Proof-testing strategies induced by dangerous detected failures of safety-instrumented systems

Abstract

Some dangerous failures of safety-instrumented systems (SISs) are detected almost immediately by diagnostic self-testing as dangerous detected (DD) failures, whereas other dangerous failures can only be detected by proof-testing, and are therefore called dangerous undetected (DU) failures. Some items may have a DU-and a DD-failure at the same time. After the repair of a DD-failure is completed, the maintenance team has two options: to perform an insert proof test for DU-failure or not. If an insert proof test is performed, it is necessary to decide whether the next scheduled proof test should be postponed or performed at the scheduled time. This paper analyzes the effects of different testing strategies on the safety performance of a single channel of a SIS. The safety performance is analyzed by Petri nets and by approximation formulas and the results obtained by the two approaches are compared. It is shown that insert testing improves the safety performance of the channel, but the feasibility and cost of the strategy may be a hindrance to recommend insert testing.

Reliability modeling of subsea SISs partial testing subject to delayed restoration

“…are introduced to represent the enabling condition of a transition, and assertions denoted as "! " represents the formulas to update one variable when the transition is fired [2,11,29].…”

“…However, several issues need to be further investigated when they are applied to the subsea SIS system. The existing literatures [2,15,33,35] focus to a large extent on the reliability assessment of SIS based on assumptions that are questionable in a subsea context, for example: 1) The failures of SISs are mostly assumed to be exponentially distributed with the constant failure rates in these studies [3,5]. But in fact for many final elements working in a subsea environment, they are more likely to deteriorate with an increasing failure rate over time especially in the wear-out period [28,29].…”

“…It is worth noting that the failure rate in the simplified formulas considers DU failures only, and that other types of failure modes (e,g,.DD failures) are omitted. Considering the non-negligible repair time (MRT) in a proof test interval [0, τ] given α=1, for 1oo1 system, there is PFDavg=λDUτ/2+λDU•MRT, and for 1oo2 system, there is PFDavg=(λDUτ) 2 /3+MRT•λDU 2 •τ+(λDU•MRT) 2 /2, which is identical to PFDavg formulas in the work [5] for systems with constant failure rates in a proof test interval [0, τ]. These proposed formulas seem complicated, however, it can provide a method to straightforwardly calculate the unavailability considering partial testing and delayed repair.…”

Reliability analysis of subsea blowout preventers with condition-based maintenance using stochastic Petri nets

“…A large number of safety instrumented systems (SISs) are in use within the oil and gas industry for drilling, production, processing and storage purposes (Liu, 2014;Liu and Rausand, 2016). The complexities associated with these systems are amplified when taking into account the myriad of challenges within the offshore environment.…”

Performance analysis for subsea blind shear ram preventers subject to testing strategies

“…Generally, existing methodologies used for BOP reliability analysis can be categorized into two types: static methods (including fault tree analysis (FTA), failure modes and effects analysis (FMEA) and Bayesian network (BN)) [9][10][11], and dynamic methods (Dynamic BN (DBN), Markov model, and Petri net etc.) [12][13][14]. FTA and FMEA are widely used in detailed BOP reliabilities studies according to the reports by Holand et al [15][16][17].…”