Contents lists available at ScienceDirect Tunnelling and Underground Space Technology journal homepage: www.elsevier.com/locate/tust Calibration of a fuzzy model estimating fire response time in a tunnel Pavel Přibyl, Ondřej Přibyl ⁎ Czech Technical University in Prague, Faculty of Transportation Sciences, Na Florenci 25, Praha 1 11000, Czech Republic ARTICLE INFO Keywords: Tunnel safety Tunnel technology Fire sensors Warning system Fuzzy system Calibration ABSTRACT Safety is one of the most important aspects when designing a road tunnel system. Apart from the general design of a road tunnel, different technological safety systems can contribute to increased safety. There is, however, no agreed methodology on how to evaluate such systems prior to their design and installation. In this paper, it is recommended that the time required to detect a fire and warn people about it in the tunnel be used as a quality criterion since it has a direct effect on the probability of saving lives. In their previous paper, the authors proposed a fuzzy system called SAFECALC for effectively evaluating fire sensors and warning systems in tunnels, even in the early design phase. The biggest challenge in designing a fuzzy system is the original identification and calibration of such a system. For that reason, this paper focuses on the identification stage and, using the example of a linear fire sensor, it suggests a new methodology for performing such early level calibration. This methodology consists of several steps and, after the original design of the system, it uses inputs provided by experts in the field (via surveys and brainstorming) for fine tuning of the system. A physical model is used to simulate the propagation of a fire in a tunnel. The results of such the process are then evaluated on a real world case study from Lochkov tunnel near the city of Prague. 1. Introduction Over recent years, the number of road tunnels has significantly in- creased. Their purpose is not only to avoid natural obstacles; tunnels currently form complex underground structures (Falconnat, 2013). Tunnel systems have lately even been adding new functionality to urban areas, including tunnel intersections, parking possibilities or even services for pedestrians. Therefore, because of the increased usability of tunnels, the prevention of fire and the safety evacuation of people lies on the critical path of the sustainable development of these important infrastructural elements. Particularly, the fire has proved to be espe- cially dangerous since it can cause massive injuries or death of people trapped in the tunnel. To minimize the negative impact of a fire, different technical safety systems have been installed to tunnels (Hrbček et al., 2014). The pri- mary role of those systems is to quickly detect such fire and efficiently warn the people in danger. It has been repeatedly reported that the first six to eight minutes are decisive in rescuing the trapped people. The design of escape routes, requirements for lighting systems and some other measures related to the construction of a tunnel are typi- cally sufficiently described in existing standards. On the other hand, a large group of technological devices, such as fire sensors or warning systems, could not be entirely prescribed by standards. It is caused mainly by rapid innovations and new possibilities in this field, where the standardization effort would limit this natural progress. Moreover, every tunnel is unique and there is not universal guide for design of all facilities. Since the number, combination and placement of different fire sensors and warning devices vary at different installations and depend significantly on the construction and geometry of the tunnel, a decision which configuration is sufficient and in which case a redesign has to be recommended is not trivial. Typically, the design of safety devices depends on the designer and usually it is discussed and finally approved by the project owner. Because of the considerably individual approach, it is important to have an evaluation system available that can provide the decision-making support whether the technological safety system provides sufficient safety for people in the tunnel. Such the assessment has to be provided before the actual installation of the technology (in order to minimize the future investment). This article provides a solution for the assessment of reaction time of fire sensors during the design stage, even though the same procedure can be used for the evaluation of the detection time of smoke sensors or the warning system as well. It focuses on quantitative evaluation, while a more common approach lies in assessment conducted on a qualitative scale, as presented, for example, in Manca and Brambilla (2011). The authors present a model using the principles of artificial in- telligence to address this problem. In this paper however, the focus is on http://dx.doi.org/10.1016/j.tust.2017.06.009 Received 14 December 2015; Received in revised form 5 June 2017; Accepted 8 June 2017 ⁎ Corresponding author. E-mail address: pribylo@fd.cvut.cz (O. Přibyl). Tunnelling and Underground Space Technology 69 (2017) 28–36 0886-7798/ © 2017 Elsevier Ltd. All rights reserved. MARK