Predicting the Consequences of Hydrogen Releases: How a Machine Learning Approach May Improve Risk-Based Inspection Planning

Giannini, Leonardo; Salzano, Ernesto; Paltrinieri, Nicola; Tamascelli, Nicola

Permalink

https://hdl.handle.net/2142/121831 Copy

Description

Title

Predicting the Consequences of Hydrogen Releases: How a Machine Learning Approach May Improve Risk-Based Inspection Planning

Author(s)

• Giannini, Leonardo
• Salzano, Ernesto
• Paltrinieri, Nicola
• Tamascelli, Nicola

Issue Date

2023

Keyword(s)

• Hydrogen
• Machine learning
• Inspection planning
• Consequence analysis

Abstract

The roadmap of the European Green Deal foresees a wide spreading of hydrogen technologies in the following decade, setting a goal of 10 million tons of domestic, renewable hydrogen production by 2030 [1]. However, the unique thermodynamic characteristics of hydrogen and this possible future implementation in densely populated areas imply that safety issues need to be tackled in the most complete, solid, and reliable way. In addition, the demand for inspection planning and maintenance activities will arise as a pivotal character in the long run, highlighting the necessity of detailed procedures and operations specifically conceived for hydrogen systems. Nevertheless, research on consequence analysis of accidental scenarios currently holds the lion’s share of hydrogen safety topics, while inspection standards for equipment working in pure hydrogen environments are scarcely considered. The present work aims at indicating how a machine learning (ML) approach could cast light on the consequence of hydrogen releases, an aspect which is crucial for risk-based inspection (RBI) planning. In this way, the data gathered from the Major Hazard Incident Data Service (MHIDAS) database are analyzed through ML algorithms and used to obtain models for consequence prediction for hydrogen releases, which may, in turn, support RBI planning. This approach aims to decrease safety-related uncertainties and optimize the associated operations. In fact, while consequence underestimation is unacceptable, overestimating this parameter could lead to unsustainable costs and redundant inspection procedures, affecting their efficacy and ultimately inhibiting the spread of a promising technology. [1] European Commission, 2019. The European Green Deal. COM(2019) 640 final, Brussels.

Type of Resource

text

Language

eng

Handle URL

https://hdl.handle.net/2142/121831

Sponsor(s)/Grant Number(s)

Research Council of Norway and SINTEF Industri Grant Number 327009

Owning Collections