Abstract
Today, transportation agencies face aging infrastructure, increasing traffic demands and funding limitations, making efficient allocation of maintenance, rehabilitation, and replacement (MR&R) resources increasingly difficult. Many State Departments of Transportation (DOTs) use asset management principles, such as deterioration modeling and future condition prediction, to manage and schedule maintenance for thousands of bridges in an inventory or network. Deterioration modeling is an essential bridge management tool for predicting future condition and for helping allocate MR&R funds. However, existing deterioration modeling methods fail to account for interaction between bridge elements, particularly subordinate deterioration.
Subordinate deterioration occurs when element deterioration is impacted by the deterioration of a separate element, for example, bridge joints affecting the condition of bearings, pier caps, beam ends and abutments. Bridge engineers recognize subordinate deterioration exists for certain bridge elements, but it is ignored in current deterioration modeling. This study provides an investigation into the effect of deck joint deterioration on the deterioration of steel bearings and reinforced concrete pier caps for bridges in the Commonwealth of Virginia. However, the techniques developed in this research are general and can be applied to investigate the subordinate interaction of other elements.
State DOTs are required by federal law to inspect and assess the condition of bridges by visual inspection. This is captured through element-level inspections, where the condition of individual members and components of a bridge are rated on a numeric scale based on condition definitions. Element-level data from Virginia’s inspection database were used to develop datasets of bridge inspection reports with steel bearing and pier cap elements. First, exploratory statistical analysis, using categorical data methods, was conducted to determine the significance of subordinate deterioration. The investigation then explored different statistical models to predict the condition of the subordinate element. Finally, this research proposed a method to develop the transition probability matrices for elements that have a subordinate relationship to be used in Markov Chain deterioration modeling.
In this research, a statistically significant association between joints and condition of subordinate elements was found. This showed that subordinate deterioration existed in the inspection data and could be incorporated into bridge management practices. Multi-category logistic regression models were developed but failed the global goodness-of-fit test, suggesting the models did not accurately reflect the inspection data. The proposed Markov Chain approach to incorporate subordinate deterioration provided useful results and was calibrated using minimization of the squared error and the goodness-of-fit test.
