Management of bridges with shallow foundations exposed to local scour

Abstract

The Serbian road network includes a large portion of bridges with shallow foundations vulnerable to local scour as tragically demonstrated during the extreme flooding in May 2014. Currently, the bridge management procedures in Serbia and worldwide do not comprehensively account for a risk of bridge failure due to flooding and fail to provide sufficient information for the decision-making. Thus, a novel methodology for quantitative vulnerability assessment is suggested as a tool to identify the most vulnerable bridges in a network. Herein, the essential task is evaluation of the conditional probability of a bridge failure due to local scour in a flooding event of a certain magnitude. To apply this approach on a network level, there is a dire need to establish precise practice-ready guidelines on an optimal set of information to be used and/or collected in situ, which is discussed on an example of the Serbian bridge database. The vulnerability of a bridge to local scour may be used as a comprehensive indicator of a bridge performance in a flooding event. For a network level, the vulnerability maps with respect to flooding of different magnitudes will give road operators crucial information to apply adequate quality control plans to vulnerable bridges.     

A risk ranking strategy for network level bridge management

This paper presents the development of a qualitative risk ranking strategy for characterising a network of bridges into groups with similar risk levels, which can form the basis for developing a risk based inspection regime over a bridge network. The factors affecting risk are identified and rationally combined to present various attributes of bridges. A qualitative scoring system is then introduced that uses the attributes to rank bridges in terms of their relative risk. Sensitivity analysis is performed to quantify the effect of relative weights of the attributes on the risk scores. The methodology is demonstrated through its application on the UK's Network Rail bridge stock and a random sample of bridges is ranked according to the proposed method. One of the factors used to evaluate the risk scores is comparable with the bridge condition index. A reasonable agreement between them is observed within the sample bridges. The proposed method will be beneficial to bridge owners in identifying the bridges according to their risk levels in a simple and systematic manner. Inspections and interventions can be planned based on this risk ranking strategy to maintain a consistent risk level across the network.     

Bridge residual service-life prediction through Bayesian visual inspection and data updating

The paper describes an integrated procedure for evaluation of the service-life curve of existing bridges, calibrated with data derived from visual inspections and aimed at assessing the degree of deterioration of their structural/non-structural elements. Visual inspections allowed a total sufficiency rating for the bridges to be calculated, together with data collection, to be used in Bayesian updating of element deterioration curves. The results were combined with information from simulations of potential bridge deterioration scenarios to quantify mean trends and related uncertainties of the remaining service-life curves of existing bridges. The proposed procedure was applied to two bridge stocks and results are critically discussed. The method used may allow public authorities/private managing companies to optimise economic resources allocated to long/medium-term maintenance plans for the most critical structures.     

Highway and Road Probabilistic Safety Assessment Based on Bayesian Network Models

A Bayesian network model is developed, in which all the items or safety related elements encountered when traveling along a highway or road, such as terrain, infrastructure, light signals, speed limit signs, intersections, roundabouts, curves, tunnels, viaducts, and any other safety relevant elements are reproduced. Since human error is the main cause of accidents, special attention is given to modeling the driver behavior variables (driver's tiredness and attention) and to how they evolve with time or travel length. The sets of conditional probabilities of variables given their parents, which permit to quantify the Bayesian network joint probability, are obtained and written as closed formulas, which allow us to identify the particular contribution of each variable to safety and facilitate the computer implementation of the proposed method. In particular, the probabilities of incidents affecting safety are calculated so that a probabilistic safety assessment of the road can be done and its most critical elements can be identified and sorted by importance. This permits the improvement of road safety making adequate corrections to save time and money in the maintenance program by concentrating on the most critical elements and effective investments. Some real examples of a Spanish highway and a conventional road are provided to illustrate the proposed methodology and show its advantages and performance.     

A model for the evaluation of intervention strategies for bridges affected by manifest and latent deterioration processes

Markov models are often used in bridge management systems to evaluate intervention strategies (ISs) for bridges affected by manifest deterioration processes (MnDPs). These models do not directly take into consideration the effect of latent deterioration processes (LtDPs) on the object, i.e. the deterioration that might occur due to natural hazards (e.g. earthquakes and floods). In cases where there is a negligible probability of the occurrence of natural hazards, this is justified, otherwise it is not. In this paper, a model is proposed that can be used to evaluate ISs for bridge elements and bridges considering both MnDPs and LtDPs. The model is an extension of the Markov models, and includes condition states (CSs) that can occur due to both MnDPs and LtDPs, as well as the probabilities of transition (p.o.ts) between them. The contributions to the p.o.ts due to MnDPs are initially estimated using well-established methods and adjusted for the contributions to the p.o.ts due to LtDPs, which are estimated using fragility curves and adjusted considering element dependencies, i.e. how the elements of a bridge work together. The use of the model is demonstrated by predicting the future CSs of a bridge affected by both MnDPs and LtDPs.     

Availability function as bridge element’s importance weight in computing overall bridge health index

Abstract

Extensive research has been conducted over the past two decades to evaluate the condition and bridge health index (BHI) of bridges using different performance measures. The International Association for Bridge Management and Safety (IABMAS) has stressed on the growing need to assess the condition state of the bridges by integrating new functionalities and methodologies into bridge management systems (BMS) for better monitoring outcomes. This article presents the development of bridge element importance weight using availability metric computed from element failure time distribution parameters and repair rates. This study investigates the following issues: the availability of bridge elements and subsystems, the evaluation of the criticality rank of bridge components and subsystems using failure and downtime criticality indices and the effect of deterioration of critical elements on the overall operability of the bridge system. The availability index and failure criticality output of selected elements showed that superstructure elements, reinforced concrete column, metal bearings, piling, pile cap/footing and reinforced concrete cap were of more importance (critical) to the bridge system deterioration compared to decks, railings and expansion joints. At the component level, it was observed that movable bridge elements (hydraulics, operator facilities, etc.) were less available based on their downtimes and operational characteristics.     

Modelling of bridge elements deterioration for Serbian bridge inventory

In order to improve bridge management practice for public roads in Serbia, a deterioration model for bridge elements was developed using condition data collected over the last 20 years. The distribution of condition data over condition states allows the estimation of reasonably reliable deterioration models even for advanced deterioration. Based on the literature review, discrete-time Markov chains were chosen for this purpose. Estimating the transition probabilities of discrete Markov chains is straightforward when the time between condition records on bridge elements closely matches the chosen time interval for the Markov chain, which is not the case with condition data in Serbia. To overcome this problem, an expectation maximisation (EM) algorithm has been applied to estimate the transition probabilities. It is shown here that the EM algorithm is a sound and robust method, which can yield a reasonable deterioration model even if limited data are available. In addition, the EM is used to determine the overall bridge deterioration using existing agency rules to derive bridge rating from element condition states.     

Bayesian Networks‐Based Probabilistic Safety Analysis for Railway Lines

A Bayesian network model is developed, in which all the items or elements encountered when travelling a railway line, such as terrain, infrastructure, light signals, speed limit signs, curves, switches, tunnels, viaducts, rolling stock, and any other element related to its safety are reproduced. Due to the importance of human error in safety, especial attention is given to modeling the driver behavior variables and their time evolution. The sets of conditional probabilities of variables given their parents, which permits quantifying the Bayesian network joint probability, are given by means of closed formulas, which allow us to identify the particular contribution of each variable and facilitate a sensitivity analysis. The probabilities of incidents affecting safety are calculated so that a probabilistic safety assessment of the line can be done and its most critical elements can be identified and sorted by importance. This permits improving the line safety and saving time and money in the maintenance program by concentrating on the most critical elements. To reduce the complexity of the problem, an original method is given that permits dividing the Bayesian network in to small parts such that the complexity of the problem becomes linear in the number of items and subnetworks. This is crucial to deal with real lines in which the number of variables can be measured in thousands. In addition, when an accident occurs the Bayesian network allows us to identify its causes by means of a backward inference process. The case of the real Palencia–Santander line is commented on and some examples of how the model works are discussed.     

Evaluation of liquefaction-induced lateral displacement using Bayesian belief networks

Liquefaction-induced lateral displacement is responsible for considerable damage to engineered structures during major earthquakes. Therefore, an accurate estimation of lateral displacement in liquefaction-prone regions is an essential task for geotechnical experts for sustainable development. This paper presents a novel probabilistic framework for evaluating liquefaction-induced lateral displacement using the Bayesian belief network (BBN) approach based on an interpretive structural modeling technique. The BBN models are trained and tested using a wide-range case-history records database. The two BBN models are proposed to predict lateral displacements for free-face and sloping ground conditions. The predictive performance results of the proposed BBN models are compared with those of frequently used multiple linear regression and genetic programming models. The results reveal that the BBN models are able to learn complex relationships between lateral displacement and its influencing factors as cause–effect relationships, with reasonable precision. This study also presents a sensitivity analysis to evaluate the impacts of input factors on the lateral displacement.     

Characterising the effect of external factors on deterioration rates of bridge components using multivariate proportional hazards regression

Within asset management of infrastructure systems, increases in maintenance needs subject to budgetary constraints have motivated the development of tools to forecast deterioration to optimise maintenance intervention. Current bridge deterioration modelling approaches, including the evolving duration-based methods, routinely rely on a priori categorisation of bridges based on design, functional, and geographic factors to account for their effects on deterioration rates. However, such preclassification is often based on engineering judgement and may not reflect the true influence of these explanatory factors. In the current study, a proportional hazards regression-based methodology was developed to identify the most critical factors affecting deterioration using the entire unsegmented bridge database. The framework designed to perform this duration-based regression on large bridge databases is presented in this paper and results from implementation on a state inventory of over 17,000 bridges are discussed. The results provide insight into the extent that explanatory factors influence deterioration rates of different bridge components. A novel aspect of the developed framework is its ability to analyse the time-dependent effects of explanatory factors on deterioration rates over the lifecycle of the structural components. This analysis can be used to develop multivariate deterioration models and inform decision-making and prioritisation strategies.     

Predicting the probability of failure of timber bridges using fault tree analysis

The resilience of a community in an extreme event depends mainly on the robustness of the critical infrastructures. Road bridges are a critical link of the road network, which plays a focal role in Australia’s economy, prosperity, social well-being and quality of life. Timber bridges are a weaker link of the Australian road network and they often provide critical access to the rural communities. This research uses a number of bridge inspection reports to develop a method to predict the probability of failure of a timber bridge. The inspected condition states of the elements in the timber bridge are used to develop a Markov chain based model and Gamma process model to predict the deterioration of each element. The probability of condition state movement for each element thus calculated were used in fault tree analysis to estimate likelihood of failure of a bridge in a given time period. Although the developed method is based on limited data and it has several limitations, model can be further refined with the availability of more inspection reports. The method developed is demonstrated using an inspection report for a timber bridge, which was not used in the development of the models.     

Life-Cycle Performance of RC Bridges: Probabilistic Approach

This article addresses the problem of reliability assessment of reinforced concrete (RC) bridges during their service life. First, a probabilistic model for assessment of time-dependent reliability of RC bridges is presented, with particular emphasis placed on deterioration of bridges due to corrosion of reinforcing steel. The model takes into account uncertainties associated with materials properties, bridge dimensions, loads, and corrosion initiation and propagation. Time-dependent reliabilities are considered for ultimate and serviceability limit states. Examples illustrate the application of the model. Second, updating of predictive probabilistic models using site-specific data is considered. Bayesian statistical theory that provides a mathematical basis for such updating is outlined briefly, and its implementation for the updating of information about bridge properties using inspection data is described in more detail. An example illustrates the effect of this updating on bridge reliability.     

Development of hybrid optimisation method for Artificial Intelligence based bridge deterioration model — Feasibility study

Bridge Management Systems (BMSs) are a common tool for bridge management to extend the life cycle of bridge networks. However, the reliability of current BMS outcomes is doubtful. This is because: (1) Overall Condition Rating (OCR) method cannot represent individual bridge elements’ condition and is unable to represent condition ratings of bridge elements in lower Condition States and due to (2) insufficient historical bridge records available. A long-term Performance Bridge (LTPB), i.e. deterioration, model is the most crucial component and decides level of reliability of long-term bridge needs. Recent development of an AI-based bridge deterioration model was undertaken to minimise these shortcomings. However, this model is computationally costly due to the process of Neural Network, generating a large data output. To improve the neural network process, optimisation is required. The hybrid optimisation method is proposed in this paper to filter out feasible condition ratings as input for long-term prediction modelling.     

State transition probabilities in bridge deterioration based on Weibull sojourn times

This paper presents an investigation of the Markov property underlying the stochastic deterioration models for highway bridges, including transition probabilities between the condition states. Using historical data of sojourn times for the ‘decay’ (no improvement intervention) deterioration, hazard functions were developed and ‘instantaneous’ 1 year transition probabilities estimated for the sojourn times in the condition states, for various bridge categories, by type of material and roadway carried. The rate of transition out of each state was found to be not constant relative to time, as assumed for Markov chain models, but rather, increasing with the time spent in the state. Best-fit distributions of the sojourn times were determined to not be exponential (Markov chain), but Weibull, with the parameters established using the maximum likelihood estimate (MLE) method. A semi-Markov model of the bridge deterioration process was formulated and developed, including kernels of transition probabilities and time-based matrices of multi-step transition probability functions.     

Modelling the spatial distribution of heavy vehicle loads on long-span bridges based on undirected graphical model

Abstract

Vehicle load modelling is highly important for bridge design and safety evaluation. Conventional modelling approaches for vehicle loads have limitations in characterizing the spatial distribution of vehicles. This article presents a probabilistic method for modelling the spatial distribution of heavy vehicle loads on long-span bridges by using the undirected graphical model (UGM). The bridge deck is divided into grid cells, a UGM with each node corresponding to each cell is employed to model the location distribution of heavy vehicles, by which probabilities of heavy-vehicle distribution patterns can be efficiently calculated through applying the junction tree algorithm. A Bayesian inference method is also developed for updating the location model in consideration of the non-stationarity of traffic process. Gross weights of heavy vehicles are modelled by incorporating additional random variables to the vehicle-location UGM, corresponding probability distributions are constructed conditioned on ignoring correlation and considering correlation, respectively. Case studies using simulated data as well as field monitoring data have been conducted to examine the method. Compared with previous studies involving vehicle load modelling, the presented method can implement probabilistic analysis for all spatial distribution patterns of heavy vehicles on the entire bridge deck.     

Structural health monitoring of a cable-stayed bridge with Bayesian neural networks

In recent years, there has been an increasing interest in permanent observation of the dynamic behaviour of bridges for long-term monitoring purpose. This is due not only to the ageing of a lot of structures, but also for dealing with the increasing complexity of new bridges. The long-term monitoring of bridges produces a huge quantity of data that need to be effectively processed. For this purpose, there has been a growing interest on the application of soft computing methods. In particular, this work deals with the applicability of Bayesian neural networks for the identification of damage of a cable-stayed bridge. The selected structure is a real bridge proposed as benchmark problem by the Asian-Pacific Network of Centers for Research in Smart Structure Technology (ANCRiSST). They shared data coming from the long-term monitoring of the bridge with the structural health monitoring community in order to assess the current progress on damage detection and identification methods with a full-scale example. The data set includes vibration data before and after the bridge was damaged, so they are useful for testing new approaches for damage detection. In the first part of the paper, the Bayesian neural network model is discussed; then in the second part, a Bayesian neural network procedure for damage detection has been tested. The proposed method is able to detect anomalies on the behaviour of the structure, which can be related to the presence of damage. In order to obtain a confirmation of the obtained results, in the last part of the paper, they are compared with those obtained by using a traditional approach for vibration-based structural identification.     

A step forward towards a comprehensive framework for assessing liquefaction land damage vulnerability: Exploration from historical data

The unprecedented liquefaction-related land damage during earthquakes has highlighted the need to develop a model that better interprets the liquefaction land damage vulnerability (LLDV) when determining whether liquefaction is likely to cause damage at the ground’s surface. This paper presents the development of a novel comprehensive framework based on select case history records of cone penetration tests using a Bayesian belief network (BBN) methodology to assess seismic soil liquefaction and liquefaction land damage potentials in one model. The BBN-based LLDV model is developed by integrating multi-related factors of seismic soil liquefaction and its induced hazards using a machine learning (ML) algorithm-K2 and domain knowledge (DK) data fusion methodology. Compared with the C4.5 decision tree-J48 model, naive Bayesian (NB) classifier, and BBN-K2 ML prediction methods in terms of overall accuracy and the Cohen’s kappa coefficient, the proposed BBN K2 and DK model has a better performance and provides a substitutive novel LLDV framework for characterizing the vulnerability of land to liquefaction-induced damage. The proposed model not only predicts quantitatively the seismic soil liquefaction potential and its ground damage potential probability but can also identify the main reasons and fault-finding state combinations, and the results are likely to assist in decisions on seismic risk mitigation measures for sustainable development. The proposed model is simple to perform in practice and provides a step toward a more sophisticated liquefaction risk assessment modeling. This study also interprets the BBN model sensitivity analysis and most probable explanation of seismic soil liquefied sites based on an engineering point of view.     

Managing Board

Abstract

While physical models of concrete structures may give a fundamental understanding of structural deterioration mechanisms, their application is more or less limited in reflecting actual scenarios as a result of inevitable theoretical assumptions, whereby it is often difficult to cover complex structural behaviour and environment in the real world. Uncertainties and fuzziness, along with complexity, add more difficulty to the estimation of structural deterioration by physical models, which are typically described by quantitative mathematical equations. To complement physical models, an approach based on a statistical causal relationship, combining logics and statistics, is established for modelling structural condition deterioration. The relationship includes three kinds of basic variables in terms of causal factors, consequences and indicators, and it is represented by a set of conditional probabilities among those variables. Level two inspection data on concrete bridge slabs, as an example, is applied to establish the relationship. On the basis of the probabilistic causal relationship, prediction and diagnosis of structural conditions are implemented. To take advantage of this approach in dealing with qualitative issues such as human factors, the influence of inspectors' judgement on structural condition rating is addressed, while the effects of weather conditions on inspectors' judgement are also illustrated.