Keywords: natural hazard risk assessment, response surface approximations, surrogate modelling, stochastic simulation, probabilistic sensitivity analysis.

The analysis of engineering systems against natural hazards requires the adoption of appropriate mathematical models for (i) the system itself, (ii) the natural hazard and (iii) the system performance. The efficiency of this task depends on how well the knowledge about the real system and its environment (representing future excitations) is incorporated into the respective mathematical models. In particular, it requires knowledge of the characteristics of future natural hazards over the entire life-cycle of operation, a feature that introduces a source of significant uncertainty. A probabilistic approach provides a consistent framework for addressing such uncertainties, using probability models to describe the relative likelihood of different properties of each natural hazard and of the system itself. In this setting, risk can be quantified as the stochastic performance, given by a probabilistic integral of some risk consequence measure over the established probability models.

For estimating this risk stochastic-simulation based techniques offer a high-accuracy solution that additionally creates no constraints on the complexity of the assumed numerical and probability models. However, they involve significant computational complexity and for this reason they are frequently combined with soft computing techniques [1-3]. This paper presents a review of such applications with emphasis on the implementation of response surface surrogate modelling and probabilistic sensitivity analysis techniques as well as on the efficient integration of these techniques within the simulation-based estimation framework. Response surface methodologies can dramatically improve the computational efficiency by approximating high-fidelity, computationally intensive models. Sampling-based, probabilistic sensitivity analysis techniques can be used to quantify the importance of the various risk factors in affecting the overall risk. This information can be exploited further to improve the risk assessment efficiency, through improvements in the implementation of importance sampling and/or moving least squares surrogate modelling.

The examples discussed extend to very diverse applications: hurricane risk prediction, seismic risk estimation, as well as evaluation of risk for offshore wind turbines under extreme environmental conditions. The discussions demonstrate the great potential benefits that the combination of soft computing and stochastic simulation provides for natural hazard risk assessment. This combination can facilitate a comprehensive and detailed characterization of risk, in terms of the models selected for the system and the uncertainty description in them, and at the same time provide a highly efficient estimation of this risk.

[1]

M. Papadrakakis, N.D. Lagaros, "Reliability-based structural optimization using neural networks and Monte Carlo simulation", Computer Methods in Applied Mechanics and Engineering, 191(32), 3491-3507, 2002.

[2]

A.A. Taflanidis, A.B. Kennedy, J.J. Westerink, M. Hope, S. Tanaka, J. Smith, K.F. Cheung, "A comprehensive approach for online fast hurricane-risk prediction", in "Proceedings of the 11th International Conference on Applications of Statistics and Probability in Civil Engineering", Zurich, Switzerland, August 1-4, 2011.

[3]

A.A. Taflanidis, E. Loukogeorgaki, D.A. Angelides, "Risk assessment and sensitivity analysis for offshore wind turbines", in "Proceedings of the 21st International Offshore (Ocean) and Polar Engineering Conference", Maui, Hawaii, June 19-24, 2011.

purchase the full-text of this chapter (price £20)

go to the previous chapter
go to the next chapter
return to the table of contents
return to the book description
purchase this book (price £85 +P&P)