Fire Detection System Reliability Analysis: An Operational Data-Based Framework

This paper describes a framework developed at CERN, conducting reliability analysis of Safety-Critical Systems (Fire detection and Alarms) based on operational data. It applies Fault-Tree Analysis on maintenance-related data, categorized based on the component on failure. This framework, a tool implemented in Python, accounts for Fire Detection components installed in tunnels and surface buildings (control panels, detectors, etc) and safety functions triggered upon detection (evacuation, alarms to the CERN Fire Brigade, compartmentalization, electrical isolation, etc). The usefulness of the results of this type of analysis is twofold. Firstly, the results are a supporting tool for estimating the yearly availability of Fire Detection Systems in critical facilities, crucial in Capital and Operational Expenditure identification. Additionally, this approach refines the frequency analysis as part of quantitative fire risk assessments performed in the context of the FIRIA (Fire-Induced Radiological Integrated Assessment) Project, launched by CERN in 2018 and aiming at assessing the risk of fire events in experimental facilities with potential radiologic consequences to the public.