions are being developed for buildings and will be incorporated in the final (EN)
version of EC8.
The main reason why isolation is not widely used today (particularly in buildings) is the
concern regarding initial cost of the project (i.e. that in most cases a seismically isolated
building costs 1 per cent to 5 per cent more than the