Data analytics for hydrogen safety management: support for preparedness and recovery

The great economic development that occurred in recent decades worldwide has led to an ever-increasing demand for energy. To date, this demand has been met, in large part, through the use of fossil fuels. They are responsible for the countless environmental damages that plague the planet Earth, including the accumulation of greenhouse gases and worsening global warming. Moreover, their prolonged and intensive consumption has resulted in their near-complete depletion. From the above, two needs are evident: fighting climate change and diversifying energy sources, moving toward green power generation. Hydrogen can play a key role in this transition process. However, some characteristics of hydrogen, such as high flammability and ability to permeate and embrittle materials, cause significant safety concerns. Learning from past events represents one of the best ways to ensure the large-scale application of hydrogen. Hence, the need arises to build databases to collect information on accidents, incidents, and near misses. Furthermore, due to the advent of the Industry 4.0 paradigm, safety management is progressively based on data analytics, which is the process of collecting, cleaning, sorting, and processing raw data to extract relevant and valuable information. However, effective support for accident preparedness and recovery can be obtained only through a structured and systematic collection of relevant information. This master thesis focuses on the development, improvement, and exploitation of accident databases and suggests a practicable way forward for the energy industry. Specifically, this work involves: 1. creation of a database by merging multiple sources; 2. application of Business Intelligence (BI), Text Mining (TM), and Machine Learning (ML) techniques.