Evaluation of Data in the National Bridge Inventory (NBI) on Missouri, USA Bridges to Predict Bridge Deterioration
Presented By: John Myers
Affiliation: Missouri S & T
Description: In the state of Missouri, USA, The Missouri Department of Transportation (MoDOT) is responsible for the inspection and maintenance of approximately 10,000 bridges and culverts on the state roadway system. The condition of these structures is assessed biannually to monitor deterioration that occurs as a result of environmental exposure and traffic loading. To effectively manage these important assets, MoDOT launched a research effort to develop deterioration models that would allow them to better project future preservation and rehabilitation activities and develop data-driven asset management plans. This study developed deterioration curves for different bridge components and culverts, identified trends in deterioration patterns, and develop recommendations for cost-effective bridge types. To conduct this analysis, records from the Federal Highway Administration’s (FHWA) National Bridge Inventory (NBI) were obtained for the years of 1983 - 2019. These records document the inspection results from biannual inspection through condition ratings (CR) that describe the condition of bridge components (deck, superstructure, and substructure) and culverts on a 0 - 9 numerical scale. This paper will discuss the key takeaways from the study in terms of influencing factors, rate of deterioration once certain CR’s occur and future recommendations in terms of data collection.
Enhancing Condition Predictions of Concrete Bridges and Budgetary Planning through Analysis of NBI, NBIE, and Maintenance Data
Presented By: Mi-Geum Chorzepa
Affiliation: University of Georgia
Description: A Recurrent Neural Network (RNN) architecture has been developed to improve bridge deterioration predictions. Specifically, the application of RNN models, including Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU), has shown promise in capturing the temporal and sequential characteristics of bridge condition data, leading to more accurate forecasts. Unlike traditional feedforward networks, RNNs excel at modeling complex temporal dependencies in time series data, which is crucial for understanding how environmental factors impact bridge conditions over time. A critical advantage of LSTM and GRU models is their three gating mechanisms, which enable selective information retention and efficient modeling of time dependencies. To further enhance prediction accuracy, this study incorporates a TimeDistributed (TD) layer within LSTM and GRU models, allowing RNNs to generate outputs based on the entire sequence of each bridge and improving adaptability to shifts in input sequences. This approach captures complex temporal dependencies and nonlinear relationships more effectively, enabling simultaneous forecasting for multiple bridges and ensuring that each bridge’s historical performance is considered in the forecasting model. The research methodology includes feature selection to identify key variables influencing bridge conditions within a state’s inventory, along with structured dataset preparation for deep learning models. The TD-enhanced architecture is trained on historical data and validated using unseen datasets, demonstrating predictive accuracy and robustness. This framework provides reliable forecasts that support proactive bridge maintenance and contribute to safer and more resilient infrastructure management, addressing the challenges of aging assets and budget constraints.
Incorporating NBI Deterioration Models in Bridge Deck Preservation Decisions
Presented By: Mohamed ElBatanouny
Affiliation: Wiss, Janney, Elstner Associates, Inc.
Description: Bridge deck deterioration poses a significant challenge for state transportation agencies. Many state Departments of Transportation allocate substantial portions of their budgets and time to maintain their bridge deck inventories, particularly in regions where de-icing salts are used. As such, several states, in collaboration with the FHWA, are funding a pooled-fund study to develop a web tool designed to assist bridge maintenance and preservation engineers in selecting optimal maintenance strategies. This presentation offers an overview of the Bridge Deck Preservation Tool, showcasing how it integrates existing NBI deterioration models to estimate the remaining service life and calculate life cycle costs for various maintenance and preservation options for the bridge deck in question.
Bridge Evaluation Based on the National Bridge Inventory
Presented By: Jun Wang
Affiliation: University of Hawaii Manoa
Description: This presentation discusses Big Data analytics on the condition evaluation of highway bridges in the United States. A large dataset comprising 1,002,172 bridge decks and superstructures is constructed, based on the National Bridge Inventory (NBI), and categorized into four service zones as specified in the American Association of State Highway Transportation Officials (AASHTO) Load and Resistance Factor Design (LRFD) Bridge Design Specifications. The condition rating of the bridge members is examined statistically and probabilistically, in conjunction with the effect of traffic and environment (i.e., temperature and precipitation). The statistical characterization of the members indicates that concrete-based superstructures are predominant in Zones 1, 2, and 3 (79%, 72%, 85%, respectively), whereas steel- and timber-based superstructures account for 51% and 21% in Zone 4, respectively. The bridges in Zones 1 and 3 are subjected to significantly high traffic-induced loading relative to those in Zones 2 and 4. Thermal loading is noticeable in Zones 1 and 4. The deterioration of bridge decks rapidly develops at the bridges’ early service life and stabilizes with time owing to maintenance and repair efforts. According to a two-factor analysis of variance, adequate selection of structural types, dependent upon service environments, enhances the performance and longevity of constructed bridges. The likelihood of deterioration of bridges constructed in Zones 1 and 3 is higher than that of the bridges in Zones 2 and 4.
Characterizing Bridge and Culvert Deterioration in the Western U.S. Using National Bridge Inventory Data
Presented By: Ben Dymond
Affiliation: Northern Arizona University
Description: The characteristics of more than 600,000 bridges in the United States are recorded in the FHWA National Bridge Inventory (NBI) database. The goal of this study was to quantitatively analyze the deterioration of concrete bridge decks and culverts between the years of 1992 and 2022 using this publicly available database. The results were determined using condition ratings on a scale from 0 to 9 that describe the existing, in-place bridge as compared to the as-built condition; a rating of 9 is excellent while a condition of 0 indicates failure. Two NBI coding items were used – Item 58 Deck that describes the overall condition rating of the deck and Item 62 Culverts that assess the overall condition evaluation of a culvert. Furthermore, the effect of several key parameters such as type of superstructure material, type of superstructure cross-sectional geometry, average daily traffic (ADT), type of wearing surface, and geographical location on deck or culvert condition were evaluated. Specifically, data and results associated with seven western states (AK, AZ, CA, ID, NV, OR, WA) will be presented. The trends are based on NBI ratings over time, and bridge owners can use the results to properly assess deck condition, predict future deck condition, and plan for future bridge maintenance.