Prediction models are one of the vital components in pavement management systems. Along with decision trees, the prediction models are ultimately used to set the priorities for maintenance planning and budget allocations at the network level. Within a pavement management system, raw performance data is often converted to aggregated performance indices, such as a riding comfort index to quantify the road roughness and a surface distress index to quantify the extent and severity of surface distresses. A historical database with pavement performance indices can be used as a source for the development of prediction models. However, in some cases the database may suffer from incomplete data. In addition to the historical performance data, the age of the pavement or the date of the last major rehabilitation is required to develop a relationship between the performance data and the age of the pavement. This paper provides a methodology to develop performance prediction models with the absence of the construction or rehabilitation dates. The models developed use the limited historical data and accounts for different parameters such as pavement thickness, traffic, and subgrade classes. The pavement construction dates, or age of the pavement, were incorporated into the proposed model and were constrained based on local experience and engineering judgment. A linear programming technique was employed to develop the performance prediction models. The approach presented in this paper can be expanded to additional parameters and can easily be adapted to different agencies based on their local experience.
Menu
