A methodology to model the variability in pavement performance

Abstract

This paper presents a methodology for modelling pavement resilience and provides an insight into the reasons why the prediction of pavement performance may not be accurate by analysing the variability of the parameters used to express pavement performance. The study identifies basic statistical properties that describe the underlying variability of the types of data considered in the development of the pavement performance model form. Thereafter the paper gives a methodology that (a) uses the Monte Carlo data generation technique to develop appropriate performance models and (b) enables the effects of the variability of the model parameters on the performance predicted to the analysed. The analysis of the results shows that, for the data sets examined, the variability of traffic loading significantly influences the pavement performance, the variability of pavement strength is less important, whilst the initial pavement roughness has an insignificant effect on pavement performance. The predictions of performance made by the models developed are assessed quantitatively and it is shown that they are in close agreement with the progression of historical data collected from in-service pavements for both high and low levels of data variability.