For the design and operation of Oil and Gas (O&G) facilities, a Quantitative Risk Assessment (QRA) should be performed to quantify the risk of accidents due to multiple hazards and sources, in terms of Risk Measures (RM) (like Individual Risk (IR) or Societal Risk (SR)), since the outcomes of the QRA would allow the effective allocation of safety barriers to prevent accidents and mitigate consequences, only if a complete picture is available of the exposure of the facility to multiple hazards. To account for the knowledge and lack of knowledge on the phenomena involved in the accident scenarios, a QRA procedure should allow the proper analysis and treatment of the uncertainties related to the frequency of occurrence of accidental scenarios and the severity of their consequences. In this thesis work, we propose a novel approach for the aggregation of risks from multiple hazards and sources, accounting for the uncertainties on the frequency and severity of the accidental scenarios. The uncertainty is propagated in the QRA model by Monte Carlo Simulation (MCS). The multi-hazard risk assessment framework proposed is applied to assess the Location-Specific Individual Risk (LSIR) for a representative offshore O&G plant, using a model based on multistate Bayesian Networks (BNs) for different functional units (i.e., risk sources, namely, gas wellhead, oil wellhead, gas manifold, oil manifold, knock out drum flare, oil separator, delivery pumps, launching trap, gas fuel system), each one undergoing an initiating event of Loss Of Primary Containment (LOPC). Estimates of frequency and severity for each possible hazard (namely Flash Fire (FF), Jet Fire (JF), Pool Fire (PF)) are aggregated to provide the overall estimate of the LSIR, its confidence intervals to describe the uncertainty in the estimate, and the separate frequency and severity contributions to risk, for targeted prioritization of the safety barriers to deploy for risk reduction. Finally, a sensitivity analysis is performed to identify the safety barriers most contributing to the LSIR assessment.
Per la progettazione e il funzionamento degli impianti oil and gas (O&G), dovrebbe essere effettuata una valutazione quantitativa del rischio (Quantitative Risk Assessment, QRA) per quantificare il rischio di incidenti legato a molteplici pericoli da diverse sorgenti, in termini di misure di rischio (Risk Measures, RMs) (come il rischio individuale (Individual Risk, IR) o il rischio sociale (Societal Risk, SR), tuttavia i risultati della QRA consentirebbero l'effettiva assegnazione di barriere di sicurezza per prevenire incidenti e mitigare le conseguenze, solo se è disponibile un quadro completo dell'esposizione dell'impianto ai molteplici rischi. Per tenere conto delle conoscenze e della mancanza di conoscenza dei fenomeni coinvolti negli scenari incidentali, una procedura QRA dovrebbe consentire un'analisi e un trattamento adeguati delle incertezze legate alla frequenza del verificarsi di scenari incidentali e alla gravità delle loro conseguenze. In questo lavoro di tesi proponiamo un nuovo approccio per l'aggregazione dei rischi legato a molteplici pericoli provenienti da diverse sorgenti, che consideri le incertezze sulla frequenza e la gravità degli scenari incidentali. L'incertezza si propaga nel modello QRA attraverso una simulazione Monte Carlo (Monte Carlo Simulation, MCS). Il quadro di valutazione del rischio multi-pericolo proposto viene applicato per valutare il rischio individuale in una specifica posizione (Location Specific Individual Risk, LSIR) per un impianto O&G offshore rappresentativo, utilizzando un modello basato su reti bayesiane (Bayesian Networks, BNs) multistato per diverse unità funzionali (ovvero le sorgenti di rischio, vale a dire, testa di pozzo del gas, testa del pozzo petrolifero, collettore di gas, collettore di petrolio, torcia di combustione, separatore di olio, pompe di consegna, trappola di lancio, sistema di carburante a gas), ognuno dei quali soggetto a un evento iniziale di perdita di contenimento primario (Loss Of Primary Containment, LOPC). Le stime della frequenza e della gravità per ogni possibile pericolo (vale a dire fuoco rapido (Flash Fire, FF), getto incendiario (Jet Fire, JF), incendio da pozza (Pool Fire, PF)) sono aggregate per fornire la stima complessiva dell'LSIR, i suoi intervalli di confidenza per descrivere l'incertezza nella stima e i contributi separati di frequenza e gravità al rischio, per una definizione mirata delle barriere di sicurezza da implementare per la riduzione del rischio. Infine, abbiamo indentificato le barriere di sicurezza che più contribuiscono alla valutazione del LSIR con un’analisi di sensitività.
A novel multi-hazard risk aggregation approach in oil and gas facilities
Martinoia, Luca
2019/2020
Abstract
For the design and operation of Oil and Gas (O&G) facilities, a Quantitative Risk Assessment (QRA) should be performed to quantify the risk of accidents due to multiple hazards and sources, in terms of Risk Measures (RM) (like Individual Risk (IR) or Societal Risk (SR)), since the outcomes of the QRA would allow the effective allocation of safety barriers to prevent accidents and mitigate consequences, only if a complete picture is available of the exposure of the facility to multiple hazards. To account for the knowledge and lack of knowledge on the phenomena involved in the accident scenarios, a QRA procedure should allow the proper analysis and treatment of the uncertainties related to the frequency of occurrence of accidental scenarios and the severity of their consequences. In this thesis work, we propose a novel approach for the aggregation of risks from multiple hazards and sources, accounting for the uncertainties on the frequency and severity of the accidental scenarios. The uncertainty is propagated in the QRA model by Monte Carlo Simulation (MCS). The multi-hazard risk assessment framework proposed is applied to assess the Location-Specific Individual Risk (LSIR) for a representative offshore O&G plant, using a model based on multistate Bayesian Networks (BNs) for different functional units (i.e., risk sources, namely, gas wellhead, oil wellhead, gas manifold, oil manifold, knock out drum flare, oil separator, delivery pumps, launching trap, gas fuel system), each one undergoing an initiating event of Loss Of Primary Containment (LOPC). Estimates of frequency and severity for each possible hazard (namely Flash Fire (FF), Jet Fire (JF), Pool Fire (PF)) are aggregated to provide the overall estimate of the LSIR, its confidence intervals to describe the uncertainty in the estimate, and the separate frequency and severity contributions to risk, for targeted prioritization of the safety barriers to deploy for risk reduction. Finally, a sensitivity analysis is performed to identify the safety barriers most contributing to the LSIR assessment.File | Dimensione | Formato | |
---|---|---|---|
2021_04_Martinoia.pdf
non accessibile
Descrizione: File Tesi
Dimensione
7.32 MB
Formato
Adobe PDF
|
7.32 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/174047