Bridges in the United States are inspected every two years, as outlined by the National Bridge Inspection Standards, to maintain safe bridge operation and prevent structural failures. Load testing during these biannual inspections is not mandatory but can provide additional insight into the performance of the bridge. There are three categories of load testing – diagnostic load, proof load, and in-service load. Diagnostic tests use a fraction of the design live loads, while proof load tests use live loads that correspond to specific design codes. In-service tests measure how a bridge responds to ambient traffic over a specified period of time. Unfortunately, only a small number of bridges undergo load testing because of the cost and challenges with instrumenting a bridge. Vision-based measurements might be the solution.

Researchers Mehrdad Ghyabi, Luke C. Timber, Gholamreza Jahangiri, David Lattanzi, Harry W. Shenton III, Michael J. Chajes, and Monique H. Head present a case study using vision-based methods for bridge load testing. In their study, “Vision-Based Measurements to Quantify Bridge Deformations” in the Journal of Bridge Engineering, the authors used two controlled load tests to demonstrate that digital image correlation and phase-based flow could be used to measure displacement magnitudes similar to more expensive traditional methods. More study is warranted, but this shows that alternative approaches offering more flexibility for bridge inspectors could ensure safer bridges. Learn more about their research at The abstract is below.


Many factors are considered when inspecting and evaluating the overall condition of a bridge. Of particular consideration here is load testing of bridges to evaluate the existing load-carrying capacity. Sensor systems are often mounted directly to girders for this assessment; however, installing sensors and data acquisition systems can be an expensive and time-consuming process, particularly given that load testing does not warrant long-term monitoring. As an alternative, noncontact remote sensing techniques have been developed for measuring structural deformations, and have the potential to be used for static load test applications. These approaches do not require sophisticated instrumentation installations, and can provide a denser array of measurements, compared with conventional sensors. A particular focus has been on techniques that use video recordings, tracking the motion between subsequent video frames via computer vision methods. There are now commercial offerings for such measurement systems, as well as an array of techniques that can be used for custom applications. While such methods have seen significant testing under laboratory conditions, there are only a limited number of studies that provide comparative methodological analyses under full-scale field conditions. This paper presents a case study on the use of vision-based methods for bridge load testing, and provides a comparison of a digital image correlation (DIC) approach with a phase-based optical flow method. Two sets of field experiments were performed on bridges in the state of Delaware. The results show that vision-based methods can provide comparable results to conventional sensor installations, given sufficient consideration of the unique technical demands of these methods, as well as operational logistics. In most cases, the DIC and phase-based methods provided comparable results, though the DIC system yielded generally better accuracy, owing to a combination of algorithmic differences and additional signal postprocessing.

Read the paper in full in the ASCE Library: