A Framework for Rapid Performance Assessment of Bridges Under Flood Hazard Using Machine Learning

Recent heavy rainfall and flood events highlight the growing threat to transportation infrastructure and their surrounding communities. Floods can severely damage bridges through pier scour, hydraulic loads, or debris impact. These effects may lead to sudden failures and impede critical emergency response and recovery operations. Traditional flood risk assessment methods often rely on expert opinion or visual inspection results, which may not accurately capture the true structural condition. Additionally, conventional stochastic approaches, while more robust, are computationally expensive and impractical for time-sensitive applications. To address these challenges, the proposed research develops a framework for rapid fragility quantification of bridges under flood conditions. The framework utilizes machine learning models that are trained to capture complex flood-structure interactions and generate predictive fragility curves for failure assessment and/or risk-based decision-making. The framework considers flood and bridge attributes to generate the fragility profile directly in real-time. Accordingly, this approach enables faster, more reliable, and computationally efficient assessment compared to traditional methods. Rather than focusing on a single bridge, the framework will be trained using high-fidelity finite element analysis results from an ensemble of models that covers a particular class of bridges. After training, the framework can generate the fragility profiles of any bridge within the class given the bridge attributes (e.g., span length, number of lanes, foundation type, etc.). At this stage, focus will be placed on single- or multi-span, simply supported girder bridges (steel or concrete). Other bridge types may be considered in subsequent studies.

• Record URL:
  • https://www.sptc.org/cy3-osu-03

Language

• English

Project

• Status: Active
• Funding: $65,000.00
• Contract Numbers:

  69A3552348306 (CY3-OSU-03)

• Sponsor Organizations:

  Southern Plains Transportation Center

  University of Oklahoma
  202 W Boyd St, Room 213A
  Norman, OK  United States  73019

  Oklahoma Department of Transportation

  200 NE 21st Street
  Oklahoma City, OK  United States  73105
• Managing Organizations:

  University of Oklahoma, Norman

  School of Civil Engineering and Environmental Science
  202 West Boyd Street, Room 334
  Norman, OK  United States  73019
• Project Managers:

  Ghasemi, Hamid

• Performing Organizations:

  Oklahoma State University, Stillwater

  School of Civil & Environmental Engineering
  Stillwater, OK  United States  74078
• Principal Investigators:

  Soliman, Mohamed

• Start Date: 20260101
• Expected Completion Date: 20270101
• Actual Completion Date: 0
• USDOT Program: UTC

Subject/Index Terms

• TRT Terms: Failure; Finite element method; Flood damage; Floods; Girder bridges; Machine learning; Predictive models; Risk assessment
• Subject Areas: Bridges and other structures; Highways; Maintenance and Preservation; Planning and Forecasting;

Filing Info

• Accession Number: 01975688
• Record Type: Research project
• Source Agency: Southern Plains Transportation Center
• Contract Numbers: 69A3552348306 (CY3-OSU-03)
• Files: UTC, RIP
• Created Date: Jan 5 2026 10:41PM