National Bridge Inventory Database Research Articles

Modelling bridge deterioration using long short-term memory neural networks: a deep learning-based approach

PurposeBridge deterioration is a critical risk to public safety, which mandates regular inspection and maintenance to ensure sustainable transport services. Many models have been developed to aid in understanding deterioration patterns and in planning maintenance actions and fund allocation. This study aims at developing a deep-learning model to predict the deterioration of concrete bridge decks.Design/methodology/approachThree long short-term memory (LSTM) models are formulated to predict the condition rating of bridge decks, namely vanilla LSTM (vLSTM), stacked LSTM (sLSTM), and convolutional neural networks combined with LSTM (CNN-LSTM). The models are developed by utilising the National Bridge Inventory (NBI) datasets spanning from 2001 to 2019 to predict the deck condition ratings in 2021.FindingsResults reveal that all three models have accuracies of 90% and above, with mean squared errors (MSE) between 0.81 and 0.103. Moreover, CNN-LSTM has the best performance, achieving an accuracy of 93%, coefficient of correlation of 0.91, R2 value of 0.83, and MSE of 0.081.Research limitations/implicationsThe study used the NBI bridge inventory databases to develop the bridge deterioration models. Future studies can extend the model to other bridge databases and other applications in the construction industry.Originality/valueThis study provides a detailed and extensive data cleansing process to address the shortcomings in the NBI database. This research presents a framework for implementing artificial intelligence-based models to enhance maintenance planning and a guideline for utilising the NBI or other bridge inventory databases to develop accurate bridge deterioration models. Future studies can extend the model to other bridge databases and other applications in the construction industry.

Intelligent condition prediction model for bridge infrastructure based on evaluating machine learning algorithms

PurposePrediction models are essential tools for transportation agencies to forecast the condition of bridge decks based on available data, and artificial intelligence is paramount for this purpose. This study aims at proposing a bridge deck condition prediction model by assessing various classification and regression algorithms.Design/methodology/approachThe 2019 National Bridge Inventory database is considered for model development. Eight different feature selection techniques, along with their mean and frequency, are used to identify the critical features influencing deck condition ratings. Thereafter, four regression and four classification algorithms are applied to predict condition ratings based on the selected features, and their performances are evaluated and compared with respect to the mean absolute error (MAE).FindingsClassification algorithms outperform regression algorithms in predicting deck condition ratings. Due to its minimal MAE (0.369), the random forest classifier with eleven features is recommended as the preferred condition prediction model. The identified dominant features are superstructure condition, age, structural evaluation, substructure condition, inventory rating, maximum span length, deck area, average daily traffic, operating rating, deck width, and the number of spans.Practical implicationsThe proposed bridge deck condition prediction model offers a valuable tool for transportation agencies to plan maintenance and resource allocation efficiently, ultimately improving bridge safety and serviceability.Originality/valueThis study provides a detailed framework for applying machine learning in bridge condition prediction that applies to any bridge inventory database. Moreover, it uses a comprehensive dataset encompassing an entire region, broadening the model’s applicability and representation.

Applying random forest algorithm for highway bridge-type prediction in areas with a high seismic risk

ABSTRACT Machine learning was used to bridge-type prediction in the preliminary design. Based on the decision tree, in the United States, the accuracy of highway bridge-type prediction showed to be until 95% in the areas with a low seismic risk, whereas it proved to be lower than 80% in the areas with a high seismic risk. This work improved the accuracy of highway bridge-type prediction in states with a high seismic risk, i.e. California, Nevada, Oregon, and Washington. Bridge basic datasets from these states were elaborated from the National Bridge Inventory database using the random forest algorithm, which was found to be better. Indeed, the average recall rate of accuracy of highway bridge-type prediction across these states was 85.4%, i.e. 5.8% higher than the most accurate recall rate according to the literature review. Additionally, this work determined the most important parameters of highway bridge-type prediction using sensitivity analyses. According to comparisons with some predictions based on important parameter combinations, the accuracy in California, Nevada, Oregon, and Washington was improved by a percentage of 10.9%. Moreover, the analysis of the ‘year built’ parameter underlined its importance. Determining bridge-types was improved, thus enabling the accumulation of technical experience and saving labor and time.

Improving the Predictive Analytics of Machine-Learning Pipelines for Bridge Infrastructure Asset Management Applications: An Upstream Data Workflow to Address Data Quality Issues in the National Bridge Inventory Database

Improving the Predictive Analytics of Machine-Learning Pipelines for Bridge Infrastructure Asset Management Applications: An Upstream Data Workflow to Address Data Quality Issues in the National Bridge Inventory Database

Survival Analysis Using Cox Proportional Hazards Regression for Pile Bridge Piles Under Wet Service Conditions

This paper studies the deterioration of bridge substructures utilizing the Long-Term Bridge Performance (LTBP) Program InfoBridgeTM and develops a survival model using Cox proportional hazards regression. The survival analysis is based on the National Bridge Inventory (NBI) dataset. The study calculates the survival rate of reinforced and prestressed concrete piles on bridges under marine conditions over a 29-year span (from 1992 to 2020). The state of Maryland is the primary focus of this study, with data from three neighboring regions, the District of Columbia, Virginia, and Delaware to expand the sample size. The data obtained from the National Bridge Inventory are condensed and filtered to acquire the most relevant information for model development. The Cox proportional hazards regression is applied to the condensed NBI data with six parameters: Age, ADT, ADTT, number of spans, span length, and structural length. Two survival models are generated for the bridge substructures: Reinforced and prestressed concrete piles in Maryland and reinforced and prestressed concrete piles in wet service conditions in the District of Columbia, Maryland, Delaware, and Virginia. Results from the Cox proportional hazards regression are used to construct Markov chains to demonstrate the sequence of the deterioration of bridge substructures. The Markov chains can be used as a tool to assist in the prediction and decision-making for repair, rehabilitation, and replacement of bridge piles. Based on the numerical model, the Pile Assessment Matrix Program (PAM) is developed to facilitate the assessment and maintenance of current bridge structures. The program integrates the NBI database with the inspection and research reports from various states’ department of transportation, to serve as a tool for condition state simulation based on maintenance or rehabilitation strategies.

An application of convolutional neural network for deterioration modeling of highway bridge components in the United States

This paper presents a deep learning-aided deterioration modeling approach for highway bridge components (i.e., decks, superstructures, and substructures). The method trains a set of deep learning models for the maximum likelihood estimation of parameters in a Markov chain. The likelihood function is based on observed numbers of transitions between different condition states. The deep learning is leveraged for efficient representation of various factors that influence the deterioration process. Aided by deep learning, the proposed method is suitable for deterioration modeling using large-scale and high-dimensional bridge datasets. The study demonstrates an application of this approach using the historical National Bridge Inventory database from 1993 to 2019. The Convolutional Neural Network is adopted as the deep learning model. The proposed method is validated with ten-fold cross-validation that encompasses a nationwide selection of 88,596, 101,414, and 96,244 bridge decks, superstructures, and substructures, respectively. The validation shows this approach achieved a robust and low prediction error with the maximum mean-squared-error near 0.5 in a 26-years-forecast. The proposed method is also compared with the conventional Artificial Neural Network and four selected Markov-chain based deterioration modeling approaches. The study shows this approach can be a promising data-driven tool for deterioration modeling of highway bridge components.

Development of a Risk Assessment Module for Bridge Management Systems in New Jersey

Bridges are critical for the mobility of our society and its economic growth. Available funds for bridge repair, maintenance, and rehabilitation are limited. The Moving Ahead for Progress in the 21st Century Act (MAP-21) introduced several new parameters for improving the management of bridge assets, such as bridge element evaluation, life-cycle analysis, and risk-based performance indicators. Risk-based methods account for the uncertainties embedded into engineering variables and long-term evaluations. The objective of this paper is to identify, assess, and quantify structural risk components to bridges using probabilistic risk methodologies and data from the National Bridge Inventory database. The aim is to simplify the implementation of risk-based ranking procedures into bridge management system packages according to the MAP-21 vision. Therefore, machine learning techniques are employed to facilitate the introduction of probabilistic risk methods into bridge management systems. The procedure is described for seven hazards that are pertinent to bridges in New Jersey: overloading, fatigue, seismic, flooding, scour, vehicle and vessel collision. Risk values are computed in monetary terms to homogenize the comparison among bridges for different hazards. The analysis is performed on 5,534 bridges, showing that seismic events and fatigue resulting from truck overloading are the most dominant hazards in New Jersey, for which about 97.0% and 29.0% of bridges show some level of risk. The main limitation of the proposed framework is the lack of accurate data from bridge inventories necessary to thoroughly perform a fully structural probabilistic analysis of bridges and to minimize engineering judgment.

Macroscopic effect of distance from seacoast on bridge deterioration – Statistical data assessment of structural condition recordings

Airborne sea chlorides that are transported inland by wind and deposited on structures’ surfaces are a principal deterioration factor for coastal bridges. In this paper, a dataset for almost 20,000 coastal bridges, which was extracted from the National Bridge Inventory (NBI) of the US Federal Highway Administration, is exploited to conduct a macroscopic statistical study on the deterioration effect of airborne chlorides. The study focuses directly at the very final stage of the overall physical phenomenon, in which deterioration due to seacoast effect is evident. To distinguish this effect, the extracted coastal bridge sample was segmented to produce consistent subsamples in terms of other coexisting factors influencing deterioration, including use of deicing salt and earthquake hazard. Year of bridge construction/reconstruction, structural materials (reinforced/prestressed concrete, structural steel) and structural types (simple/continuous spans) were also taken into account. Moreover, bridge groups based on the bridges’ distances from the seacoast were formed using high accuracy coastline and bridge coordinates. Hence, structural condition ratings for bridge components recorded in the NBI database were statistically assessed with respect to corresponding coastal distances. It was statistically deduced that, under certain conditions, bridges can be noticeably affected by sea chlorides at coastal distances up to 2–3 km inland.

Bridge condition rating data modeling using deep learning algorithm

This paper presents a deep learning-based bridge condition rating data modeling approach using selected data from the National Bridge Inventory (NBI) database. The objective of this research is to develop a data-driven approach that enables prediction of future conditions of highway bridge components from historical inspection data. The problem is solved by training a Convolutional Neural Network (CNN) model with online available NBI data. One prominent feature of the CNN model is that if well-trained it can represent the high dimensional data in the dataset abstractions for which conventional mathematical models may be difficult to describe. A case study of Maryland and Delaware highway bridges using historical data (1992–2017) sourced from the NBI database has been performed to demonstrate the proposed method. CNN models for three primary components of these highway bridges including the deck, superstructure, and substructure have been established. Optimization of model parameters is achieved through a parametric study. Research findings suggest that the deep learning model offers a promising tool as a data-driven condition forecasting approach for bridge components with a demonstrated prediction accuracy over 85%.

Multi-state deterioration prediction for infrastructure asset: Learning from uncertain data, knowledge and similar groups

Infrastructure assets such as bridges need to be inspected regularly for signs of deterioration. Although a fixed inspection interval could be used, an estimate of the rate of deterioration allows us to schedule the next inspection more cost-effectively. Our earlier work outlined a Bayesian framework that uses both data and knowledge to predict the transition between assets, which has been extended and realised in this paper for asset deterioration prediction. In the Bayesian model, censorship is modelled to incorporate the uncertainty from inspection records and prior of the parameter is used to express expert knowledge. In particular, we also suggest how the prior probabilities of the parameters of a Weibull distribution can be set in practice using expert estimates such as the maximum and average times of a transition from one state to another.Furthermore, assets with similar characteristics may deteriorate similarly. We propose to separate related assets into groups and learn deterioration between these groups. This assumption allows us to tackle the challenge of limited data further and is experimented with the deck inspection records from the National Bridge Inventory database in Wyoming. This database includes over 100 features of each bridge such as structure type and average daily traffic: we use a modified random forest to select a subset of important features to separate assets into groups. The model is extended into hierarchical Bayesian models to learn between groups with the help of hyper-parameters and an aggregated variable from the feature levels. Performance of our method is compared with other existing approaches from various aspects.

Structural information integration for predicting damages in bridges

Abstract Industrial Information Integration Engineering interacts with each of the twelve engineering disciplines defined by the U.S. National Academy of Engineering. Civil Engineering is one of the twelve engineering sections. Integration of information can be a very valuable method for civil engineering. In the management of the highway bridges, bridge design and operation parameters and inspection results are recorded according to national and state guides and the data is coded and archived in what is known as the National Bridge Inventory (NBI) database. The integration of the information in the NBI database can be a very valuable method for managing of bridges. The major challenges of the information in such a database are unbalance, complexity, subjectivity, and incompleteness, which make developing a damage prediction model from such information difficult. Particularly, the prediction and management of bridge abutment distress is difficult since their generation mechanism is not clear. The purpose of this study was to integrate the information of the inspections database and establish models to predict abutment distress. Input variables for the networks were selected from knowledge-based evaluations, statistical analyses, and trial and error. Four neural network models were developed and evaluated: the multilayer perceptron and support vector machine models predicted abutment condition better than a supervised self-organizing map and radial basis function networks.

Senior Community Resilience with a Focus on Critical Transportation Infrastructures: An Accessibility-Based Approach to Healthcare

The importance of bridges to mobility in transportation is well known. However, the identification of bridges that influence the mobility of senior members of communities has not been evaluated. This is imperative because of human frailties associated with aging. In this paper, senior community resilience is assessed through accessibility of seniors to hospitals after bridge damage caused by hurricane events using Pinellas County in the Tampa Bay area as case study. The paper presents: (i) exposure probabilities for hurricane events at bridge locations; (ii) bridge damage state functions and damage state rating assignments using historical data from the National Bridge Inventory database; (iii) identification of bridges at risk of hurricane-induced damage; (iv) bridges identified as serving areas (census districts) with dense population of aging people; and (v) estimated effects of bridge closures on mobility and resilience of the aged population, based on accessibility to hospitals by using congested and free-flow travel times obtained from traffic assignment modeling. Findings showed that: (i) 66 bridges prone to hurricane-induced damage were observed to affect 140 selected aging population areas; (ii) bridge closures resulted in about 15% and 75% increase in free-flow and congested travel times, respectively; (iii) complete loss of accessibility to hospitals for some aging-dense zones; and (iv) resilience indexes of 0.94 and 0.81 were computed for free-flow and congested travel times, respectively. These results highlight significant loss in senior accessibility to hospitals and emphasize the need for policy discussions on the capabilities of highway bridges for efficient senior mobility.

Bridge Type Classification: Supervised Learning on a Modified NBI Data Set

A key phase in the bridge design process is the selection of the structural system. Due to budget and time constraints, engineers typically rely on engineering judgment and prior experience when selecting a structural system, often considering a limited range of design alternatives. The objective of this study was to explore the suitability of supervised machine learning as a preliminary design aid that provides guidance to engineers with regards to the statistically optimal bridge type to choose, ultimately improving the likelihood of optimized design, design standardization, and reduced maintenance costs. In order to devise this supervised learning system, data for over 600,000 bridges from the National Bridge Inventory database were analyzed. Key attributes for determining the bridge structure type were identified through three feature selection techniques. Potentially useful attributes like seismic intensity and historic data on the cost of materials (steel and concrete) were then added from the US Geological Survey (USGS) database and Engineering News Record. Decision tree, Bayes network and Support Vector Machines were used for predicting the bridge design type. Due to state-to-state variations in material availability, material costs, and design codes, supervised learning models based on the complete data set did not yield favorable results. Supervised learning models were then trained and tested using 10-fold cross validation on data for each state. Inclusion of seismic data improved the model performance noticeably. The data was then resampled to reduce the bias of the models towards more common design types, and the supervised learning models thus constructed showed further improvements in performance. The average recall and precision for the state models was 88.6% and 88.0% using Decision Trees, 84.0% and 83.7% using Bayesian Networks, and 80.8% and 75.6% using SVM.

Load-Capacity Rating of Bridge Populations through Machine Learning: Application of Decision Trees and Random Forests

The functionality of the U.S. transportation infrastructure system is dependent upon the health of an aging network of over 600,000 bridges, and agencies responsible for maintaining these bridges rely on the process of load rating to assess the adequacy of individual structures. This paper presents a new approach for safety screening and load-capacity evaluation of large bridge populations that seeks to uncover heretofore unseen patterns within the National Bridge Inventory database and establish relationships between select bridge attributes and their load-capacity status. Decision-tree and random-forest classification models were trained on the national concrete slab bridge data set of over 40,000 structures. The resulting models were validated on an independent data set and then compared with a number of existing judgment-based schemes found in an extensive survey of the current state of practice in the United States. The proposed approach offers a method that provides guidance for improved allocation of resources by informing maintenance decisions through rapid identification of candidate bridges that require further scrutiny for either possible load restriction or restriction removal.

Impact of Overweight Trucks on the Service Life of Bridge Girders

State agencies are responsible for making major decisions on allocating the available but limited funds for the maintenance and rehabilitation of bridges. Over the past two decades, the frequency of overweight trucks has kept increasing. Although the AASHTO load and resistance factor design specifications mandate a design life of 75 years, the actual service life will be affected by site-specific loading; local environmental attack, including the application of deicing salts; each agency’s bridge preservation strategy; the funding situation; and routine maintenance. Nowadays, agencies issue more permits for trucks with gross vehicle weights that exceed legal load limits. There is a need, therefore, to quantify the impact of overweight trucks on the service life of bridge girders. This study presents a procedure for investigating the impact of truck loads on bridges in New Jersey through the utilization of bridge inspection reports, truck weigh-in-motion data, and the National Bridge Inventory database. Actual bridge deterioration modes in New Jersey were identified from the bridges’ respective inspection reports, and the expected bridge service life was successfully correlated with truck loading from weigh-in-motion data. The deterioration mode of prestressed concrete (PC) girders was found to be the corrosion of prestressing tendons near the beam ends induced by cracking and spalling of enclosing concrete. The maintenance of deck joints is a major factor that affects the deterioration of PC girders. Additionally, a strong correlation was found between the expected service life of PC girders and overweight truck counts.

CONFLATION OF NATIONAL BRIDGE INVENTORY DATABASE WITH TIGERBASED ROAD VECTORS

Abstract. The National Bridge Inventory (NBI) database provides detailed bridge information with full coverage in the United States. In the database, each bridge record has an associated geographic location, which makes it possible to match the NBI records with road vectors in a geospatial database. This paper presents our approach to conflating the NBI database with U.S. Census Bureau's Topologically Integrated Geographic Encoding and Referencing (TIGER) road vectors based on a point-to-line matching algorithm, which combines both geometric and semantic similarity measurements. Experiments have shown that most NBI bridges can be successfully matched to TIGER road vectors with an overall success rate over 80%. The matched results are then used to help define linear features for bridges in the road database. The resultant road databases have enhanced bridge information and are further used in creating a 3D road database. The proposed methodology has been fully implemented and has been used in various applications.

Estimated effects of climate change on flood vulnerability of U.S. bridges

We assessed the potential impacts of increased river flooding from climate change on bridges in the continental United States. Daily precipitation statistics from four climate models and three greenhouse gas (GHG) emissions scenarios (A2, A1B, and B1) were used to capture a range of potential changes in climate. Using changes in maximum daily precipitation, we estimated changes to the peak flow rates for the 100-year return period for 2,097 watersheds. These estimates were then combined with information from the National Bridge Inventory database to estimate changes to bridge scour vulnerability. The results indicate that there may be significant potential risks to bridges in the United States from increased precipitation intensities. Approximately 129,000 bridges were found to be currently deficient. Tens of thousands to more than 100,000 bridges could be vulnerable to increased river flows. Results by region vary considerably. In general, more bridges in eastern areas are vulnerable than those in western areas. The highest GHG emissions scenarios result in the largest number of bridges being at risk. The costs of adapting vulnerable bridges to avoid increased damage associated with climate change vary from approximately $140 to $250 billion through the 21st century. If these costs were spread out evenly over the century, the annual costs would be several billion dollars. The costs of protecting the bridges against climate change risks could be reduced by approximately 30% if existing deficient bridges are improved with riprap.

Using Soft Computing to Analyze Inspection Results for Bridge Evaluation and Management

The national bridge inventory (NBI) system, a database of visual inspection records that tallies the condition of bridge elements, is used by transportation agencies to manage the rehabilitation of the aging U.S. highway infrastructure. However, further use of the database to forecast degradation, and thus improve maintenance strategies, is limited due to its complexity, nonlinear relationship, unbalanced inspection records, subjectivity, and missing data. In this study, soft computing methods were applied to develop damage prediction models for bridge abutment walls using the NBI database. The methods were multilayer perceptron network, radial basis function network, support vector machine, supervised self-organizing map, fuzzy neural network, and ensembles of neural networks. An ensemble of neural networks with a novel data organization scheme and voting process was the most efficient model, identifying damage with an accuracy of 86%. Bridge deterioration curves were derived using the prediction models and compared with inspection data. The results show that well developed damage prediction models can be an asset for efficient rehabilitation management of existing bridges as well as for the design of new ones.

Identifying Critical Sources of Bridge Deterioration in Cold Regions through the Constructed Bridges in North Dakota

This paper presents the performance of constructed bridges in cold regions through examining the bridges in North Dakota that is one of the coldest regions in the United States. Unique approach of a combined multiple regression and geographic information system technology is employed to evaluate the performance and to identify the critical sources affecting deterioration of the 5,289 bridges sampled from the National Bridge Inventory database inspected between 2006 and 2007. Typical parameters examined include physical, material, and environmental factors associated with the existing bridges. The importance of maintenance and repair is highlighted. Traffic volume significantly influences the level of deterioration of the bridge decks. Year built is the most significant contribution to the structural deficiency of the bridges, followed by structural characteristics and traffic volumes. The presence of water particularly influences the deterioration. Concrete bridges are more durable than steel bridges. Tru...

Deficient Bridge Mobility Performance Measures

The structural and functional performances of bridges affect their safety, mobility, and life span. The concepts of structural deficiency and functional obsolescence of bridges are essential engineering and management considerations with public and economic implications. This article discusses management aspects of deficient bridges. National bridge inventory database with conditional data and mobility measures were studied to compile practical data for more dependable decision-making. A forecast was developed to predict the number of structurally deficient and functionally obsolete bridges in the future. To reach a systematic conceptualization of deficiency, areas of bridges were also discussed as compared to the count of bridges. The resulting study is practical for transportation agencies, officials, and researchers. The presented result supports better bridge management practices by better planned policies, financial resources, and strategies.