Quant CR for Transformative Bridge Asset Management

Project Information

Link to Latest Report: Coming Soon


We propose developing an artificial intelligence (AI)-powered quantitative condition rating (QUANT CR) model which operates on a low-cost geographic information system (GIS) platform, aiding local and state bridge owners in maintenance, repair, and replacement (MRR) decisions while preserving the established inspection and condition rating practices. The next generation asset management system leverages the knowledge gained from 50+ years of bridge inspection practices but is predictive, forward-looking, and transformative. QUANT CR embodies insights gained from the understanding of human behavior to better assist bridge owners in decision-making. Thus, we envision QUANT CR will be operated in parallel with the existing bridge condition ratings and provide simple decision aids for bridge owners. We believe bridge condition ratings can be better predicted by modern machine learning methods, leveraging the historic data, evolving element condition ratings, and detailed defect items. Additionally, deep learning widely used for text recognition enables an analysis of inspectors’ narratives describing bridge conditions. Lastly, computer vision and deep generative learning help bridge owners visualize the outcomes of their decisions.

QUANT CR will:
1. Reduce human errors and aid in training bridge inspectors.
2. Close the knowledge gap between predicted and actual bridge performance, noting that it only gets better with time/data.
3. Aid bridge owners in budgetary planning by providing access to performance prediction data and MRR options.
4. Enable the use of technologies such as a drone, autonomously assigning condition ratings and ultimately writing an inspection report that is better than one written by a human-writer. Therefore, this technology is expected to improve bridge inspection outcomes, reduce costs, increase access to performance predictions, aid in MRR decisions, which are essential for asset management at local and state government levels. This innovation will particularly serve to aid rural areas with limited resources.


This project aims to empower local and state bridge owners to make informed decisions on MRR,optimizing budget plans across various scenarios. Thus, the research approach centers on developing an AI-powered quantitative condition rating (QUANT CR) system and decision analysis tool for bridges, which leverages historical data, including the extensive records of the National Bridge Inventory (NBI)’s element data, Long-Term Bridge Performance Program, bridge MRR scenarios, inspectors’ narratives, and ultimately risk management including environmental,traffic, vehicle, and other relevant data to predict long-term bridge performance. A web-based Geographic Information System (GIS) platform will store and visually represent bridge network assets, providing uniform access to all databases. A hierarchical approach to quantify bridge condition ratings and assigning them to the network analysis tool will improve long-term bridge performance predictions. Deep learning algorithms will analyze historical data and detailed bridge condition narratives to improve the bridge performance  predictions at element level. Incorporating the element-level inspection results and detailed defect items  is a key task for building a Quant CR model. Text recognition techniques will further interpret inspectors’ descriptive reports, transforming qualitative comments into actionable quantitative data.


The proposed project includes several tasks

  • Task 1 – Acquisition of bridge performance/condition data into geospatial format
    The research team will acquire the bridge performance, condition, structural, and traffic data for nationwide bridge network assets from different sources and process into a geospatial data format.
    In addition to the deck condition data, National Bridge Inventory (NBI) provide data related to operational conditions, functional descriptions, and inspection data in geospatial format. Additionally, Highway Performance Monitoring System (HPMS) database is available and
    includes detailed roadway and bridge deck surface condition data.
  • Task 2 – Development of a bridge condition rating prediction model
    To create a Quant CR model, the research team will adopt both Multi-Criteria Decision Analysis (MCDA) and Analytic Hierarchy Process (AHP). The use of MCDA and AHP together ensures a systematic approach to decision-making by combining multiple criteria through MCDA [4-5] and
    evaluating their relative importance through AHP [6-7]. For example, three independent criteria may be applied for evaluating bridge performance: (a) structural rating; (b) condition rating; and (c) scour rating. After determining the relative importance of the elements of multi-criteria, the spatial layers will be in Geographical Information Systems. In addition to the state Department of Transportation open data sources, attributes data can be collected from the NBI bridge dataset available online. Structural ratings can be calculated using Annual Average Daily Traffic (AADT) and structure type, scour ratings are calculated using reported scour risk factors and condition ratings from the available databases. Combining these three ratings, we will generate a composite index of overall bridge condition with a 5-score scale, with 1 being the best condition and 5 being the most vulnerable. The ‘Quant CR’ is an AI-based performance prediction model which will be further refined based on large datasets at the network-level. The proposed prediction model will allow for the quantitative decision analysis on a multitude of scenarios. For example, it will allow evaluation of
    bridge performance under truck platoon loading and extreme weather events.
  • Task 3 – Enhance with a graph-based network model
    The research team will develop graph-based networks of bridges and roadways and integrate them
    using the geographical interdependence among them. These models will be used to conduct various topological analysis for the bridge network. In graph theory, a graph G is a triple consisting
    of a vertex (node) set V (G), an edge (link) set E(G), and relation that associates with each edge, two vertices called its endpoints [8]. In the context of freight network, bridges serve as nodes (V) and while the relationships/interdependencies among them serve as edges (E). The research team will identify three (3) types of interdependencies to construct the coupled network-of-networks of the bridge systems. Interdependence includes: (1) Physical – where two nodes are physically connected by a link to exchange material outputs, so that the case of a failed state of one influences the other; (2) Geographic – when multiple network assets are in close geographical proximity, making them susceptible to fail from the same external shock events; and (3) Logical – which explains how network asset failures may start cascading failures that go beyond physical, or geographic interdependencies.
  • Task 4 – Verify model and explore benefits from employing other technologies
    The research team aims to use the Quant CR system to conduct pilot studies to explore rewards
    from employing advanced bridge monitoring technologies and roadway maintenance data (e.g.,
    patching on bridge deck surfaces), improve its user-interface that enables state and local government planners to input multiple management strategies, assess potential impacts of automated and heavy trucking technologies, bridge monitoring technologies, and climate conditions, and evaluate alternative routes. By incorporating real-world scenarios, the system is expected to prioritize roadway maintenance and evaluate resiliency against environmental changes
    and increased freight traffic.

Research Team:
Principal Investigator: Bjorn Birgisson
Co-Principal Investigator: Mi G. Chorzepa, Sajib Saha