GNGTS 2018 - 37° Convegno Nazionale

526 GNGTS 2018 S essione 2.3 evaluated for considering both, the magnitude of the impact on the primary risk receptor, and the impacts on final risk receptors of interest. Since major risks are rare events for which scarce or no data are usually available, Garcia- Aristizabal et al. (2018) adopt a Bayesian framework (Fig. 1d) that allows us to coherently integrate all the useful information and to update assessments as site-specific data are retrieved. In this way, the assessments can be adjusted to the dynamic environment that usually characterises operations for the development of geo-resources. The MHR assessment approach presented in Garcia-Aristizabal et al. (2018) includes five classes of probabilistic models for implementing the stochastic characteristics of FT’s basic events (namely: homogeneous Poisson process, binomial, Weibull, static- and dynamic- physical reliability models). The performance of the proposed approach is illustrated through a simple synthetic example. The analysis is focused on the assessment of the possible pollution of an underground reservoir of drinking water during the management of flowback fluids after hydraulically fracturing a geologic formation for unconventional gas development. Two scenarios are considered in the analysis: (1) groundwater pollution caused by a spill outside of the site related to a volume of flowback fluid being transported from the well site to a disposal site; (2) groundwater pollution caused by a surface spill within the site related to the failure of a storage unit (e.g. a tank) containing the flowback fluids (for details see Garcia-Aristizabal et al. , 2018). Fig. 1 - Example of a generic bow-tie structure for a determined scenario of interest (modified from Garcia-Aristizabal et al. , 2017); (a) fault tree component of the bow-tie structure; (b) critical event linking the fault tree and the event tree; (c) event tree component of the bow-tie structure; (d) Bayesian inference of the parameters of the probabilistic models used to define basic events in fault trees and nodes in event trees.