Measuring the Effect of Uncertainty in Unit Cost and Pre-Treatment Condition on Pavement Maintenance and Rehabilitation Decisions




Journal Title

Journal ISSN

Volume Title



A pavement maintenance and rehabilitation (M&R) project normally extends over 2-10 mile long roadway segment. At the M&R planning stage, these projects are called pavement management sections, which are often comprised of multiple data collection sections. The fact that a management section is comprised of multiple data collection sections introduces variability into the condition of the pavement within each M&R project. Also, variability is often found in the cost of M&R projects of the same M&R type. These variability are poorly understood and qualified in the pavement management literature. Accounting for these uncertainties in pre-treatment pavement condition and in the M&R treatment cost is essential for obtaining realistic estimate of needed funding. This research addresses this knowledge gap by a) developing probability density functions for pavement pre-treatment condition indicators and M&R unit cost, and b) developing a novel probabilistic methodology for estimating M&R funding needs for pavement networks that accounts for these uncertainties.

Data was obtained from the Bryan district pavement management plan for 2012 and from the Texas Department of Transportation (TxDOT) Pavement Management Information System (PMIS). Probability distribution functions were fitted for distress score, ride score, and unit cost using the @Risk software. Also, a simplified decision tree was developed to help simulate the maintenance and rehabilitation treatment selection process. This decision tree considers ride score, distress score, and traffic volume. After fitting the probability distributions of pavement condition indicators and unit cost, the impact of uncertainty in them on funding needs estimate was investigated using Monte Carlo simulation, The analysis shows that the needs estimate produced by TxDOT for the studied projects falls within the 90 percent confidence interval of the simulated need estimate.