This paper presents a comprehensive study to establish the threshold limit of CTIndex for the Marshall mixes. The study also aims to assess the impact of different design factors on the cracking resistance of bituminous mixtures using the Indirect Tensile Asphalt Cracking Test (IDEAL-CT) test. This study considers 2 different aggregate sources, 2 different gradations, 5 different types of Design Aggregate Gradation (DAG), 3 binder types, and 5 levels of compactive effort. The test results show that higher cracking resistance can be achieved using a smaller nominal maximum size of the aggregate (NMAS), finer gradation, modified binder, and decreased compactive effort with aggregates having low abrasion and absorptive characteristics. This study also comprehends the influence of volumetric parameters of the bituminous mixes on fracture resistance. It was found that at a particular Optimum Binder Content (OBC), higher bulk specific gravity of the compacted specimen (Gmb) and voids filled with asphalt (VFA) result in reduced CTIndex, specifying poor cracking performance. While higher air voids (AV) and voids in mineral aggregate (VMA) lead to increased CTIndex, indicating better-cracking resistance. Statistical analysis tools were used to evaluate the significance of the influential factors that are affecting the cracking potential of the mix. Different Machine learning models were also developed to predict CTIndex based on the design factors considered in the study. The random forest (RFR) model showed strong accuracy, reflected by low Mean Absolute Error (MAE=3.16), Mean Absolute Percentage Error (MAPE=9.57), Root Mean Square Error (RMSE=4.23), and a high coefficient of determination (R²=0.95) value, notifying a precise fit and reliable predictions. Additionally, a GUI has been also developed to enhance the practical usability of the model for wider usage. Further, the present study proposes the threshold value of CTIndex for the selection of crack-resistant bituminous mixtures. Moreover, the study investigated the correlation between laboratory and field compaction methods and validated the initial threshold specification of CTIndex for the Marshall mixes. Despite of the variations in different compaction methodologies and specimen thickness, a strong positive correlation (R² > 0.76) between laboratory and field cores of BC-1 and DBM-2 indicates that the performance criteria are adequate and justified.
Read full abstract