Systematic Statistical Approach to Populate Missing Performance Data in Pavement Management Systems

Abstract

Transportation agencies use pavement management systems (PMS) for their maintenance and rehabilitation planning, programming, and budgeting. PMS is used to make decisions regarding when maintenance and rehabilitation should be applied. To support these decisions, it is important to have reliable data on pavement conditions and accurate performance models for predicting pavement condition. The data on pavement condition typically come from regular field surveys resulting in distress, condition, and ride scores. PMS data sets are often incomplete (for some locations and some years) as a result of operational limitations reducing the predictive power of the performance models. Model-free and model-based replacement techniques for estimating missing data points have been designed and successfully used in other application areas like statistics, economics, marketing, medicine, psychometrics, and political science. It is therefore reasonable to apply these methods to the PMS databases. Statistical techniques are assembled and used in a robust approach to systematically analyze the effect of applying these techniques to rebuild missing performance data. As a case study, continuous reinforced concrete pavement (CRCP) sections were selected to test the proposed statistical systematic approach from a pavement management information system (PMIS) maintained by the Texas Department of Transportation (TxDOT). A major effect was observed in the results of predicting the distress scores when applying the developed approach.