Quantitative reliability modelling and functional safety calculations of Svend topside High Integrity Pressure Protection System
Author
Glæsner, Jacob
Term
4. semester
Education
Publication year
2017
Pages
125
Abstract
En Lag-af-beskyttelse-analyse (LOPA) for Svend olie- og gasplatformen viste, at en farlig hændelse kunne medføre op til 10 dødsfald og et økonomisk tab på op til 1 mia. USD. Derfor blev en opgradering af High Integrity Pressure Protection System (HIPPS) foreslået. Et HIPPS er et Safety Instrumented System (SIS), der skal være tilstrækkeligt pålideligt for at opfylde det krævede sikkerhedsintegritetsniveau, SIL 2. Specialet beskriver HIPPS-arkitekturen på Svend og tre kvantitative metoder til at beregne pålidelighed: Reliability Block Diagrams (RBD), Fault Tree Analysis (FTA) og Markov-modellering. Vi beregner den gennemsnitlige sandsynlighed for fejl ved behov (PFDavg)—altså sandsynligheden for at systemet svigter, når det skal aktiveres—med hver metode og sammenligner resultaterne. RBD og FTA ligner hinanden i tilgang, mens Markov-modellering hurtigt bliver mere kompleks, fordi antallet af tilstande kan vokse eksponentielt. Til gengæld kan et SIS beskrives mere detaljeret med Markov-modeller. Beregningerne viser, at PFDavg afviger mindre end 1 % mellem metoderne. Med de foreslåede komponenter og den valgte arkitektur opnår HIPPS på Svend topside det krævede SIL 2. Konklusionen er, at det er vigtigere, at brugeren behersker den valgte metode, end hvilken metode der vælges.
On the Svend oil and gas platform, a Layers of Protection Analysis (LOPA) predicted that a hazardous incident could lead to up to 10 fatalities and economic losses of up to USD 1 billion. This led to a recommendation to upgrade the High Integrity Pressure Protection System (HIPPS). A HIPPS is a Safety Instrumented System (SIS) that must be reliable enough to meet the required Safety Integrity Level, SIL 2. This thesis describes the Svend HIPPS architecture and compares three quantitative reliability methods: Reliability Block Diagrams (RBD), Fault Tree Analysis (FTA), and Markov modeling. We calculate the average Probability of Failure on Demand (PFDavg)—the chance the system fails when called upon—with each method and compare the results. RBD and FTA use similar approaches, whereas Markov modeling becomes more complex as the number of states can grow exponentially; however, it can represent the SIS in greater detail. Across all three methods, PFDavg values differ by less than 1%. With the proposed components and architecture, the Svend topside HIPPS achieves the required SIL 2. Overall, the competence of the practitioner in applying a method is more important than which method is chosen.
[This abstract was generated with the help of AI]
Documents