Tracking the normal modes of an overpass highway bridge using Distributed Acoustic Sensing

TL;DR

The paper addresses the challenge of non-invasively characterizing bridge dynamics with high spatial resolution. It employs Distributed Acoustic Sensing (DAS) to convert fiber-optic cables into dense strain-rate sensors and applies Frequency Domain Decomposition on the cross-spectral density of DAS signals to perform Operational Modal Analysis (OMA) and extract modal shapes directly from strain-rate data. The authors identify eight normal modes of a strut-frame overpass with submetric spatial sampling, validate DAS results against a previous seismometer-based campaign, and document strong seasonal effects that shift or suppress certain modes due to temperature-driven boundary-condition changes. The work demonstrates DAS as a cost-effective, scalable SHM tool capable of long-term monitoring, enabling continuous observation of seasonal dynamics and modal evolution with high spatial fidelity.

Abstract

Distributed Acoustic Sensing (DAS) of ambient vibrations is a promising technique in the context of structural health monitoring of civil engineering structures. The methodology uses Rayleigh backscattered light from small deformations at different locations of the sensed fiber-optic cable, turning it into a large array of equally distributed strain sensors. In this paper, we demonstrate the feasibility of using DAS technology to record dynamic strain used for modal identification through the Operational Modal Analysis (OMA) of a strut-frame bridge overpassing the A8 highway in southeastern France. Modal identification using DAS data is successful despite its predominantly axial sensitivity (along fiber), though the help of three-component seismometers is useful for discriminating the main motion direction of each identified mode. The identification of 1 bridge's normal modes with unprecedented spatial resolution is obtained from the lowest (transverse and longitudinal) modes to high-order modes that present significant vertical motion. In addition, strong seasonal effects are observed in both the absolute frequency values and the modal shapes of the first transverse and longitudinal modes of the bridge, comparing ambient vibration testing and DAS surveys carried out in the summer and winter periods.

Figures (8)

• Figure 1: a): Location of strut bridge over the A8 highway in southeast France. The path of the fiber is indicated in orange. The DAS interrogator was located inside a building just west of the bridge. A scale bar and a North arrow are provided on the bottom right. Imagery © 2023 Maxar Technologies, Map data © 2023 Google. b) view from the highway of the strut-frame bridge under study.
• Figure 2: Pictures taken during the DAS survey on June 2nd, 2022. a) general view of the bridge deck b) detail of one seismometer station and the layout of the fiber secured under the orange tape, and c) two seismometer stations side-by-side.
• Figure 3: a) extract of five minutes of strain rate data recorded by the optic fiber. b) Fourier spectra of each channel. Channels 1 to 76 correspond to the western side of the deck, while channels 97 to 171 to the eastern side of the deck.
• Figure 4: Schema of the deformation of an infinitesimal portion of the bridge deck, when the optic fiber is located at a distance $y$ from the longitudinal neutral axis of the beam.
• Figure 5: Comparison of the first singular values of several datasets: ambient vibration testing (AVT) in the winter of 2019, AVT in the summer of 2022, and DAS in the summer of 2022 and winter of 2024. In vertical solid lines, the first four normal modes identified in 2022 are indicated.