Mode Shape Expansion for Lively Pedestrian Bridges through Kriging

Abstract

This paper investigates the use of kriging as a powerful tool for mode shape interpolation in flexible pedestrian bridges. Kriging can be used to interpolate a set of sparse mode shape coordinates obtained through experimental modal identification techniques onto a dense structure geometry, leading to better mode shape visualization and efficient model updating. Kriging does not require a finite-element (FE) model (analytical modes) to be available a priori, as it is purely statistical in nature. Despite this key advantage, kriging has not been adopted for mode shape interpolation due in part to the relative lack of exposure within the modal analysis community and the arbitrariness of the parameter defining the spatial correlation between discrete geometry points. The first objective of this paper is to describe kriging in detail, especially focusing on pedestrian bridges. Second, modifications to the computational procedure for kriging are presented with a new metric to automatically tune the correlation parameter and a batch process to undertake kriging for mode shape interpolation. Details of the procedure are discussed through both numerical and experimental field data obtained from flexible pedestrian bridges. It is anticipated that through the modifications described in this paper, kriging will spur greater interest from the pedestrian bridge modal analysis community.