Stochastic regression deterioration models for superstructure of prestressed concrete bridges in California

Abstract

At the beginning of 2018 about 6% of California’s bridges are structurally deficient, and approximately 17% of California’s bridges are estimated to cost about $12.2 billion for repairs. The subjectivity in determining the condition rating is an imprecise process and may significantly affect the maintenance process, which may vary from inspector to another. Most research works in prestressed concrete bridges condition rating has focused predominantly on modeling and has neglected to study the effect of non-periodical maintenance on condition rating. This study aims to identify the variables affecting superstructure deterioration and build models for predicting the superstructure condition. This paper has used National Bridge Inventory for California State in order to build models for predicting the superstructure condition of four structure types (Slab; Stringer/Multi Beam or Girder; T-Beam; and Box Beam or Girder) using Regression technique and Monte Carlo simulation. This research shows the impact of eight significant variables on the superstructure deterioration with high coefficient of determination (R2 = 86%). The developed models have been validated with a satisfactory result “93%” using Average Validity Percentage method. The developed models will help departments of transportation and infrastructure agencies to predict the condition rating and priorities the maintenance process for bridges..