Abstract:Based on the longterm pavement performance (LTPP) faulting data, the sensitivity analysis was divided into two steps. In the first step, 5 methods of correlation analysis, variance analysis, grey correlation analysis, principal component analysis and neural network method were compared. Graphical analysis was used to verify the accuracy. The optimal analysis method and the key design feature were presented. In the second step, the whole data were divided into two datasets of dowel and nodowel sections. Correlation analysis and variance analysis were used to analyze the sensitivity of design feature. With the graphical analysis, the design feature sensitivity for dowel and nodowel sections was also analyzed, and the suggestions for controlling faulting were proposed. The results indicate that correlation analysis performs the best. For sections with dowel bar, setting dowel bar is the most effective method. For sections without dowel bar, decreasing joint space is more effective method.