In order to establish a correct balance between safety and cost reduction around underground structures such as tunnels, research and analysis of factors that during construction contribute to increment the hazard of dangerous accidents are described, with the intent to reduce or eliminate them. For several tunnels typical hazards are explosive gas invasion, like methane. In this paper the development of a risk classification is described, using Favourability Functions, to create maps with different risk classes regarding the possibility of gas invasion during the excavation of a tunnel. This classification was first applied to a real case study, the system of galleries for high-speed trains construction between Bologna and Florence, Italy. To develop the Favourability Functions models, the Bayes theorem and the Fuzzy Logic were used, which are well adaptable to a predicting study. To process the geographic data for this study, a Geographic Information System called ILWIS was used.
This study deals with the problem of methane invasion in tunnels under construction with a risk analysis process based on Fuzzy Logic and bayesian probability models. The models calculate on the same geographic area and with the same database, different degrees of probability related to the event to be predicted (in this case the event is "methane invasion in tunnels") and they may be called Favourability Functions. The geographic area involved in the study is limited at north by the city of Bologna and at south by Florence. This area was subjected to excavations for the construction of various tunnels belonging to the complex project of high-speed trains. The input data was prepared at two different levels: firstly, the various hypothetical causes of methane creation and accumulation (lithology, faults, anticlines and synclines) and secondly the evidences of methane existence (cited in the literature and registered during preliminary geological studies, investigations by drill holes, previous tunnel excavation and geological studies). The study purposes a final map where the investigated area is divided in 20 risk classes, in Which the zero of this classification means that a very low predisposition of ground for having methane, and values around 20 mean that the probability is consistent.
These functions associate, for every element of a studied area domain, a number that is related to the predisposition of that domain for the event under prediction. The favourability of an event to occur is the possibility for the event to become real in the area in which it is defined.
The geographic area involved in this study was limited at north and south by the cities of Bologna and Florence respectively, in which many excavations were conducted for the construction of various tunnels belonging to the high-speed train project. Thus the extents of studied area were 87 km in latitude and 59,3km in longitude. According to the experiences gathered in previous construction of highways and railroad tunnels in that area, the probability of finding geological formations containing methane during the excavations was really high.
