Premium
Robustness of simplified analysis methods for rocking structures on compliant soil
Author(s) -
Sieber Max,
Klar Sebastian,
Vassiliou Michalis F.,
Anastasopoulos Ioannis
Publication year - 2020
Publication title -
earthquake engineering and structural dynamics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 2.218
H-Index - 127
eISSN - 1096-9845
pISSN - 0098-8847
DOI - 10.1002/eqe.3294
Subject(s) - nonlinear system , robustness (evolution) , bilinear interpolation , structural engineering , stiffness , engineering , response analysis , mathematics , statistics , biochemistry , chemistry , physics , quantum mechanics , gene
Summary Recognizing the beneficial effect of nonlinear soil–foundation response has led to a novel design concept, termed ‘rocking isolation’. The analysis and design of such rocking structures require nonlinear dynamic time history analyses. Analyzing the entire soil–foundation–structure system is computationally demanding, impeding the application of rocking isolation in practice. Therefore, there is an urgent need to develop efficient simplified analysis methods. This paper assesses the robustness of two simplified analysis methods, using (i) a nonlinear and (ii) a bilinear rocking stiffness combined with linear viscous damping. The robustness of the simplified methods is assessed by (i) one‐to‐one comparison with a benchmark finite element (FE) analysis using a selection of ground motions and (ii) statistical comparison of probability distributions of response quantities, which characterize the time history response of rocking systems. A bridge pier (assumed rigid) supported on a square foundation, lying on a stiff clay stratum, is used as an illustrative example. Nonlinear dynamic FE time history analysis serves as a benchmark. Both methods yield reasonably accurate predictions of the maximum rotation θ max . Their stochastic comparison with respect to the empirical cumulative distribution function of θ max reveals that the nonlinear and the bilinear methods are not biased. Thus, both can be used to estimate probabilities of exceeding a certain threshold value of θ. Developed in this paper, the bilinear method is much easier to calibrate than the nonlinear, offering similar performance.