Artificial Neural Network (ANN) models for PA lifespan

More than 60% of the Dutch motorways are covered with a porous asphalt wearing course (PA). In spite of many advantages of PA its lifespan is mostly short which causes high maintenance costs. Lifespan of PA is defined by its main damage being raveling. The main factors which cause raveling are asphalt construction/mixture and environmental factors. Therefore it is important to determine to what extent the development of raveling is dominated by these factors. Artificial Neural Networks (ANNs) are especially good in modeling complex problems. Their ability to extract relations between inputs and outputs of a process suits the problem of the PA lifespan well, since it is nonlinear and complex. The paper discusses ANN models developed to analyze the dominancy of parameters and to forecast raveling of PA. The study also shows that ANN models can only be as good as allowed by the data.