3D Rocking Motion: Blind Prediction Contest Results and Influence of Evaluation Metric on the Rankings

It has been claimed that numerical rocking motion models are not only inaccurate, but that all rocking structures behave unpredictably. This paper revisits the issue of rocking motion unpredictability and explores whether not the response to a single ground motion, but the statistics of the responses to an ensemble of ground motions is predictable. To this end, a 3D rocking podium structure was constructed and tested on a shake table. A blind prediction contest was organized, where the contestants were invited to predict the CDF of the time-maxima of the responses. There were 13 participant teams that used FEM, DEM and rigid body models. It was found that several teams were able to predict the CDF with a decent accuracy. Both FEM and DEM can perform well, depending on the input parameters. Hence, it can be concluded that 3D rocking motion is predictable in the statistical sense.