Experimental–numerical Framework for Intelligent Finite Element Model Updating of a Full-scale Laboratory Footbridge Structure

Key findings from this study

• The study found that vibration modes measured experimentally did not appear in the initial finite element model, necessitating systematic model updating.
• The researchers demonstrate that component-level testing of spliced beams and composite deck panels revealed stiffness properties inconsistent with preliminary modelling assumptions.
• The authors report that boundary condition refinement, material property adjustment, and section parameter modification within an optimisation framework reduced natural frequency discrepancies to below 8 percent.

Overview

A full-scale laboratory footbridge comprising spliced steel girders and composite sandwich deck panels served as a test-bed for structural dynamics investigations. Preliminary finite element modelling and experimental modal analysis revealed discrepancies between predicted and measured dynamic behaviour. Component-level testing of structural elements informed systematic model updating via optimisation-based adjustment of boundary conditions, material properties, and section parameters.

Methods and approach

Experimental modal analysis identified dynamic properties of the assembled footbridge structure. Preliminary finite element models predicted structural response based on initial design assumptions. Component-level testing characterised splice connections in primary beams and composite action in sandwich deck panels. Sensitivity-driven optimisation adjusted model parameters iteratively to minimise discrepancies between computed and measured natural frequencies and mode shapes. Modal assurance criterion quantified agreement between experimental and numerical mode shapes.

Results

The updated finite element model achieved natural frequency discrepancies below 8 percent relative to measured values. Mode shapes demonstrated high agreement with experimental data according to modal assurance criterion evaluation. Model updating revealed that splice connections reduced effective stiffness in primary beams compared to initial assumptions. Sandwich deck composite action required increased stiffness representation relative to preliminary modelling. Boundary condition refinement and material property adjustment within the optimisation framework proved essential to capture measured structural behaviour.

Implications

The study demonstrates that component-level testing provides critical information for identifying deficiencies in global structural models. Splice connections and composite material systems require explicit mechanical characterisation to ensure adequate finite element representation. Sensitivity-driven optimisation frameworks enable systematic identification of parameters requiring adjustment when measured and predicted behaviour diverge. The methodology supports development of reliable predictive models for subsequently conducted experimental investigations on the test-bed structure. Laboratory-scale structures with reconfigurable elements benefit substantially from rigorous model updating prior to complex dynamic testing programmes.

Scope and limitations

This summary is based on the study abstract and available metadata. It does not include a full analysis of the complete paper, supplementary materials, or underlying datasets unless explicitly stated. Findings should be interpreted in the context of the original publication.