Blind-prediction of the dynamic behavior of 3D-printed masonry-like structures under shake-table loading via a high-fidelity numerical modeling approach

Abstract

This paper presents a modeling approach for high-fidelity blind-prediction of dynamic responses of 3D-printed masonry-like structures, as part of a contest organized by Pacific Earthquake Research Center (PEER) for simulating shake table tests on 29 identical ⨅-shaped 1:15-scaled sand-based 3D-printed specimens, each subjected to a different earthquake. The contest challenged participants to predict experimental outcomes without access to test results. Leveraging their modeling approach originally developed for regular masonry, the authors proposed an innovative methodology to simulate these structures, implementing extensions to overcome challenges such as representing their continuum nature within a discrete block-and-joint framework and simulating their small-scale response via 1:1-scale counterparts. The numerical model blind-predicted the experimental outcomes with highest accuracy among participants. Parametric studies, before and after access to modal characteristics, showed the importance of such information for simulation accuracy, and the ability of the approach to investigate variability of dynamic responses in complement to physical tests.