ABSTRACT
The present study aims to review the efficacy of three widely used roughness prediction models (Highway Development Model/HDM-4, Australian and South African) in the determination of blading frequencies to control roughness in the management of unsealed roads. The review comprises an assessment of the information used in the formulation of the models, the contribution of the input parameters and the effect if used in pavement management. The models are found to be based on dissimilar environmental and traffic conditions, to have relatively low coefficients of determination and variable contributions of the input parameters. The consequences are very diverse predictions of roughness and the determination of blading frequencies. The results of the assessment were used to develop blading frequency relationships using traffic volumes, rainfall category and blading cost as input for a range of commonly used models and two maintenance approaches, threshold level or user cost criteria. These relationships provide an alternative for the determination of blading frequencies if accurately calibrated models are not available.