| Bridge management is an important activity of transportation agencies in the US and in many other countries. A critical aspect of bridge management is to reliably predict the deterioration of bridge structures, so that appropriate or optimal actions can be selected to reduce or minimize the deterioration rate and maximize the effect of spending for replacement or maintenance, repair, and rehabilitation (MR&R). In the US, Pontis is the most popular bridge management system used among the state transportation agencies. Its deterioration model uses the Homogeneous Markov Chain, with a statistical regression to estimate the required transition probabilities. This method works under the assumption that deterioration is constant over time.;The limitations of the Pontis method is studied based on which Non-Homogenous Markov Chain methodology is proposed in this research. This method works with the assumption of deterioration varying according to age, which is more reasonable. Two Optimization based models, one Non-Linear and the other Linear have been proposed for effective prediction of bridge condition states.;These models have been applied to Michigan Department of Transportation (MDOT) data, and the results compared to the Pontis method. Results show that both these models have performed much better than the Pontis approach there by proving its credibility. Then these models have been applied to the National Bridge Inventory (NBI) data, where it is found to be reliable too. Finally a simple Optimal Policy Model is presented which would aid in finding the best possible MR&R policy. |
