Applying a K-means model to TSD data to find categories for the structural assessment of flexible pavements

• Modeling of measured TSD data with K-means to find evaluation categories for pavement • Cluster analysis of slope values as measured data from TSD alongside SCI300 values • Identification of categories for the structural status assessment of flexible pavements • Correlation found between K-means models and mechanistic models • Increased applicability of TSD measurements for the pavement management system Since 2018, the German Federal Highway Research Institute (BASt) has been using a traffic speed deflectometer (TSD) for measurements on network level to assess the structural condition of asphalt pavements. TSD collects a variety of data, such as deflections, temperature and slope values at each measuring point. But for the evaluation of this data, there is no established methodological framework yet. The common methodological approach for falling weight deflectometer (FWD) data uses threshold values for different categories in order to assess the condition of the pavement. But as the load setup and properties of the TSD are different from FWD, the existing FWD- based thresholds are not directly applicable to TSD-based data. The objective of this study is to develop a new, data-driven approach for the analysis of TSD data in order to find categories for the structural status assessment of flexible pavements. A database with more than 113,000 data points of TSD measured data is used for K-means clustering slope values and SCI300 values in order to divide the data into different categories relevant for the assessment of the structural condition of pavements. The resulting threshold values of the categories found showed obvious correlations to the results of mechanistic models. The K-means model is thus a good supplement to mechanistic models and may even support their validation. The results support the practical applicability of the TSD as new measuring device for the pavement management system.