Methods for determining PFD/SIL for workover control systems with short test-intervals and imperfect testing: Metoder for å bestemme PFD/SILfor workover kontrollsystemer med korte testintervall og ikke-perfekte tester

Abstract

Safety related-systems are subject to periodic proof testing, and it is often assumed that the proof test is perfect. In recent years, partial proof testing has been introduced in order to improve the system s reliability, and partial proof testing is often referred as imperfect proof testing. In this master s thesis we show that partial proof testing is different from imperfect proof testing, and, a mathematical model for modelling the effect of both partial and imperfect proof testing is proposed.In addition to imperfect and partial proof tests, the system can also be subject to proof testing with short test intervals (intervals in the order of two or three weeks). In theory, a system that is subject to short test intervals is highly reliable, however, there are some factors like (i)human errors that are introduced during proof testing, and (ii)wear, that impact the system's reliability. We present and discuss the major contributors to the \emph{unreliability} of a system, which is often known as the system's unavailability.In this master's thesis we study four reliability assessment methods for estimating the probability of failure on Demand, $\textrm{PFD}_\textrm{avg}$ of safety-related systems that are subject to partial and imperfect proof testing. In addition, the effects of short test intervals are also studied.Additional factors that contribute to high reliability of a system during the system's life cycle are discussed and highlighted. A Workover Control System functions as a safety barrier during workover and intervention operations for subsea production wells. In addition to partial and imperfect proof testing, this type of system is also subject to proof testing with short test intervals.This system is used as a case study to illustrate the use of the proposed model and to discuss the effects partial and imperfect proof testing.