Parameter investigation of relaxed uncertain power spectra for stochastic dynamic systems

authored by

Marco Behrendt, Marius Bittner, Liam Comerford, Matteo Broggi, Michael Beer

Abstract

In structural dynamics the considerations of statistical uncertainties are imperative to ensure a realistic modelling of loading and material parameter setup. It is well known that any deterministic analysis only constitutes a narrow result for the given input parameters. Because of aleatoric or epistemic uncertainties, many factors must be considered either in certain interval margins or with subjective probabilities. Especially in the case of seismic ground motion, due to significant uncertainties, a reliable prediction of future event characteristics is important in designing safe structures. This work attends to the well known statistical procedure of simulating time histories of a mechanical model under an artificially generated earthquake loading, which is modelled by a stochastic process. The stochastic processes in this work are throughout synthesized using the Spectral Representation Method (SRM). A key aspect of this procedure is the estimation of the Power Spectrum Density (PSD). The PSD determines dominant frequencies and their magnitude of influence on the stochastic process and in nature on the earthquake signals. There are numerous methods to estimate the PSD from source data, however, the amount of data available is seldom enough to do this accurately and reliably. To address this issue, the authors suggest that the PSD model itself could be defined as a random vector in the frequency domain, thereby encompassing a range of possible valid PSD models for a given data set. For a stationary process, the random dimension of said model would be dependent upon its frequency discretisation. On a simple mechanical system, artificially generated stochastic processes with the novel described underlying power spectra are analysed utilizing a Monte Carlo simulation. A reliability statement of the mechanical system in the form of a first-passage problem is defined to acquire a probability of failure which is used alongside first- and second-order moments of the system's output as reasonable benchmark values.

Organisation(s)

Institute for Risk and Reliability

External Organisation(s)

University of Liverpool
Tongji University

Type

Conference contribution

Volume

2

Pages

3803-3815

No. of pages

13

Publication date

2020

Publication status

Published

ASJC Scopus subject areas

Building and Construction, Architecture, Civil and Structural Engineering

Electronic version(s)

https://generalconferencefiles.s3-eu-west-1.amazonaws.com/eurodyn_2020_ebook_procedings_vol2.pdf (Access: Open)