This research documents the development and evaluation of artificial neural network (ANN) models to predict the condition ratings of concrete highway bridge decks in Michigan. Historical condition assessments chronicled in the national bridge inventory (NBI) database were used to develop the ANN models. Two types of artificial neural networks, multi-layer perceptrons and ensembles of neural networks (ENNs), were developed and their performance was evaluated by comparing them against recorded field inspections and using statistical methods. The MLP and ENN models had an average predictive capability across all ratings of 83% and 85%,respectively, when allowed a variance equal to bridge inspectors. A method to extract the influence of parameters from the ANN models was implemented and the results are consistent with the expectations from engineering judgment. An approach for generalizing the neural networks for a population of bridges was developed and compared with Markov chain methods. Thus, the developed ANN models allow modeling of bridge deck deterioration at the project (i.e., a specific existing or new bridge) and system/network levels. Further, the generalized ANN degradation curves provided a more detailed degradation profile than what can be generated using Markov models. A bridge management system (BMS) that optimizes the allocation of repair and maintenance funds for a network of bridges is proposed. The BMS uses a genetic algorithm and the trained ENN models to predict bridge deck degradation. Employing the proposed BMS leads to the selection of optimal bridge repair strategies to protect valuable infrastructure assets while satisfying budgetary constraints. A program for deck degradation modeling based on trained ENN models was developed as part of this project.

Bridges are at the heart of transportation systems connecting the roads to and between the mainlands. Thus, bridges are an integral part of the economic growth of any country. They are subjected to dynamic loads of the vehicles and the environmental effects. These loads cause stress and strain cycles causing its deterioration by initiating microcracking. The deterioration is then accelerated due to the chloride attack which causes the corrosion of the steel reinforcement resulting in cracking and delamination of concrete and ultimately leads to failure. It is essential to analyze the bridge with its actual condition which is difficult with a visual inspection. This analysis can help in determining the degree of repairs needed and an approximate idea about its service life. The development of the Non-Destructive Test (NDT) methods helps assess the condition of the bridge without any kind of damage to the original structure. In the past few decades, the Non-Destructive Evaluation (NDE) with the help of Ground Penetration Radar (GPR) has gained popularity due to its ease in the evaluation of the larger areas such as bridge deck and parking lot in a shorter amount of time with sufficient training. The NDE using GPR for Structural Health Monitoring (SHM) has been still evolving with new improvements in its technology as well as the development of new methods for the analysis of its data. A positive step towards detecting the subsurface materials present in the cracks has been undertaken in this study. A methodology to detect the subsurface cracks/gaps in concrete using GPR has been developed here by preparing three concrete samples of dimensions 50 x 25 x 5 cm3, 50 x 25 x 10 cm3, and 50 x 25 x 20 cm3 in the laboratory. The detection of reinforcement of 6 mm, 10 mm, 18 mm, 20 mm diameter, as well as a 21.8 mm Fiber Reinforcement Polymer (FRP) bar, are studied along with the detection of the air gap, water gap, and gap with the salt solutions of thickness 3 mm, 4 mm, 4.8 mm, 5.8 mm and 8.8 mm under the depth of 5 cm, 10 cm, and 15 cm. The amplitude values of these parameters are studied, and a comparison is made to check the ability of GPR to detect this material in cracks and/or delamination with changes in depths. This will be helpful in analyzing the GPR data with more reliability. Along with this, a non-linear finite element model (FEM) of a bridge superstructure using a fiber element is developed. The FE model of the bridge deck is updated and analyzed using a GPR defect map. This procedure of model updating is less tedious than the previous method available in the literature and proves to be time-saving. This model updating procedure will prove to be helpful in estimating the capacity of the bridge and make a prediction for future deterioration with the help of NDE methods (here GPR).

An effective bridge management system that is equipped with reliable deterioration models enables agency engineers to carry out monitoring and long-term programming of bridge repair actions. At the project level, deterioration models help the agency to track the physical condition of bridge elements and to specify when bridge maintenance, rehabilitation and replacement can be expected. Also, with reliable deterioration models, the agency can customize bridge repair or replacement schedules that incorporate element condition, functional obsolescence, and pre-specified performance thresholds. At the network level, component-specific deterioration models are useful for system-wide needs assessment over a specified future time horizon, and to quantifying the system-wide consequences of funding shortfalls or funding increases in terms of specified performance measures including average values of bridge condition and remaining service life. The bridge deterioration models that are currently in use in the Indiana Bridge Management System were developed over two decades ago. Since then, significant changes have taken place in inspection methods, technologies used, advanced statistical tools for data analysis. Also, because of the lack of reliable data, such items as the truck traffic and climate conditions were not included in past modeling efforts. In recent years, these obstacles have been minimized and therefore, there is an opportunity to update the deterioration models for the various bridge components. In addressing this research need, the present study developed families of curves representing deterioration models for bridge deck, superstructure, and the substructure. The National Bridge Inventory database was used, and the models use the NBI condition ratings as the response variable. The model families were categorized by administrative region, functional class, and superstructure material type. The explanatory variables include traffic volume and truck traffic, design type, and climatic condition, and design features. Deterministic and probabilistic models were developed.

The mathematical verification of the safety of structures can be done by determining the probability of failure or by using safety elements. Observed damages and collapses are usually assessed within the framework of expert reports, which seems reasonable due to the large number of unique structures in the construction industry. However, there should also be an examination of observed safety across all structures. Therefore, in this book the collapse frequencies are determined for different types of structures, such as bridges, dams, tunnels, retaining structures and buildings. The collapse frequency, like the failure probability, belongs to stochasticity. Therefore, the observed mean collapse frequencies and the calculated mean failure probabilities are compared. This comparison shows that the collapse frequencies are usually lower than the calculated failure probabilities. In addition, core damage frequencies and probabilities are given to extend the comparison to another technical product.

The Virginia Department of Transportation is deeply committed to the development and implementation of an efficient, cost-effective maintenance management system for its bridges. Much effort is being applied towards the development of a management system that will ensure that appropriate maintenance takes place at the optimum times. Within such systems, the ability to anticipate with reasonable accuracy how rapidly and in what fashion bridges will deteriorate is essential in optimizing expenditures of limited maintenance funds. The research described in this report was undertaken to provide this predictive capability. Specifically, the objective was to use existing bridge inspection data in conjunction with multiple regression analyses to develop models relating the rate of deterioration of structural components with variables such as age, loadings, and environmental factors and to evaluate the relative importance of these variables. In addition, these efforts include a discussion of current bridge management activities and recommendations on needed modifications to the VDOT record-keeping system.