The Pavement Performance Modeling: Deterministic vs. Stochastic Approaches

Abstract

The pavement performance modeling is an essential part of pavement management system (PMS). It estimates the long-range investment requirement and the consequences of budget allocation for maintenance treatments of a particular road segment on the future pavement condition. The performance models are also applied for life-cycle economic evaluation and for the prioritization of pavement maintenance treatments. This chapter discusses various deterministic and stochastic approaches for calculating the pavement performance curves. The deterministic models include primary response, structural performance, functional performance, and damage models. The deterministic models may predict inappropriate pavement deterioration curves because of uncertain pavement behavior under fluctuating traffic loads and measurement errors. The stochastic performance models assume the steady-state probabilities and cannot consider the condition and budget constraints simultaneously for the PMS. This study discusses the Backpropagation Artificial Neural Network (BPN) method with generalized delta rule (GDR) learning algorithm to offset the statistical error of the pavement performance modeling. This study also argues for the application of reliability analyses dealing with the randomness of pavement condition and traffic data.