Speaker
Description
This paper presents results of a site-specific seismic hazard study for a wind farm in South Iceland. The site was characterized using ambient vibration tests, resonant column and down-hole testing. The stratigraphy consists of sand with embedded lava layers, which is commonly found in Iceland due to the extensive volcanism. The effect of local geology on expected ground motion is included by means of a period-dependent soil (de)amplification function, AF(T), where T is a generic oscillator period. AF(T) is defined as the ratio of the spectral acceleration including non-linear site-effects to the spectral acceleration at the bedrock. The estimates of the statistics of the amplification function are obtained by linear-equivalent analyses of the soil column with uncertain properties (Bazzurro and Cornell, 2004a). The probabilistic seismic hazard assessment (PSHA) is based on Monte Carlo simulation of synthetic earthquake catalogues. This approach allows for a straightforward implementation of the soil effects in calculating hazard at the site. Epistemic uncertainties are also easily handled, without having to rely on logic trees, by switching over different candidate models of ground motion prediction through a roulette sampling approach. Near-fault effects, which are important for long-period structures such as the ones considered here, are incorporated deterministically from the results of physics-based numeral simulations of wave propagation of the most relevant earthquake scenarios from the deaggregation of the seismic hazard. Design ground motions and response spectra are therefore defined from a combination of probabilistic hazard assessment and simulation of potential scenario events incorporating aleatory uncertainties in key parameters defining the seismic rupture.
