Development of pavement condition forecasting models

Pavement management is a process that helps maintain a pavement network in a safe and serviceable condition in a cost effective manner. A key component of an effective pavement management system is the ability to predict the future condition of pavements Future condition of pavements can be predicted using pavement condition forecasting models. The objectives of this study are to investigate both the network level and project level forecasting models and to discuss the advantages and disadvantages of the models. The accuracy of a forecasting model depends on the pavement group, or family from which it is developed. A single forecasting model developed for all the pavements will have lower prediction accuracy than a forecasting model developed for a specific group of similarly behaving pavements. Therefore, before developing forecasting models, data were separated into similar performing groups. Pavement performance data were first divided into three groups (high, medium and low performance) based on their deterioration trends. The logistic regression model was then used to predict the likelihood that a pavement will belong to a particular family. The results show that separating data into similar performing groups and developing separating forecasting model for each group gives more reliable results than a single forecasting model developed for entire dataset. The validation of project level forecasting models show that the Markov model has the highest accuracy. The remaining service lives estimated from the network level derived performance model were comparable to those estimated from the project level Markov model.