|
ORDERED PEAK
STATISTICS THROUGH DIGITAL SIMULATION |
|
B. BASU 1,
V. K. GUPTA 1, D. KUNDU 2 |
|
1Department
of Civil Engineering, Indian Institute of Technology, Kanpur 208016, India |
|
Keywords |
random processes; seismic response; peak factors;
order statistics; peak dependence; simulated joint density
|
Abstract |
|
A detailed knowledge of the seismic response time history
in the form of the first, second, third, |
largest
peak amplitudes is useful to understand the progressive damage in a
structure. This information is however not available from the conventionally
used response spectra. This paper provides a formulation of the statistics of
these higher order peaks by proposing a digitally simulated joint density
function for the peaks in a stationary, Gaussian process. A digital
experimentation has been done by generating an ensemble of the
single-degree-of-freedom oscillator response to several artificial
accelerograms, and the corresponding ordered peak distributions have been
found to be in close agreement with the predictions based on the formulation
in this paper. The peak factors based on the proposed formulation have also
been compared with those from the existing formulations based on the Markov
theory and on the assumption of peak independence. It has been found that
whereas the peak factors from the proposed and Markovian formulations are
compatible, the assumption of independence may give reasonable estimates only
for the first few orders of peaks.