Vector-valued Intensity Measures Incorporating Spectral Shape For Prediction of Structural Response

Jack W. Baker, C. Allin Cornell

Journal of Earthquake Engineering · 2008 · 123 citations · 22 references

Concepts

TL;DR

The second period in the intensity measure must be chosen intelligently to capture the most relevant spectral shape properties for effective response prediction. The study develops vector‑valued ground‑motion intensity measures and proposes two methods for selecting effective periods to improve the efficiency of structural response predictions. The authors construct a three‑parameter vector IM comprising first‑mode spectral acceleration, a spectral‑shape indicator at a second period, and the ground‑motion parameter ε, and use it to predict the probability distribution of structural response while accounting for collapse effects. The study shows that incorporating a spectral‑shape parameter improves prediction efficiency, but the ground‑motion parameter ε still needs to be accounted for, either via informed record selection or by including ε in the vector IM.

Abstract

Vector-valued ground motion intensity measures (IMs) are developed and considered for efficiently predicting structural response. The primary IM considered consists of spectral acceleration at the first-mode structural period along with a measure of spectral shape which indicates the spectral acceleration value at a second period. For the IM to effectively predict response, this second period must be selected intelligently in order to capture the most relevant spectral shape properties. Two methods for identifying effective periods are proposed and used to investigate IMs for example structures, and an improvement in the efficiency of structural response predictions is shown. A method is presented for predicting the probability distribution of structural response using a vector IM while accounting for the effect of collapses. The ground motion parameter ε is also considered as part of a three-parameter vector. It is seen that although the spectral shape parameter increases the efficiency of response predictions, it does not fully account for the effect of ε. Thus, ε should still be accounted for in response prediction, either through informed record selection or by including ε in the vector of IM parameters.

References

22