## Model updating of structural systems

- March 16, 2015
- 1 p.m.
- LeConte 312

## Abstract

Numerical models of structural systems such as bridges and buildings play an important role in design and maintenance of our civil infrastructure. These models have proven useful when trying to estimate the behavior of a new structure for design purposes. During design two steps are followed. First, a numerical model of the structure is used to determine the likely forces that structural members will have to resist during the structureâ€™s life time. Then, the size of the elements is determined to resist these forces. Safety factors when sizing elements usually covers for any errors in the estimation of the structural forces.

When a structure is retrofitted the engineer would like to have a meaningful model of the structure in its current state. A large body of literature has demonstrated the the numerical models do not necessarily predict the behavior of the existing structure. In these cases model updating techniques are used to enhance the numerical model of the bridge or building. Features of the existing structure are estimated based on experimental data and the model is updated based on the measured features. Several model updating techniques available in the literature can result in numerical models that represent the input/output relationship of the system but the physical parameters of the model can lack physical meaning. For example, in the area of structural dynamics, an updated model of a bridge could have natural frequencies representing the actual structure, but the stiffness or mass of the structural elements might not be realistic. This lack of physical meaning in the updated parameters prevents the use of updated models in some engineering applications and it is likely due to simplifications in the model, limited sensors in the experimental process, or simplifications in the model updating techniques.

The Structural Dynamics and Intelligent Infrastructure laboratory (http://sdii.ce.sc.edu) has performed research to obtain meaningful physical parameters of structural systems for the last 10 years. This presentation discusses advanced performed in sensing and model updating techniques with the specific focus of obtaining numerical models with physical meaningful parameters. In particular, the presentation will discuss mobile sensors and advances in Bayesian inference for structural model updating.