Title: Reliable condition assessment of structures using uncertain or limited field modal data

Authors: Mojtaba Dirbaz; Mehdi Modares; Jamshid Mohammadi

Addresses: MWH Americas, Inc., 305 N. Orleans Street Suite 1301, Chicago, IL 60654, USA ' Department of Civil, Architectural, and Environmental Engineering, Illinois Institute of Technology, 3201 South Dearborn Street, Chicago, IL 60616, USA ' Department of Civil, Architectural, and Environmental Engineering, Illinois Institute of Technology, 3201 South Dearborn Street, Chicago, IL 60616, USA

Abstract: A new method for reliable condition assessment and damage detection of structures is developed that uses stochastic Finite Element (FE) analysis and uncertain or limited field modal data quantified as bounded random variables. The steps of this method are to (1) construct the stiffness and mass matrices of undamaged structure, (2) quantify the uncertainty in measured modal data as bounded random variables, (3) perform random sampling simulations to obtain the stiffness of existing structure through iterative optimisation method, (4) utilise the element stiffness matrices in each random sampling to identify the damage members based on the change in element stiffness and (5) determine the bounds on location, and extent of damage, that caused the degradation of the system. It is shown that in the presence of uncertainty in, or with limited information on, the modal data, this method is capable of determining the bounds on location and extent of damage.

Keywords: damage detection; optimisation; probabilistic methods; structural reliability; degradation; condition assessment; uncertainty; stochastic FEM; finite element method; random sampling; simulation; structural stiffness.

DOI: 10.1504/IJRS.2015.072721

International Journal of Reliability and Safety, 2015 Vol.9 No.2/3, pp.220 - 234

Received: 22 Oct 2014
Accepted: 23 Aug 2015

Published online: 27 Oct 2015 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article