Regression models for permanent deformation parameters using in-service pavement data from the SPS-1 experiment

摘要

This paper uses the results of multivariate regression analyses on rutting data from the SPS-1 experiment in the Long Term Pavement Performance (LTPP) program to develop models for predicting permanent deformation parameters (α and μ) for a three layers pavement system. All available material, structural and climatic data were extracted from the LTPP database and using simple linear regression, α and μ were regressed against each data category as an independent variable. The variables that have relatively high R~2 values were introduced into the multiple linear regression models. Backward regression analysis was used to develop the final models from statistically significant variables, which differed between layers. Since α- and μ-values were backcalculated from time scries data, α-prediction models for all layers are more accurate than μ-prediction models. Also, μ-values were significantly affected (positively) by their corresponding α-values, suggesting pavements with lower μ-values (lower initial rutting) will show lower α-values (higher rut growth with time). The regression equations developed in this paper should be used within the range of the data presented herein to obtain reasonable predictions. Also, the parameters α and μ are not material properties, but parameters to be used within the prescribed empirical procedure to predict the rut depth.