Statistical analysis and applications of WIM data from A3 highway

Weigh-In-Motion (WIM) systems are a relatively recent technology, capable of providing data about traffic characteristics that are useful for various purposes and applications. These devices can be used indeed to monitor road traffic on a transport infrastructure, providing the possibility of controlling cases of vehicle overloading beyond the legal limits and of promoting better maintenance and management of the infrastructure thanks to a data-driven approach. In addition to this, they can be integrated with devices for Structural Health Monitoring, supplying input data related to the weight of vehicles that can be correlated with the output results detected with traditional monitoring techniques. Since through WIM devices it's possible to provide information about real vehicles' loads for a specific site, another interesting application consists of generating stochastic models for traffic load starting from WIM data. These models can then be compared with the traffic actions, prescribed in the structural codes for the design or assessment of bridges and derived from a semi-probabilistic approach, with a given exceeding probability in a time of interest. In the following thesis, this last aspect has been mainly studied. Starting from a statistical characterization of the available data, it has been possible to obtain stochastic models for traffic loads, using a Monte Carlo simulations approach. Several case studies have been then evaluated, considering the single or multiple lanes and separating the analysis based on the measurements of a specific site of the road segment or device. Correlations among some variables provided by WIM measurements have been also analysed, resorting to interpolation models. The results observed from studying the relationship among these variables have been then used to achieve more accurate simulations. The simulations output obtained have been then examined for different devices, fitted with generalized extreme value (GEV) distributions and then evaluated for different bridge lengths. From this latest analysis, it has been possible to get the parameters of a general GEV distribution and subsequently to compare the latter with real GEV distributions coming from simulations' results. Finally, updating algorithms have been adopted in order to fit the elaborated GEV distribution with the actual one, related to the single device or to a specific site. The WIM dataset used comes from the systems installed on the A3 highway, in the segment connecting the cities of Napoli, Pompei and Salerno. Due to the morphological context in which it's located, several bridges and viaducts are present in the tract, mainly built in the 50s and designed for different actions compared to the actual ones prescribed in the standards. For this reason, WIM devices have been installed to monitor the vehicles' characteristics and to prevent cases of overload. The available data come from one year of measurements of the devices installed in the road segment.