Employment of Bayesian networks for risk assessment of excavation processes in dense urban areas

Excavation processes can frequently manifest critical issues regarding permanent damages in surrounding buildings. The employment of bulkheads or pile-sheets in dense urban areas cannot overlook a proper monitoring of the excavation process. It can be necessary to combine different kinds of information and computations since excavation processes gather together dissimilar issues and require a wide assortment of solution strategies. Aim of this work is to propose a strategy for monitoring purposes, based on the employment of Bayesian networks. The proposed strategy lays on the statistics of soil parameters and external loads which are available in the literature. Computing the results of land survey measurements of the site, a Bayesian network is able to update the model’s parameters and to estimate many quantities of interest. A proper choice of them can be employed in order to set a limit state condition aiming to avoid permanent damages in surrounding buildings. The proposed procedure has been applied to a real excavation site. The inference process shows how, for the considered site, an efficient reliability analysis cannot leave out of consideration the detection of ongoing soil displacements. This is because geotechnical tests are not sufficient to properly define the model features. Furthermore, it takes advantage of soft evidences which are provided by employed survey methodology. Eventually, the proposed strategy aims to be fast and efficient in order to promptly give advice of critical issues while excavation is carried out.