Domino scenarios triggered by fire pose severe risks to workers, assets, and the environment. Accurate quantitative models are needed to support mitigation actions addressing the prevention of fire escalation, especially considering sensitive targets such as atmospheric tanks containing large quantities of dangerous substances. A novel approach based on neural networks was developed, allowing the accurate quantification of the time-to-failure (TTF) of atmospheric tanks exposed to external fires accounting for mitigation actions. Data from a lumped parameter model were used to train and assess neural networks' performance. The toolbox of models obtained provides the TTF of atmospheric tanks both in the case of unmitigated fire scenarios and considering safety barriers and protection measures, such as water deluges and fire monitors. Model predictions are fast, accurate, and supplemented with confidence intervals. The comparative analysis demonstrated the better performance of the model developed compared to simplified correlations widely used in the literature to predict TTF. The approach developed, based on the integration of neural networks in consequence analysis tools, shows significant potential for the advancement of a quantitative assessment of domino scenarios, providing accurate and user-friendly tools for a quick evaluation of domino fire scenarios under both mitigated and unmitigated conditions.

Tamascelli, N., Scarponi, G.E., Amin, M.T., Sajid, Z., Paltrinieri, N., Khan, F., et al. (2024). A neural network approach to predict the time-to-failure of atmospheric tanks exposed to external fire. RELIABILITY ENGINEERING & SYSTEM SAFETY, 245, 1-13 [10.1016/j.ress.2024.109974].

A neural network approach to predict the time-to-failure of atmospheric tanks exposed to external fire

Tamascelli, Nicola;Scarponi, Giordano Emrys;Cozzani, Valerio
2024

Abstract

Domino scenarios triggered by fire pose severe risks to workers, assets, and the environment. Accurate quantitative models are needed to support mitigation actions addressing the prevention of fire escalation, especially considering sensitive targets such as atmospheric tanks containing large quantities of dangerous substances. A novel approach based on neural networks was developed, allowing the accurate quantification of the time-to-failure (TTF) of atmospheric tanks exposed to external fires accounting for mitigation actions. Data from a lumped parameter model were used to train and assess neural networks' performance. The toolbox of models obtained provides the TTF of atmospheric tanks both in the case of unmitigated fire scenarios and considering safety barriers and protection measures, such as water deluges and fire monitors. Model predictions are fast, accurate, and supplemented with confidence intervals. The comparative analysis demonstrated the better performance of the model developed compared to simplified correlations widely used in the literature to predict TTF. The approach developed, based on the integration of neural networks in consequence analysis tools, shows significant potential for the advancement of a quantitative assessment of domino scenarios, providing accurate and user-friendly tools for a quick evaluation of domino fire scenarios under both mitigated and unmitigated conditions.
2024
Tamascelli, N., Scarponi, G.E., Amin, M.T., Sajid, Z., Paltrinieri, N., Khan, F., et al. (2024). A neural network approach to predict the time-to-failure of atmospheric tanks exposed to external fire. RELIABILITY ENGINEERING & SYSTEM SAFETY, 245, 1-13 [10.1016/j.ress.2024.109974].
Tamascelli, Nicola; Scarponi, Giordano Emrys; Amin, Md Tanjin; Sajid, Zaman; Paltrinieri, Nicola; Khan, Faisal; Cozzani, Valerio
File in questo prodotto:
File Dimensione Formato  
RESS_AI_TTF.pdf

accesso aperto

Tipo: Versione (PDF) editoriale
Licenza: Creative commons
Dimensione 4.52 MB
Formato Adobe PDF
4.52 MB Adobe PDF Visualizza/Apri
ScienceDirect_files_25Mar2024_11-22-38.721.zip

accesso aperto

Tipo: File Supplementare
Licenza: Creative commons
Dimensione 5.25 MB
Formato Zip File
5.25 MB Zip File Visualizza/Apri

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11585/960543
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 2
  • ???jsp.display-item.citation.isi??? 2
social impact