Introduction. There is a world-wide trend to also increase the sustainability of the road sector. The growing use of various industrial by-products, together with economical and eco-friendly construction and maintenance techniques can be observed in many countries.
Problem Statement. The utilization of warm mix asphalt and the use of relatively high share of reclaimed asphalt materials in new asphalt mixtures can have negative features, as well.
Purpose. Modelling indirect tensile strength of warm mix asphalt with variable reclaimed asphalt pavement (RAP) content was aimed at based on Hungarian laboratory test series.
Materials and Methods. Three models were developed for the prediction of indirect tensile strength, this important asphalt mechanical parameter of warm mix asphalt as a function of Foamed Bitumen Content (FBC) and the RAP share in the new asphalt mixture. Among others, linear regression analysis and support vector regression (SVR) models were applied.
Results. A comparison performed between Random Forest and Neural Network models illustrates and proves the versatility of machine learning techniques in predicting asphalt indirect tensile strength values both in wet and dry conditions. The research work enhances our understanding of the multifaceted dynamics influencing the performance of asphalt mixtures, offering valuable insights for optimizing pavement design and construction practices in diverse environmental conditions. The model developed successfully captures the relationship between the ITS (wet and dry) metric and its contributing factors, Foamed Bitumen Content (FBC) and RAP, with a high R-squared value.
