Resumen
Surfaces of airport pavements are subject to the friction decay phenomenon. A recurrent problem for the runways is represented by the deposits of vulcanized rubber of aircraft tires. This happens mainly in the touch-down areas during landing operations, and the loss of grip compromises the safety of both take-off and landing operations. This study moves from the International Civil Aviation Organization and the Italian Civil Aviation Authority provisions concerning runway friction measurement and reporting to a better way to analyze friction data. Being data mining the computational process of discovering patterns in a large data sets, data mining techniques are very helpful to reach this target. Unsupervised and supervised classification methods to analyze friction data detected by Grip Tester Trailer were employed. First, K-means and Subtractive Clustering were applied to divide data into a certain number of clusters representing the different areas of consumption. In a second time two different Classification and Regression Trees models, CART and GCHAID, were employed to split the data points of the runway into nodes. At the end of the process scatterplots were built and better visualized through non-linear regressions. The decay curves obtained were of service to compare the results achieved using data mining techniques versus the International Civil Aviation Organization and the Italian Civil Aviation Authority provisions in order to find out the best way to analyze friction data. The final goals are to assure an optimum scheduling of the Airport Pavement Management System, as well as users safety.