Share Email Print
cover

Proceedings Paper

Multiscale model updating of a curved highway bridge
Author(s): Priscilla Mensah-Bonsu; Shinae Jang
Format Member Price Non-Member Price
PDF $14.40 $18.00

Paper Abstract

Finite element model updating based on a multi-scale data is demonstrated on a skewed in-service highway bridge. The multi-scale data approach provides an evidence-based method to create a bound of model class to ensure that the optimum model retains physical connectivity to the real structure. The need for this hybrid approach to model selection comes from the challenges of applying model updating to online structural health monitoring (SHM) strategy based on output-only measurements of in-service highway bridges, which should consider various uncertainties. In vibration-based FE model updating methods, the optimum model is selected by minimizing the error between modal parameters of the model and the real structure. A major drawback of model selection is the probability that the optimum model has no physical connectivity with the real structure. This is because a large set of updating parameters is required to increase accuracy. In this paper, an evidence-based approach to model selection using temperature-induced tilts is implemented in which the cyclical static behavior of a bridge is used to create a bound of possible models. This approach has a strong potential to be applicable to large civil structures with sparse array of sensors.

Paper Details

Date Published: 6 April 2012
PDF: 9 pages
Proc. SPIE 8345, Sensors and Smart Structures Technologies for Civil, Mechanical, and Aerospace Systems 2012, 834547 (6 April 2012); doi: 10.1117/12.915583
Show Author Affiliations
Priscilla Mensah-Bonsu, Univ. of Connecticut (United States)
Shinae Jang, Univ. of Connecticut (United States)


Published in SPIE Proceedings Vol. 8345:
Sensors and Smart Structures Technologies for Civil, Mechanical, and Aerospace Systems 2012
Masayoshi Tomizuka; Chung-Bang Yun; Jerome P. Lynch, Editor(s)

© SPIE. Terms of Use
Back to Top