Estimation of extreme value distribution of crosswind response of wind-excited flexible structures based on extrapolation of crossing rate

• An approach based on crossing rate analysis is proposed.
• It is for estimating extreme statistics of narrow band non-Gaussian crosswind response.
• Its accuracy is confirmed by simulated response and full-scale vibration data.
• It is very effective when the number of response time histories is very limited.

This paper presents an approach based on response crossing rate analysis for estimating extreme value distribution of crosswind response of wind-excited structures with significant nonlinear aeroelastic effect. The crossing rates at various thresholds are calculated from response time histories, and are then curve-fitted by a prescribed parametric model. The influence of narrow band characteristic of response is accounted by using envelope process with two-state description of crossings and a further consideration of mean clump size. The curve-fitting and extrapolation of crossing rate permit estimation of extreme value distribution using Poisson distribution of crossings. The effectiveness and accuracy of the approach are examined using simulated crosswind responses covering a wide range of non-Gaussian characteristics, and also using full-scale vibration measurement data of a wind-excited traffic-signal-support structure. The results illustrated that the approach can produce robust estimations of extreme value distributions of hardening non-Gaussian crosswind responses. The narrow band characteristic of response process has very limited effect on the extremes of hardening non-Gaussian responses. The approach presented is especially effective in practice, where the number of available response time histories is often very limited, and a direct use of extreme samples fails to provide accurate estimation of extreme statistics.