Investigating and improving bridge management system methodologies under uncertainty

This dissertation presents a novel procedure to select explanatory variables, without the influence of human bias, for deterioration model development using National Bridge Inventory (NBI) data. Using NBI information, including geometric data and climate information, candidate explanatory variables can be converted into normalized numeric values and analyzed prior to the development of deterministic or stochastic deterioration models. The prevailing approach for explanatory variable selection is to use expert opinion solicited from experienced engineers. This may introduce human influenced biases into the deterioration modeling process. A framework using Least Absolute Shrinkage and Selection Operator (LASSO) penalized regression and covariance analysis are combined to compensate for this potential bias. Additionally, the cross validation analysis and solution path is used as a standard for the selection of minimum number of explanatory variables. The proposed method is demonstrated through the creation of deterministic deterioration models for deck, superstructure, and substructure for Wyoming bridges and compared to explanatory variables using the expert selection method. The comparison shows a significant decrease in error using the presented framework based on the L2 relative error norm. The final chapter presents a new method to develop stochastic deterioration models using logistic regression. The relative importance amongst explanatory variables is used to develop a classification tree for Wyoming bridges. The bridges in a subset are commonly associated with several explanatory variables, so that the deterioration models can be more representative and accurate than using a single explanatory variable. The logistic regression is used to introduce the stochastic contribution into the deterioration models. In order to avoid missing data problems, the binary categories condition rating, either remaining the same or decreased, are considered for logistic regression. The probability of changes in bridges’ condition rating is obtained and the averages for same condition ratings are used to create transition probability matrix for each age group. The deterioration model based on Markov chain are developed for Wyoming bridges and compared with the previous model based on percentage prediction and optimization approach. The prediction error is analyzed, which demonstrates the considerable performance of the proposed method and is suitable for relatively small data samples.