Integration of Fault Rupture Models in Seismic Hazard Analysis: A Perspective from the Consultancy Side

Seismic hazard analysis (SHA) is an essential element in the site characterisation of a nuclear installation. A SHA comprises two major tasks: Seismic Source Characterisation (SSC) and Ground Motion Characterisation (GMC). Their respective outputs constitute the input for the hazard analysis, which includes obtaining ground motions at a reference rock, calculating the site response, and deriving compatible time histories. The paper identifies, from the perspective of an engineering consultant, the aspects of a SHA that could benefit from the use of fault rupture models, analysing the advantages, identifying their costs, and bringing up the points that need to be analysed to establish their relevance. Traditionally the GMC output has been a set of Ground Motion Prediction Equations (GMPEs). There are already GMPEs whose database has been completed with ground motions derived from simulations, which is a first (indirect) presence of fault rupture modelling in SHAs. The key question is whether generic GMPEs could be replaced by ad hoc fault rupture models in SHAs for the nuclear industry. Some ideas and even practical applications can already be found in the technical literature and in technical documents issued by official organisations like the IAEA. The site response analysis and the generation of time histories, both usually uncoupled from the rest of the analyses, would probably be more easily integrated in the overall process when employing fault rupture models. All these benefits would entail costs. Firstly, there is an obvious increase in the computational cost; second, the SHA team needs to be more multidisciplinary, enlisting specialists in fault rupture modelling, although this may be more an investment than a cost. A SHA must take uncertainties into account. When deciding about the relevance of fault rupture models, the previously detected pros and cons have to be weighted together with the level of uncertainty of the input parameters and the impact on the reliability of the results. (author)