Effects of ground motion scaling on the response of structures considering the interdependency between intensity measures and scale factors

• Published in: Engineering structures 2020-04, Vol.209, p.110007, Article 110007
• Main Authors: Wen, Weiping, Ji, Duofa, Zhai, Changhai
• Format: Article
• Language: English
• Subjects: Acceleration; Bias; Ground motion; Ground motion scaling; Intensity measure; Interdependency; Nonlinear analysis; Nonlinear dynamics; Scale factor; Scaling; Seismic design
• ISSN: 0141-0296; 1873-7323
• DOI: 10.1016/j.engstruct.2019.110007

•The effects of ground motion scaling on structural responses are quantified.•The method is proposed to quantify the interdependency between IMs and SFs.•Housner spectrum intensity is the best IM to scale ground motions.•The allowable SFs are determined for various IMs based on the acceptable bias. Ground motion scaling has been widely applied in seismic design and evaluation to perform nonlinear dynamic analysis. However, the quantitative interdependency between intensity measures (IMs) and scale factors (SFs) has not received sufficient attention in previous studies. In this study, effects of ground motion scaling are investigated by considering the effects of IMs and SFs on the response of structures. The bias introduced by ground motion scaling is defined as the difference in engineering demand parameter values obtained from seismic demand models developed with scaled and unscaled ground motions. The bias and allowable SF are comprehensively used to quantify the interdependency between IMs and SFs. Eight different IMs are selected, and 1593 ground motions are collected from the PEER NGA-West2 database. The results indicate that the Housner spectrum intensity is superior to the other IMs based on the entire period region, and that it can control the bias to within about 20% even for a SF as large as 10.0. The spectral acceleration and peak ground velocity are secondary alternatives, and the corresponding biases are generally within 40% when the ground motions are scaled within five times (i.e., the SF varies from 0.2 to 5.0). The allowable SFs are determined for various IMs when the acceptable bias is assumed as 20%.